Citation
Inclusive pathways to gifted education : examining gifted referral processes

Material Information

Title:
Inclusive pathways to gifted education : examining gifted referral processes
Creator:
Durtschi, Melissa Dayle
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Doctor of education)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
School of Education and Human Development, CU Denver
Degree Disciplines:
Leadership for educational equity
Committee Chair:
Fulmer, Connie L.
Committee Members:
Blunck, Rodney L.
Christensen, James

Notes

Abstract:
This study explored the affect race had on gifted referrals at the 90th and 95th percentiles, based on the Cognitive Abilities Text 7 (CogAT 7). Achievement, Opportunity and Excellence Gaps are defined and put into historical context including systems of power and privilege. Giftedness is defined. The history of Gifted Education is explored through a Culturally and Linguistically Diverse (CLD) lens. Research questions explored race and its affect on the Cognitive Abilities Test, CogAT 7, and the significance of difference between racial groups at the 90th and 95th percentiles. The results of this study indicate: (a) CogAT 7 test scores are affected by race, (b) 95th percentile gifted referrals were significantly different across race, (c) 90th percentile gifted referrals were significantly different across race, (d) when comparing 90th and 95th percentile referrals, there was a significant difference, and (e) gifted referrals at the 90th and 95th percentiles were significantly not equitable across race. Recommendations for further study include the development of district equity goals, second screening processes, and talent enrichment programs and/or strategies for CLD learners.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright [name of copyright holder or Creator or Publisher as appropriate]. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
INCLUSIVE PATHWAYS TO GIFTED EDUCATION:
EXAMINING GIFTED REFERRAL PROCESSES
by
MELISSA DAYLE DURTSCHI B.S., Brigham Young University, 2004 M.A., Adams State University, 2016
A dissertation submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Education Leadership in Educational Equity
2019


©2019
MELISSA DAYLE DURTSCHI ALL RIGHTS RESERVED
11


This dissertation for the Doctor of Education degree by
Melissa Dayle Durtschi has been approved for the Leadership for Educational Equity by
Connie L. Fulmer, Chair Rodney L. Blunck James Christensen
Date: May 18, 2019


Durtschi, Melissa Dayle (EdD, Leadership in Educational Equity)
Inclusive Pathways to Gifted Education: Examining Gifted Referral Processes Dissertation directed by Professor Connie L. Fulmer
ABSTRACT
This study explored the affect race had on gifted referrals at the 90th and 95th percentiles, based on the Cognitive Abilities Text 7 (CogAT 7). Achievement, Opportunity and Excellence Gaps are defined and put into historical context including systems of power and privilege. Giftedness is defined. The history of Gifted Education is explored through a Culturally and Linguistically Diverse (CLD) lens. Research questions explored race and its affect on the Cognitive Abilities Test, CogAT 7, and the significance of difference between racial groups at the 90th and 95th percentiles. The results of this study indicate: (a) CogAT 7 test scores are affected by race, (b) 95th percentile gifted referrals were significantly different across race, (c) 90th percentile gifted referrals were significantly different across race, (d) when comparing 90th and 95th percentile referrals, there was a significant difference, and (e) gifted referrals at the 90th and 95th percentiles were significantly not equitable across race. Recommendations for further study include the development of district equity goals, second screening processes, and talent enrichment programs and/or strategies for CLD learners.
The form and content of this abstract are approved. I recommend its publication.
Approved: Connie L. Fulmer
IV


DEDICATION
For David and Mom.
If I had a degree to hand out, I would give it to my husband, David, for most supportive partner. You have stood by my side through the late nights, all-nighters, melt-downs, watched the kids, and loved me through this. Without your loving support, this would not be possible.
For Mother, who encourages me and teaches me that I can be and do anything.
v


ACKNOWLEDGEMENTS
I want to acknowledge the efforts of my support network through this process. It is because of the following people that I was able to complete this work.
My Dissertation Chair, Dr. Connie L. Fulmer. Without Dr. Fulmer’s guidance, I would have been lost in the vastness of “All but Dissertation” land. She helped guide and direct me, while holding me accountable to deadlines I set for myself. Thank you Dr. Fulmer. Thank you for your expertise and wisdom having been down this road before.
My Father, Editor Extraordinaire. Having written detailed reports from his days in the military, my father knows what it takes to be concise and clear in written communication. With his writing expertise, I am a better writer for it. Thank you Dad - for your efforts to make my drafts the best they could be.
Graduate Assistant and Sister from Another Mother, Dara Marin Prais. What can I say Dara? You are the other half of my brain. You have always understood me and for some reason things are always better after talking with you. Thank you for your time and dedication to mentor me through this process. You listened, guided, nudged, and generally taught me to be the best version of myself possible.
My fellow Doctoral Candidates. I could not have done this without you. Sarah Famularo, who studied with me at Kneaders and listened to the hardships of each week along the way. Jessica Slattery, who always helped me gain perspective and keep my eyes focused on my dreams. Jeremy Koselak, who kept me grounded in strong principles of school systems. Rob Thelen, who would send me funny texts to keep me light-hearted. And Jon Ail, who showed me that through trial and hardships, you can emerge victorious. Thank you to each of you. Having others go through this experience with me helped forge friendships I hope to continue well beyond this Doctoral program.
My mentor, Ashley Gehrke. She is an amazing example to me of the kind of leader I hope to be someday. She has taught me to put the “Gehrke Touch” on situations and weave it into conversations with others. Her sincere desire to serve others and collaborate resonates with me and I will be forever grateful for her guidance. Her leadership is instrumental in shaping who I aspire to be as a leader.
vi


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION.......................................................... 1
Background............................................................ 2
Achievement Gap, Opportunity Gap, and Excellence Gap.................. 4
The Problem of Practice............................................... 6
District Perspective........................................... 7
Conceptual Framework..................................................10
Research Questions................................................... 11
Research Assumptions..................................................12
Significance of Study.................................................12
Limitations and Delimitations.........................................13
Summary.............................................................. 13
II. LITERATURE REVIEW.................................................... 15
Systems of Power and Privilege....................................... 15
School Systems and Power and Privilege........................ 15
Historical Context............................................ 16
More Recent Policies.......................................... 18
Definitions of Giftedness.............................................19
Culturally and Linguistically Diverse Gifted Education................21
Gifted Identification with a Culturally and Linguistically Diverse Lens. 23
Colorado Department of Education...............................24
District...................................................... 25
vii


Cognitive Abilities Test, CogAT
25
Summary...............................................................28
III. METHODOLOGY.......................................................... 29
Research Design...................................................... 29
Site Selection................................................ 30
Subjects...................................................... 30
Variables..................................................... 30
Conceptual Framework................................................. 31
Cognitive Abilities Test...................................... 32
Racial Groups................................................. 32
Gifted Referral............................................... 32
Percentiles................................................... 32
Research Questions....................................................33
Instrument, CogAT 7...................................................34
Data Collection and Storage...........................................34
Data Analysis.........................................................35
Limitations and Delimitations.........................................36
Ethical Considerations................................................37
Participants...................................................37
Sponsors.......................................................37
Field of Research..............................................38
viii
Research Community
38


IV. RESULTS...............................................................39
Descriptive Statistics.............................................. 40
Valid and Missing Cases..............................................41
Testing for Normalcy.......................................... 41
Central Tendency by Race and Battery.......................... 42
Comparisons Across Administration Years Within Batteries.......43
Verbal Battery..........................................43
Quantitative Battery....................................45
Nonverbal Battery.......................................47
Summary..............................................................49
Research Question 1: Does race affect test scores from the
Cognitive Abilities Test, CogAT 7?.............................49
Verbal Battery.................................................50
Quantitative Battery.......................................... 51
Nonverb al B attery........................................... 53
Answering Research Question 1: Does race affect test scores
from the Cognitive Abilities Test, CogAT 7?..........................54
Research Question 2: What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above,
between racial groups?...............................................55
Research Question 3: What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above,
between racial groups?.............................................. 58
IX


Research Question 4: What is the significance of difference of gifted referral between racial groups, based on the CogAT 7,
from the 90th percentile to the 95th percentile?.................... 62
95th Percentile Referral by Race, Chi-Square Goodness-of-Fit..63
90th Percentile Referral by Race, Chi-Square Goodness-of-Fit..64
Overall Gifted Referral Rate Disaggregated by Race............66
Summary..............................................................67
V. DISCUSSION..........................................................69
Summary of Study.....................................................69
Findings............................................................ 70
Race and Test Scores.................................................70
95th Percentile Referrals.................................... 71
90th Percentile Referrals.................................... 73
Comparison Between 90th and 95th Percentile Referrals.........74
Conclusions..........................................................75
Implications and Future Research.................................... 77
Equity Goals................................................. 77
Second Screening as an Equity-Driven Alternative............. 79
Talent Enrichment Programs for High-Ability CLD Students......81
Summary..............................................................82
Final Thoughts.......................................................82
REFERENCES........................................................................84
x


APPENDIX
A. University of Colorado Approval of Research Letter...............92
B. District Research Approval Letter................................ 93
xi


LIST OF TABLES
TABLE
1. CMAS proficiency levels 2016-2017, English language arts, grade 3
CMAS proficiency levels 2016-2017, mathematics, grade 3..................... 3
2. CMAS % exceeded expectations 2016-2017, English language arts
and mathematics, Grade 3..................................................... 6
3. Disaggregated student enrollment, District K-12, 2015-2018................... 9
4. District enrollment compared to gifted identification,
disaggregated by ethnicity.................................................. 10
5. Chances of being poor in America.............................................21
6. Operationalizing variables...................................................31
7. Observations by battery 2016, 2017, and 2018 combined........................41
8. Testing for normalacy by battery 2016, 2017, and 2018 combined...............42
9. Descriptive statistics, all CogAT 7 battery scores by race,
including 2016, 2017, 2018 data..............................................43
10. Comparison of CogAT scores for administration years (2016, 2017, 2018),
verbal battery...............................................................45
11. Comparison of verbal battery CogAT scores for administration years, t-tests.45
12. Comparison of CogAT scores for administration years (2016, 2017, 2018),
quantitative battery.........................................................47
13. Comparison of quantitative battery CogAT scores for administration years.....47
14. Comparison of CogAT scores for administration years (2016, 2017, 2018),
nonverbal battery............................................................48
xii


15. Comparison of nonverbal battery CogAT scores for administration years........ 48
16. Population descriptives, Verbal battery, used for ANOVA testing.............. 50
17. Games-Howell post hoc comparing verbal CogAT scores across race...............51
18. Population descriptives, Quantitative battery, used for ANOVA testing.........52
19. Games-Howell post hoc comparing quantitative CogAT scores across race........52
20. Population descriptives, Nonverbal battery, used for ANOVA testing...........53
21. Games-Howell post hoc comparing nonverbal CogAT scores across race.......... 54
22. Relative frequency of gifted referral (95th percentile) by race for each
CogAT battery.................................................................56
23. Relative frequency of gifted referral (95th percentile) by race.............56
24. Population Descriptives, 95th Percentile Gifted Referrals,
Used for ANOVA Testing........................................................57
25. Games-Howell post hoc comparing 95th percentile gifted referral across race.58
26. Relative frequency of gifted referral (90th percentile) by race for each
CogAT battery.................................................................59
27. Relative frequency of 90th percentile gifted referral by race................60
28. Population descriptives, 90th percentile gifted referrals,
used for ANOVA testing....................................................... 61
29. Games-Howell post hoc comparing 90th percentile gifted referral across race.61
30. Number, mean, and standard deviation of stratified random sampling,
one-way MANOVA................................................................62
31. Gifted referral by race at the 95th percentile, Chi-Square Goodness-of-Fit...63
32. Gifted referral by race at the 90th percentile, Chi-Square Goodness-of-Fit...65
xiii


33. Percent of gifted referral by race at the 90th and 95th percentiles...............67
34. Relative frequency of 90th and 95th percentile gifted referral by race........... 73
35. Percent of gifted referral by race at the 90th and 95th percentiles...............74
36. Colorado state student population trends 2010-2017............................... 76
37. District enrollment compared to gifted identification,
disaggregated by race............................................................78
xiv


LIST OF FIGURES
FIGURE
1. Inclusive excellence adaptive framework.................................. 8
2. Conceptual framework including comparison between 90th and
95th percentile referrals............................................... 11
3. Referral processes as outlined by Colorado Department of Education.... 24
4. Conceptual framework................................................... 31
xv


CHAPTER I
INTRODUCTION
As a Gifted Specialist, who has worked in varying ethnic and socio-economic environments, I am deeply concerned about gifted opportunity inequities across demographic groups. I see financially-limited parents struggling to make a living; wanting the best for their children, yet unsure how to provide that opportunity. There are children yearning to learn, yet struggling to “catch-up” to their same-age, White peers. After receiving my Masters in Culturally and Linguistically Diverse (CLD) education, I realized, although important, implementing CLD strategies was not enough. School systems need to change. Access to gifted education needs to systematically shift to provide equitable opportunities in gifted education. This study focused on recent screening practices for gifted referral within an urban to suburban school district in Colorado. For the purposes of this study, the school district name has been de-identified and given a pseudonym of District.
Discussion within District abounded regarding the underrepresentation of diverse populations in gifted education. With recent changes at the Colorado Department of Education (CDE) in respect to gifted identification (CDE, 2016), there had been discussion around screening and referral processes, particularly for diverse populations. The scope of this current study was to determine if equitable referral representation existed across racial groups from the Cognitive Abilities Test (CogAT), Form 7 (Lohman, 2012), a universal screener used in District. In this study, universal screener refers to a given test being administered across a group of students, in this case all second graders. It is not to be confused with the CogAT screening test, which includes three subtests. The test used was the full nine subtest version of CogAT 7. Since the school district used 90th percentile cutoffs for referral purposes, the researcher was
1


interested in comparing the 90th percentile cutoff with the 95th percentile cutoff; which, if any, was a better representation of the student body.
Background
Despite the Civil Rights movement’s contribution to school desegregation, persistent inequities in the United States school system continue to exist (Ford, 2011). Prejudice is usually based on very limited perspective (Tatum, 2013). Prejudice is defined here as, “preconceived judgments toward people or a person because of race, ethnicity, nationality, religion, socioeconomic status, gender, age, disability, and other stereotypes” (Ford, 2013a, p. 63). The phrase minoritized is used in this study to denote, “people who endure mistreatment, and face prejudices that are enforced upon them because of situations outside of their control” (Odyssey, 2016). Minoritized populations here are Black, Hispanic, and American Indian populations.
Within school systems, limiting perspective plays out with minoritized groups’ over identification for special education and underrepresentation in gifted programs. This discrepancy is related to the achievement gap, a statistical analysis of performance-based assessments determining how well a student does in school, usually between minoritized and/or low socioeconomic students and their White and/or Asian peers (NEA, 2017). The continued underrepresentation of minoritized groups in gifted education and the overrepresentation in special education, “strongly indicates systemic problems of inequity, prejudice, and marginalization within the education system” (Sullivan, 2011). In her book, Multicultural Gifted Education, Donna Ford (2011) cites statistics in the Condition of Education 2010 report (NCES, 2010) stating that while minoritized populations continue to increase, there has not been a significant change in gifted identification. White students are continuing to be over identified, while minoritized groups are underrepresented.
2


Data from the 2016-2017 school year, Table 1, from CDE (2017) indicate student proficiency levels in grade three for English Language Arts (ELA) and Mathematics.
Table 1
CMAS Proficiency Levels 2016-2017, English Language Arts, Grade 2
Demographic # Valid Scores Mean Scale Score Standard Deviation % Did Not Yet Meet Expectations % Partially Met Expectations % Approaching Expectations % Met Expectations % Exceeded Expectations
American 418 720 39 32.3 23.0 22.2 21.5 1.0
Indian
Asian 1,950 751 41 12.7 11.9 21.9 46.4 7.1
Black 3,031 722 38 31.6 21.1 22.5 23.6 1.2
Hispanic 21,215 723 37 29.0 22.8 23.8 23.2 1.1
White 33,907 748 38 11.5 14.2 23.9 45.8 4.6
Multi-Racial 2,906 745 39 13.2 16.1 25.1 41.5 4.2
CMAS Proficiency Levels 2016-2017, Mathematics, Grade 2
Demographic # Valid Mean Standard % Did Not % Partially % % Met % Exceeded
Scores Scale Deviation Yet Meet Met Approaching Expectations Expectations
Score Expectations Expectations Expectations
American 422 724 37 26.5 25.8 24.2 17.5 5.9
Indian
Asian 1,981 757 39 7.7 12.1 19.4 40.3 20.5
Black 3,034 723 35 26.7 24.1 26.3 20.1 2.8
Hispanic 22,954 725 34 23.7 25.7 26.4 21.3 2.9
White 33,931 749 35 8.3 14.8 26.2 38.4 12.4
Multi-Racial 2,915 744 37 11.8 16.6 26.9 34.1 10.5
3


American Indian, Black, and Hispanic students score significantly lower than their Asian and White counterparts. This trend in lower performance, known as the achievement gap, for ethnically diverse students is seen across the nation, not just in Colorado (Ford, 2013b; Miller, 2004; Olszewski-Kubilius & Thomson, 2010; Plucker & Peters, 2016). With these trends, the need for equitable practices has become a necessity (Ford, 2012).
Achievement Gap, Opportunity Gap, and Excellence Gap
The phrase achievement gap can be controversial, as it can denote more than just a lack of performance of minoritized groups (Perry, Steele, & Hilliard, 2003). Hillard (2003) explains, “Framing the problem this way [as achievement gap] is itself problematic. Importantly, it establishes European average achievement as the universal norm” (p. 137). While the researcher advocates for change in using the term achievement gap, its use is well-documented in the literature and therefore, for the purposes of this study, will be used when referring to the discrepancies in academic performance. In addition, the researcher uses the term achievement gap to indicate the discrepancy of achievement scores without placing a value on various demographic groups.
The discrepancies in proficiency levels begin well before the third grade (Ford, 2011; Miller, 2004). They exist in opportunities outside of school, to which, low socio-economic youth may not have access (Miller, 2004). This is referred to as the opportunity gap. The use of the term opportunity gap refers to “the unequal or inequitable distribution of resources and opportunities—while achievement gap refers to outputs—the unequal or inequitable distribution of educational results and benefits” (Great Schools Partnership, 2013, p. 1). It is the,
“cumulative differences in access to key educational resources” (Darling-Hammond, 2013, p.
4


77). When denoting the cause to this achievement gap, the researcher will use the phrase opportunity gap.
The opportunity gap can also be thought of as an educational debt, wherein groups of people have been denied access to quality education for extended periods of time (Ladson-Billings, 2013). Ladson-Billings discussed this educational debt as one of moral importance in a nation where groups of people have been oppressed and then released. For example, Ladson-Billings cited President Lyndon B. Johnson’s policy on Affirmative Action, stating, “it is unfair to keep people shackled for centuries, unshackle them, and then expect them to compete against those who have never known such restrictions” (p. 17). Therefore, there is a moral obligation for those making educational structures and systems to look through the cultural lens of equity. Rothstein (2013) stated that in order to address the issues facing school systems regarding the opportunity gap, school improvement and efforts that focus on socioeconomic inequalities need to be combined.
One other gap that needs to be addressed here is the excellence gap, those gaps in high-performance between demographic groups on cognitive and academic testing (Plucker & Peters, 2016). Not only do discrepancies exist within basic proficiencies, but also in advanced scores (Plucker & Peters, 2016). Take, for example, the data shared regarding CDE (2017) third grade results for ELA and Mathematics (see Table 2).
Table 2 indicates, especially in mathematics, a clear excellence gap discrepancy; from 2.8 to 5.9 percent for Black and Hispanic, and American Indian populations to 12.4 and 20.5 percent in White and Asian populations respectively. The gaps in achievement are not localized to achieving proficiency; they are pervasive, even in gifted populations. This systemic discrepancy leads to further gaps when, because of excellence gaps, minoritized groups are left out of gifted
5


programming. Majority groups end up achieving higher, as minoritized groups fall further behind. Data suggests that the United States’ educational system perpetuates the gaps through current screening processes for gifted referral and evaluation (Ford, 2011).
Table 2
CMAS % Exceeded Expectations 2016-2017, English Language Arts and Mathematics, Grade 2
Third Grade English Language Arts Third Grade Mathematics
Demographic % Exceeded Expectations Demographic % Exceeded Expectations
American Indian 1.0 American Indian 5.9
Asian 7.1 Asian 20.5
Black 1.2 Black 2.8
Hispanic 1.1 Hispanic 2.9
White 4.6 White 12.4
Multi-Racial 4.2 Multi-Racial 10.5
In a study, focused on social class standing in relation to educational access, Anyon (1980) relates observations and themes that emerged from a study done with students attending various schools with varying social classes. He relates that the lower the class standing at a given school, the more rote memorization and menial tasks were observed. They related this to preparing the students for a workforce that would keep them in their current social class. As one moved up in social class as a school, there were more resources available and accessed by students. There was also a marked increase in creativity and expectations of teachers. In conclusion, this article relates higher class to higher expectations of students, and therefore, higher performance and autonomy. The researcher argues that inequitable access is at the heart of underrepresentation in this case and generally in gifted education.
The Problem of Practice
All the changes in standards, such as Common Core, and re-evaluations of laws (i.e., Elementary and Secondary Education Act and No Child Left Behind) have done little to effect
6


change in the achievement gap (Plucker & Peters, 2016). The Every Student Succeeds Act (ESSA), authorized by President Obama in 2015, authorized funds for the Javits Gifted and Talented Students Education Program. It also allowed Title I funds to be used for programming needs of gifted students (Plucker & Peters, 2016). In order to continue moving toward more equitable gifted representation, access to gifted programming needs to be addressed (Plucker & Peters, 2016). One of the ways to do this is through gifted referral, which leads to identification. Without changes to gifted referral, gifted programming continues to be overrepresented by White and Asian students; while Black, Hispanic, and American Indian populations continue to be underserved.
District Perspective
The school district, as a whole, is striving for excellence for all its students. The Office of Inclusive Excellence, states its purpose is to, “serve as a cultural liaison between district and its diverse community, working to promote racial, ethnic, linguistic, religious and cultural understandings . . .” (Office of Inclusive Excellence, 2018, p. 2). In recent years, there has been a systematic push for inclusive excellence through required professional development, Beyond Diversity, for each staff member. This professional development increases staffs racial self-awareness, thus heightening sensitivity for diverse populations.
Part of the mission of the Office of Inclusive Excellence is to promote the following framework (see Figure 1). District is fulfilling its goals of Racial Consciousness and Cultural Competence through professional development, such as Beyond Diversity, and training with Dr. Stembridge, a multicultural researcher. Capacity of the district is being built to encompass a more culturally responsive environment.
7


Figure 1. Inclusive excellence adaptive framework.
Along with the work around Beyond Diversity, District has also partnered with Dr. Yemi Stembridge, who provides “technical assistance for school improvement with a specific focus on equity” (My Reflection Matters, 2018, p. 1). This work is tracked in various formats, including Twitter, where Dr. Stembridge (2017) related, “Racial consciousness is the foundation of equity work because our country has deeply problematic and unresolved issues with race” (Retrieved from https://twitter.com/DrYemiS). The inclusive excellence work happening courageously in District needs to expand in efforts for more inclusive programming within gifted education.
One of the goals from the Office of Inclusive Excellence includes, “promoting student access to rigorous learning opportunities” (Office of Inclusive Excellence, 2018, p. 8). One of the varied ways to fulfill this goal would be to evaluate current pathways to gifted education.
The ultimate goal would be to provide additional access to gifted programming for our diverse gifted learners, through updating referral systems with an equity lens. Table 3 details the trends
Systems and Structures of Opportunity
Racial Cultural
Consciousness Competence
Culturally
Responsive
Education
8


in current student enrollment in District. Student enrollment is disaggregated by ethnic groups; grouped by minoritized (American Indian, Black, and Hispanic) and majority (White and Asian). Asian is included in the majority designation, even though considered a minority, because they are traditionally overrepresented in gifted populations (Plucker & Peters, 2016; Worrell, 2014).
Table 3
Disaggregated Student Enrollment, District K-l2, 2015-2018
Ethnic Groups 2015-2016 2016-2017 2017-2018
American Indian /Black/Hispanic # of students in ethnic groups Total # of students enrolled 16,051 51,663 16,240 51,678 16,752 53,059
Percent of ethnic group compared to total population 31.07 31.43 31.57
Multi-Racial # of students in ethnic group Total # of students enrolled 3,216 51,663 3,424 51,678 3,665 53,059
Percent of ethnic group compared to total population 6.22 6.63 6.91
White/Asian # of students in ethnic group Total # of students enrolled 32,396 51,663 32,014 51,678 32,642 53,059
Percent of ethnic group compared 62.71 61.95 61.52
to total population
(Office of Assessment and Evaluation, 2018)
As evidenced by comparison data shown in Table 4, current and historical gifted populations do not equitably represent the District student population; White and Asian groups are disproportionately overrepresented, while Black, Hispanic, and American Indian groups are underrepresented (Ford, 2013b; Office of Assessment and Evaluation, District, 2018). Current research that identifies the reasons for this inequity is pervasive across the nation (Borland, 2003; Borland, 2009; Felder et al., 2015; Ford, 1998; Ford, 2011; Ford, Moore, & Scott, 2011; Ford, 2013b; Whiting & Ford, 2009; Worrell, 2014).
9


Table 4
District Enrollment Compared to Gifted Identification, Disaggregated by Ethnicity
Ethnic Group 2016 2017 2018
American Indian/ Hispanic/ % District Population 31.07 31.43 31.57
Black % District Gifted Education 18 17 17
% Underrepresentation 42 46 46
Multi-Racial % District Population 6.22 6.63 6.91
% District Gifted Education 6 6 6
% Underrepresentation 3.5 9.5 13.2
White/Asian % District Population 62.71 61.95 61.52
% District Gifted Education 76 77 77
% Underrepresentation 0 0 0
From these data, it is evident that excellence gaps exist and are similar to Colorado state and national data on underrepresentation of diverse populations. Through this study, the researcher intends to identify equitable referral of diverse populations.
Conceptual Framework
District uses the Cognitive Abilities Test 7 (CogAT 7) as a cognitive measure of student ability. The researcher was interested in determining whether race affected gifted referral in District, based on the use of CogAT 7. Figure 2 represents the conceptual framework of this study. The conceptual framework elements include: CogAT, racial groups, gifted referral, and percentiles. These are operationalized in Chapter III. It is sufficient to include the overview of the conceptual framework here.
10


Gifted Referral
(95th percentile or above)
Comparison of Percentile Gifted Referral Levels
Which, if any, is an equitable representation of district demographics.
Gifted Referral
(90th percentile or above)
Figure 2. Conceptual framework including comparison between 90th and 95th percentile referrals.
Research Questions
The research questions explored in this study include:
1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7?
Ho = Race does not affect test scores from the Cognitive Abilities Test.
Hi =Race does affect test scores from the Cognitive Abilities Test.
2. What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups?
Ho = There is no difference in gifted referral between racial groups at the 95th percentile and above.
Hi = There is a difference in gifted referral between racial groups at the 95th percentile and above.
3. What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups?
11


Ho = There is no difference in gifted referral between racial groups at the 90th percentile and above.
Hi = There is a difference in gifted referral between racial groups at the 90th percentile and above.
4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?
Ho = There is no significant difference in gifted referral between racial groups when using the 90th percentile compared to the 95th percentile.
Hi = There is a significant difference in gifted referral between racial groups when using the 90th percentile compared to the 95th percentile.
Research Assumptions
The research assumptions of this study include: (a) the instrument, CogAT 7, will elicit valid and reliable scores, (b) the data collected will be from second grade screening processes in District, and (c) screening for gifted referral elicits more equitable referrals than teacher referral alone.
Significance of Study
This study was created to compare referral processes across demographics using the Cognitive Abilities Test, Form 7 (Lohman, 2012), a universal screener used in District. In the end, the researcher compared screening processes at the 90th percentile and above compared to that the 95th percentile and above to identify any difference in representative referral demographics for gifted evaluation. The study was intended to be used when making decisions around using percentile cut-offs for gifted referral.
12


Limitations and Delimitations
Since this study was conducted within a school district, it is limited to that school district. This study can be replicated through the use of similar methodology, but was not intended to be generalized to a larger population. The limitations are defined as the restrictions not within the researcher’s control, while the delimitations are limits the research has placed on the study (Baltimore County Public Schools, 2017). Limitations to this study include (a) this study is not intended to be generalizable to a greater population, as it was done within District and (b) this study is not seeking to determine causation, only relation. However, this study might be useful as a framework for other district equitability studies. The following are delimitations of the study (a) data is narrowed to 2nd graders administered the CogAT 7 during the 2015-2016, 2016-2017, and 2017-2018 school years, (b) students with invalid scores in one or more batteries will be eliminated from the data set, and (c) students whose parents opted out of testing will not be included in the data set.
Summary
This chapter discussed underrepresentation of minority groups namely: American Indian, Hispanic, and Black in current gifted programs across the nation, state, and district. The constructs of achievement gap, opportunity gap, and excellence gap are discussed and operationalized. Achievement gap is defined as the discrepancy between minority (American Indian, Hispanic, and Black) and their majority counterparts (White and Asian). Opportunity gap is defined as the discrepancy of opportunities available to minority groups compared to their majority counterparts. Finally, the excellence gap is defined as the discrepancy of achievement scores between minority and majority groups. These gaps lead to underrepresentation in gifted populations.
13


The study was situated in District, an urban to suburban school district in the western United States. The equity work in District is described, including collaborative efforts with Dr. Yemi Stembridge. The discrepancy of demographic representation is presented. A conceptual framework for comparing the current gifted referral screener, CogAT 7, at the 90th percentile and the 95th percentile is established. A comparison model is presented for the study focus. The significance of the study is defined as establishing equitable gifted referral practices that are inclusive to diverse, traditionally underserved populations. Research questions, assumptions, limitations, and delimitations are outlined and described.
14


CHAPTER II
LITERATURE REVIEW
In this review, recent literature is explored on giftedness through a culturally and linguistically diverse lens. In order to set context for this diversity lens, the history of systems of power and privilege is also related. Four major themes emerged from the literature review and form context and meaning around the current study. These themes include: (a) definitions of giftedness, (b) culturally and linguistically diverse (CLD) gifted education, (c) gifted identification with a CLD lens, and (d) Cognitive Abilities Test (CogAT).
Systems of Power and Privilege
Power and privilege exist in every part of our world. Since we are social beings and interact with others around us, power and privilege are inherent in society (Gallagher,
2003). Power and privilege are defined as (a)power being the ability to choose the norm or what normal looks and acts like and (b) privilege as being part of the dominant group that chooses, regardless of whether or not personally participating in the decision (Johnson,
2006). These two ideas work together to create stratification and difference in social systems. In this section, the system of school is defined through historical policies, structure, and norms as they relate to racial identity. Then, institutionalization of privilege and oppression is discussed and its implications on society at large and specifically within a school system for gifted learners. School Systems and Power and Privilege
The NAGC 2012-2013 State of the Nation reports that half of the states are citing more inclusive definitions of giftedness, only five included culturally diverse populations, five included students from low-socioeconomic (SES), three included English language learners (ELLs), and only two included those with disabilities (NAGC, 2013). Clearly, with only a few
15


states reporting conditions for minority groups, there is need for diverse reform in gifted education. If there are not definitions that are inclusive, how are systems searching for marginalized groups to ensure inclusion in gifted programs?
Historical Context
The fight for social justice has been around for centuries, beginning with the ideas of Socrates’ and Aristotle’s ideas of justice. They searched for the meaning of justice and how to define it (Boyles, Carusi, & Attick, 2009). For the purpose of this study, it is enough to state that over the centuries definitions of social justice have been redefined. Yet, equity remains at the ecenter of it all. This literature review is focused on the development of social justice over the last fifty years, specifically on social justice in gifted education.
The legal end of racial segregation within school systems is evident in the landmark case of Brown us. the Board of Education in 1954. This case marked a step forward in systemic change. However, with it came a restructuring that resulted in great controversy and push-back (Boyles, et al., 2009). Even though schools were integrated, how integration was to be implemented differed across the country. Although this was one step for racially diverse groups of students, it was not until the 1970s that federal mandates for special education were implemented. Prior to the 1970s, states were allowed significant freedom regarding enrollment, often marginalizing disabled and racial groups (Martin, Martin, & Terman, 1996). This resulted in refusal of services and misplacement of students (Martin, et al., 1996).
In addition to the racial discrepancies in services for marginalized individuals, it was not until the 1950s that national advocacy organizations for gifted learners began to have a place on the educational stage. The National Association for Gifted Children (NAGC) was established in 1954. Shortly after, in 1958, the National Defense Education Act was passed, the first of federal
16


policies in relation to gifted education. This enacted policy to provide funding to improve school systems and promote secondary education (Encyclopedia Britannica, 2016). The Equality of Educational Opportunity report (Coleman et al., 1966) brought to light the wide-spread achievement gaps between racial groups. This seminal work re-emphasized the vast achievement and opportunity gaps in racial performance on standardized tests, that only widened as the grade level increased.
The 1970s brought reforms encouraging states to establish a working definition of giftedness and the inclusion of the Office of Gifted and Talented within the US Office of Education. In 1983, a pivotal report was submitted to the United States Secretary of Education, T. H. Bell, called A Nation at Risk (National Commission on Excellence in Education, 1983). In this report, the authors call out the lack of high standards for children across the United States. They also state that the vast majority of gifted students were not reaching their identified potential. The underperformance and laissez-faire attitude found in school systems, if unchecked, would lead to economic disaster (National Commission on Excellence in Education, 1983). This report led to higher federally-mandated standards, especially in high schools. It brought much needed reform to school systems, including research-based programs for studying giftedness. By the turn of the century, much had been done to establish definitions for giftedness. However, programming was a problem as each state, district, and even school interpreted programming options in a wide spectrum of services. Another seminal report came in 2004, with A Nation Deceived: How Schools Hold Back America’s Brightest Students. Within this report, the authors conclude that because America is set on age-level grades, gifted students will be continually marginalized because they cannot accelerate at their own pace (Colangelo, Assouline, & Gross, 2004).
17


More Recent Policies
More recent policies include the No Child Left Behind Act of 2002 and the Every Student Succeeds Act of 2015. The No Child Left Behind Act (NCLB) showed that the achievement gap between racial groups continues (Jennings & Rentner, 2006). Although the intention of NCLB was to highlight the progress and gaps in student learning, it became increasingly difficult for schools to fulfil the prescriptive requirements. Gifted education also suffered, as national efforts and state funding focused on students below proficiency (Beisser, 2008; Jolly & Makel, 2010). In 2012, the Obama administration began granting leniency on NCLB requirements in exchange for plans for state-designed plans for equity work and high standards for all (Malin, Bragg, & Hackmann, 2017).
With the enactment of Every Student Succeeds Act (ESS A) in 2015, there was greater focus on equity work focused on disadvantaged and high-need students (Egalite, Fusarelli, & Fusarelli, 2017). It brought about a call for high standards preparing all students for college and career readiness (Malin, Bragg, & Hackmann, 2017). With ESSA, came funding for the Javits Gifted and Talented Students Education Act, which is the only federally funded gifted program (NAGC, 2018). The Javits fund was originally established in 1988, as part of the Elementary and Secondary Education Act, and is dedicated to providing research regarding gifted students (Senate, 1987). With the reauthorization of the Javits Fund, through the ESSA and the current work done through the NAGC, there is great advocacy in the United States currently for progression in serving underrepresented populations, namely Black and Hispanic groups. This advocacy and work now needs to become more localized to school systems.
18


Definitions of Giftedness
Now, with increasingly diverse populations, definitions for gifted education have become outdated. Before, definitions of giftedness were based on myths regarding what giftedness is and upon theory lacking in research (Missett & McCormick, 2014; Sternberg, Jarvin, & Grigorenko, 2011; Worrell, 2014). Having a solid research-based practice for gifted referral is essential to a well-functioning program, as it provides access for gifted curriculum (Missett & McCormick, 2014).
The debate over what giftedness means is vast and has been going on for decades. There is the IQ-based conception of giftedness, with its roots in Louis Terman’s (1926) study of giftedness, where Terman studied students with IQs of 140 and above. This study affected the way people viewed giftedness, even persisting into the present (Missett & McCormick, 2014; Subotnik, Kassan, Summers, & Wasser, 1993). This concept of giftedness, based on IQ, has been criticized as marginalizing racial groups, as well as leaving out creativity and twice-exceptionality (Missett & McCormick, 2014; Reis & Renzulli, 2011).
In recent years, the National Association for Gifted Children (2016) reported that-A common view that gifted students do not need specialized services contribute to a vast disparity of programs and services across states and often within states, leaving many high-ability students without the supports they need to achieve at high levels, which is a disservice to them and to the nation, (p. 1)
In an interview with M. Rene Islas, the Executive Director of NAGC and S. Dulong Langley, a NAGC board member getting her doctorate in underrepresented gifted populations, Langley stated, “Without federal mandates, states vary by how they define giftedness, their criteria for identifying students who would benefit from gifted services, and their program services
19


available” (Langley, 2015, Interview). Another research report conducted by Mccain and Pfieffer (2012) reported that even though there are half the states reporting specialized identification for underrepresented populations, “these procedures are not identified” (p. 78).
Without identification procedures being explicit, there is too much room for interpretation of the state laws and how each location identifies students. This, in turn, leads to problems with ensuring best practices when screening for gifted referral (Missett & McCormick, 2014). One persistent cut-off score that is constent to gifted identification is the 95th percentile (Mccain & Pfeiffer, 2012). This is systemically used as the qualifying percentile for many states. This 95th percentile holds to the IQ-based ideology, discussed previously, and continues to be congruent with the most research-based theory around giftedness (Lohman & Korb, 2006; Sternberg et al., 2011). In the last thirty years, theories of giftedness have expanded from an IQ-only based pedagogy to include: task commitment, motivation, creativity, multiple intelligences, talent development vs. natural ability, and practical intelligence (Ford & Thomas, 1997; Mccain & Pfeiffer, 2012; Reis & Renzulli, 2011; Sternberg et al., 2011; Subotnik et al., 1993).
Viewing giftedness not just as a set ability level, but also as potential for achievement, is important as issues of underrepresentation are addressed. This, however, requires more work and time to create protocols in identification of and programming for Culturally and Linguistically Diverse (CLD) gifted learners. Through Colorado Department of Education (CDE), the members of the Exceptional Student Services Unit (2016) state that, according to compliance with the Exceptional Children’s Educational Act, identification for gifted ability is only one of the ways a child can be identified. They can also be identified through aptitude, which they define as, “exceptional capability or potential in any academic content areas” (p.
99). This aptitude identification can happen with either demonstration of, “advanced level on
20


performance assessments or 95th percentile and above on standardized achievement tests” (Exceptional Student Services Unit, 2016, p. 100). There is room for alternate CLD student identification, through aptitude vs. ability identification; however, the methodology used is unclear. The issue is not whether alternative assessments or identification processes for underrepresented populations should be used; that much is clear (Naglieri & Ford, 2005). What is not clear is what those processes should look like and how to embed them in current practices.
Culturally and Linguistically Diverse Gifted Education
Within the literature, the issue surrounding underrepresentation of gifted minority groups, namely Black, Hispanic/Latino, and American Indian, is well-documented (Plucker, Burroughs, & Song, 2010; Worrell, 2014). For the context of this study, gifted underrepresentation is defined as the disproportionality of ethnically diverse students (Black, Hispanic/Latino, and American Indian) in gifted education when compared to the same population in relation to the whole (National Association for Gifted Children, 2008; Worrell, 2014). In other words, the gifted population should similarly match the demographics of the whole student body.
Race is not the only factor when speaking of cultural and linguistic diversity in gifted education. There is a great deal of overlap with race and class standing, as evidenced by the data in Table 5 on the likelihood of being poor in America (Mantosis, 2013).
Table 5
Chances of Being Poor in America
Chance of being poor Parents in home Race
1:10 2 White
1:5 1, female White
1:5 2 Hispanic
1:3 1, female Hispanic
1:4 2 Black
1:3 1, female Black
21


Not only is there marginalization of racial groups, but this marginalization makes it difficult for working class families to change their status. Mantosis (2013) explains that, “Class standing has a significant impact on chances for educational achievement” (p. 155). Most school systems use subjective referral systems to identify gifted students, which systems work from middle-class values and norms (Felder, Taradash, Antoine, Ricci, Stemple, & Byamugisha, 2015). There is not a lot of consideration to account for poverty. Many working-class families face dilemmas, such as (a) more hours at work, (b) latchkey children, and (c) increased health problems due to work stress and pressures leading to increased health car costs (Collins & Yeskel, 2013). These problems lead to less time and energy to help students with homework, to explain and problem-solve, and to be involved in a child’s education. Whereas, upper middle-class and upper class families have the resources available to hire tutors, attend parent nights and school nights, and generally appear more involved in their child’s education. Working families can also have the added barrier of navigating two languages: their native language and English. This barrier creates difficulty in advocating for their child, making connections with school employees who only speak English, and navigating the school system. With the added stress of trying to navigate a new language system as well as a school system that is foreign, English Language Acquisition (ELA) families can struggle to get the appropriate education for their child. This is not because they don’t want the education or lack of advocacy, but because a system is scant on resources also does not speak their native language (Hos, 2016).
In addition to economic barriers, there can be deficit mindset in school staff. “Deficit thinking (e.g., stereotypes, biases, low and negative expectations) compromises teacher referrals, as well as how nomination forms and checklists are completed, the test and instruments selected, the specific cut-off score selected, and the ultimate placement criteria and decision” (Ford,
22


2013b, p. 10). As educators begin to examine personal biases and become more asset-focused, they will become better at referring students for special education services. They will appreciate the nuances of gifted tendencies more and teachers, “so many of whom are wedded to stereotypical and traditional notions of gifted grounded in high test scores and good behavior,” (Ford, 2013b, p. 8) will become more aware of differing characteristics of diverse giftedness. As teachers are trained and become more familiar with gifted characteristics of students that are (a) CLD, (b) disabled, and (c) from low socioeconomic status (SES), they will be better prepared to engage in more equitable referral systems (Felder, et al., 2015).
Gifted Identification with a Culturally and Linguistically Diverse Lens The battle for equitable representation in gifted education has been a long one. Frasier (1991) states that a major challenge to the underrepresentation within gifted populations is the lack of identification of diverse populations. She goes on to state that referral-only based models preclude students from diverse populations. Frasier advocates for data from multiple sources in order to identify gifted minoritized students.
This social dilemma of underrepresentation in gifted programming is rooted in the systems of power and privilege discussed earlier. In an interview conducted with Dr. Mary M. Frasier, founder of the Torrance Center for Creative Studies, Dr. Frasier states,
You are dealing with a very sensitive social problem [underrepresentation of minority students in gifted education]. There is no one who would tell you, ‘Well, the reason that these kids aren’t in the program is because I am prejudiced. I discriminate. I am biased in my opinion about giftedness’.. .people just won’t say that. (Grantham, 2002, p. 51) Whether intentional or unintentional, biases and systemic views of minoritized groups have excluded and continue to exclude access for many talented CLD students. For the current study,
23


the researcher is interested in looking at the systemic processes in place for equitable access to gifted identification and thus, access to programming. The next sections outline how the State Department of Education and District are striving to minimize excellence gaps for inclusive excellence.
Colorado Department of Education
As part of Colorado Department of Education (CDE), the Office of Gifted Education developed a Gifted Identification manual to be used throughout the state to guide administrative units at district and school levels. In this manual, it outlines (see Figure 3) that there should be varied methods of referral, including universal screening measures (Medina, 2016).
Figure 3. Referral processes as outlined by Colorado Department of Education.
Through the use of multiple entries for gifted referral, CDE is ensuring more inclusive pathways to gifted referral and identification, leading to more inclusion in gifted programming.
24


District
Within District, universal screening processes have been established at varying grade levels, including second grade (Office of Advanced Academics and Gifted Services, 2011). In second grade, the universal screener is the CogAT 7, a nationally norm-referenced cognitive test. From this test, students scoring at the 90th percentile or above are screened for further evaluation (Office of Advanced Academic and Gifted Services, 2018). The researcher was interested in determining if screening at the 90th percentile versus the 95th percentile overcomes some of the biases discussed earlier, namely the achievement gap, opportunity gap, and the excellence gap. Thus, yielding a more diverse gifted screened population, better matching District student population.
Cognitive Abilities Test, CogAT
The debate over equitable practices for identifying CLD gifted learners is long. Controversy exists between two major writers of cognitive tests; Lohman, author of CogAT, and Naglieri, author of NNAT. Lohman and Naglieri debate over best practices for identifying CLD gifted learners. The researcher examines some of the controversy here, but focuses efforts on describing CogAT and its utilization in context to District, as that is where the study is taking place and District uses the CogAT as a gifted screener in second grade.
Lohman (2005) explains that there are two types of giftedness: high-accomplishment and high-potential. Highly accomplished students are ones that are already performing high on standardized tests or academic tests. High potential students are ones that are not yet demonstrating their unique abilities in a particular domain. Lohman (2005) adds that these groups of students possess differing programming needs. He cites that using nonverbal tests for identifying English Language Learners for gifted programming should use more than “figural
25


reasoning abilities” (Lohman & Gambrell, 2012, p. 41). He advocates for verbal and quantitative reasoning batteries over a nonverbal-only test (Lohman & Gambrell, 2012). His argument is that there is unequivocal data demonstrating that the “predictors of achievement in reading, mathematics, social studies, and science are the same for White, Black, Hispanic, and Asian American students” (Lohman, 2005, p. 344). This is vastly different from the findings of Naglieri and Ford (2005, 2015).
Naglieri and Ford (2015) have long written about the discrepancies in gifted representation of minority populations, especially for Black and Hispanic students. Data from the Office for Civil Rights is clear in the underrepresentation of Black and Hispanic students in gifted education. Naglieri and Ford (2015) cite data from the Office for Civil Rights stating that, “Black students are the most underrepresented racial group, followed by Hispanic students” (p. 234). They state that, “On a daily basis, educators struggle with finding the most effective ways to both identify and serve gifted students who are not reaching their potential, as measured by tests or as perceived by teachers, counselors, or parents” (Naglieri & Ford, 2005, p. 34). Others have also reported regarding these gifted discrepancies of ethnically diverse students (Feldman, 2003; Fraiser, Hunsaker, Lee, Finley, Garcia, & Martin, 1995; McBee, 2006).
Lohman, Korb, and Lakin (2008) counter Naglieri and Ford (2005) when they conducted a study comparing cognitive measures including the Raven, NNAT, and CogAT. Their assumptions include that, “one cannot assume that nonverbal tests level the playing field for children who come from different cultures or who have had different educational opportunities” (p. 293). The authors state that nonverbal tests help to provide useful information, but caution against use of nationally normed data that are normed for different populations, saying that this
26


can be misleading. Although, later Lohman and Gambrell (2012) advocate use of local norms for increasing gifted referral from ethnically diverse populations.
As part of the ever developing and fine-tuning process, the CogAT test has been adapted recently to better meet the needs of ELLs. At the annual 2013 NAGC conference, Jim Nicholson, President of HMH-Riverside, announced changes to the structuring of CogAT, including less emphasis on the language components that can hinder second language learners. Nicholson stated, “The reduced language emphasis on CogAT increases access to enrichment programs for students from a variety of backgrounds” (Business Wire, 2013, p. 1). The researcher could not find published studies regarding the effects of this change on student populations being referred for gifted evaluation.
There is great controversy in the gifted field around what test and data should be used for CLD learners (Lakin & Lohman, 2011; Naglieri & Ford, 2005). What can be agreed on is that no one test or one method is going to guarantee that all students needing gifted referral will receive the referral (Greenfield, 1997; Helms, 1992; Naglieri & Ford, 2005). Given this, there are ways to improve referral practices to make selection more equitable. The researcher also agrees that, “Regardless of their linguistic and cultural background, any student who demonstrates a need for more demanding curricula should be challenged” (Naglieri & Ford, 2005, p. 34). The question is, how do systems effectively screen for the most equitable outcome? This study was developed to evaluate current practices for racial equitability and to determine if equitable numbers of students can be referred for gifted evaluation at the 90th percentile compared to the 95th percentile.
27


Summary
This chapter reviewed current literature regarding the four major themes of this study including: (a) definitions of giftedness, (b) culturally and linguistically diverse (CLD) gifted education, (c) gifted identification with a CLD lens, and (d) Cognitive Abilities Test (CogAT). It framed this work in literature and historical context of systems of power and privilege. This study focused on determining if equitable representation for gifted referral can be reached if the percentile cut-off at the 95th percentile is dropped to the 90th percentile. In this way, the researcher is looking to affirm current District practices or call for a change in policy. The battle for equitable gifted practices continues on the national, as well as local level. Striving to find most equitable practices for gifted referral in District can help other districts in their efforts to do the same.
28


CHAPTER III
METHODOLOGY
This chapter describes and outlines the methodology used in this study. The conceptual framework is reviewed from Chapter 1. The variables are operationalized. The method discussed includes a non-experimental, comparative study. Research design elements, including: site selection, subjects, and variables are described and operationalized. After that, data collection processes are outlined. Data analyses, including descriptive statistics, relative frequency, t-tests, ANOVA, MANOVA, and Chi-Square are used for this study. Finally, CogAT 7, the instrument being researched, is described with current reliability and validity.
This is a non-experimental, comparative study; examining the difference between racial group CogAT 7 scores. It is also a relational study; examining the relationship between racial group test scores and gifted referral. Ultimately, the researcher examined the relationship between racial designation and gifted referral at the 90th percentile and 95th percentile, through the use of the CogAT 7 as a screener.
Since this study is not manipulating variables and its purpose is to compare variables (race and CogAT 7) to test the hypothesis, a non-experimental, comparative study is appropriate (Johnson & Christensen, 2012). The focus of this study also lends itself to descriptive and relationship, which is part of the non-experimental design, as outlined by Johnson and Christensen (2012). The researcher’s purpose is to check for equitable referral representation across racial groups.
Research Design
Design elements of this study include site selection, subjects, and variables. The site selection was from a large Denver Metro school district. Subjects were selected from the current
29


screening processes of the Office of Advanced Academic and Gifted Services in District. Finally, variables are operationalized in terms of this study.
Site Selection
Site selection for this study was convenient, as the researcher worked in District as an Advanced Academic Specialist (formerly known as Gifted and Talented). Since the purpose of this study is to examine current practices for Gifted referral, the researcher desired to examine processes supporting referral of diverse populations by examining current practices.
Subjects
The target population is 2nd graders in District, because they take the CogAT 7 screener. The sampling approach is convenient, because it is within the school district where the researcher is currently employed. The data is disaggregated by race. All CogAT 7 scores, within the 2nd grade District population, will be pulled from Powerschool and used for testing the statistical significance of racial distribution. No other identifying information was gathered to protect the anonymity of students.
Variables
The variables compared and used in this study included: CogAT 7, categorical racial groups, and gifted referral. The CogAT 7 contains the three batteries: Verbal, Quantitative, and NonVerbal. The categorical data came from the District data, which is Powerschool. Finally, the gifted referral data was from the referrals at the 90th and 95th percentiles. These were measured using the procedures as described in Table 6.
30


Table 6
Operationalizing Variables
Variable How it will be measured
CogAT 7
independent
continuous
CogAT 7 has three batteries: (a) Verbal, (b) Quantitative, and (c) NonVerbal. Each of the batteries will be used separately in the screening process. A student scoring 95th percentile or higher on any of the batteries will be referred for gifted evaluation and further testing.
Racial
Groups
independent
categorical
Using Powerschool, each racial group will be categorized and grouped together for statistical analysis. The CogAT 7 will be disaggregated into racial groups for purposes of determining the percentage of students referred for gifted evaluation. These results will then be compared for statistical significance of scores.
Gifted
Referral
dependent
categorical
A student scoring 95th percentile or higher on any of the batteries will be referred for gifted evaluation and further testing. In this study, the 90th percentile cutoff will also be explored for gifted referral.
Conceptual Framework
District uses CogAT 7 as a cognitive measure of student ability. The researcher is interested in determining whether race affects gifted referral in District, based on the use of CogAT 7. Figure 4 visually represents the conceptual framework of this study.
Gifted Referral
(95th percentile or above)
Comparison of Percentile Gifted Referral Levels
Which, if any, is an equitable representation of district demographics.
Gifted Referral
(90th percentile or above)
Figure 4. Conceptual framework.
31


Cognitive Abilities Test
CogAT 7, “measures students’ learned reasoning abilities in the three areas most linked to academic success in school: Verbal, Quantitative and Nonverbal” (Houghton Mifflin Harcourt, 2017, p.7). It is currently used as a universal screener, at the end of 2nd grade, for gifted referral. A universal screener is a test given to a whole grade level, across a district, to identify students that need to be referred (CDE: Office of Gifted Education, 2016). If a child performs at the 95th percentile or above, on any given battery, they are referred for additional testing (District, 2017).
Racial Groups
Ethnic and racial categorical data is taken from PowerSchool, a District student database, and can be sorted accordingly. Racial populations including: White, Black, Hispanic, Asian, American Indian, and Multi-Racial will be included in this study.
Gifted Referral
Within District, the first step in identifying a student for giftedness is a referral and evaluation (District, 2017). This referral is taken from various sources, one being CogAT 7. Other sources for gifted referral include: parents, teachers, or other professionals working with the student i.e., doctors, psychologists, etc. (District, 2017).
Percentiles
Given national norms, the CogAT 7 uses age-based national norms and grade-based national norms (Houghton Mifflin Harcourt, 2017). District uses the age-based national norms at the 95th percentile and above for referral and identification (District, 2017). This evaluation at the 95th percentile, using national norms, and above meets the CDE guidelines for gifted
32


evaluation (CDE, 2016). This study compared screening at the 90th percentile or 95th percentile or above is more equitable for minoritized populations.
Research Questions
The research questions explored in this study include:
1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7?
Ho = Race does not affect test scores from the Cognitive Abilities Test.
Hi =Race does affect test scores from the Cognitive Abilities Test.
2. What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups?
Ho = There is no difference in gifted referral between racial groups at the 95th percentile and above.
Hi = There is a difference in gifted referral between racial groups at the 95th percentile and above.
3. What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups?
Ho = There is no difference in gifted referral between racial groups at the 90th percentile and above.
Hi = There is a difference in gifted referral between racial groups at the 90th percentile and above.
4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?
Ho = There is no significant difference in gifted referral between racial groups when using the 90th percentile compared to the 95th percentile.
33


Hi = There is a significant difference in gifted referral between racial groups when using
the 90th percentile compared to the 95th percentile.
Instrument CogAT 7
The CogAT 7 was chosen as the study instrument to be analyzed, using district demographic distributions, because it is used in District as a universal screener for gifted referral. The following describes CogAT 7 sampling and norming processes, along with reliability and validity.
CogAT 7 has a substantial standardization sample of 65,350 K-12 students, normed with achievement tests such as Iowa Tests of Basic Skills and Iowa Tests of Educational Development. Reliability for the sample was high, with split-half correlations between .80 and .92 for Verbal, Quantitative, and Nonverbal batteries. Global scores had split-half reliability values between .91 and .97 (Lohman, 2012; Warne, 2015).
The validity of CogAT 7 was tested through concurrent validity studies with the Naglieri Nonverbal Ability Test (NNAT-2) and Iowa assessments (Lohman, 2012). The correlation with the NNAT was done in a study sample of 149 second-grade students, with all three CogAT batteries correlating (r= .51) with the NNAT. The correlation scores with the Iowa assessments yielded the following ranges for differing grade levels (r = .42-.83 for Verbal, r = .30-.79 for Quantitative, r = .32-.69 for Nonverbal, and r = .40-.85 for the total CogAT 7 battery (Lohman, 2012; Warne, 2015). With the large standardization sample, high correlations, and concurrent validity studies, the CogAT 7 is a reliable and valid measure.
Data Collection and Storage
Sampling was of all second graders taking the CogAT 7; thus, resulting in a reliable study, as it was representative of the target population. A sample size of approximately 9,000
34


(some 3,000 for each of three years) helped reduce the margin of error. Students that are opted out of district assessments were excluded from this study. Also, students who recently moved into District and who have not taken the CogAT 7 were excluded from the study. Validity was ensured through use of CogAT 7, a valid measure of cognitive ability. Through the use of PowerSchool, students were grouped using the family-selected racial designation, providing valid racial groupings.
Data for this study, including three years of CogAT data (2015-2016, 2016-2017, 2017-2018) was collected from the Office of Assessment and Evaluation within District. The data had the following identifiers: (a) racial designation, (b) Verbal battery score, (c) Quantitative battery score, and (d) Nonverbal battery score. The data was grouped by school, but no other identifying information was given to the researcher. The data was stored electronically on a District secured computer, ensuring the security of the data. When the data was downloaded, it will be on a District secured computer. The data went through Excel and SPSS for data processing. These Excel and SPSS files were stored on the District secured computer. The data was shared in raw form only with District members assigned to this study or for University of Colorado staff assigned to this study as a committee member or in the role of methodology coach.
Data Analysis
Three years of data (school years: 2015-2016, 2016-2017, and 2017-2018) was collected. This provided a representative sample set of over 9,000 students (approximately 3,000 students per year), ensuring reliability of findings. Prior to any analysis testing of the data, descriptive statistics and t-tests will be run, checking for variance of the samples. The t-test data will be compared to ensure similar sample distributions prior to the ANOVA testing. Relative frequency will also be given of gifted referrals by race.
35


For research questions: (a) Does race affect test scores from the Cognitive Abilities Test, CogAT 7, (b) What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups, and (c) What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups, an ANOVA test will be conducted. Confidence level is set at 95%. Descriptive statistics will first be conducted focusing on CogAT 7 scores in each battery (Verbal, Quantitative, and Nonverbal), across the following races: American Indian, Black, Hispanic, Asian, White, and Multi-Racial. Statistical assumptions of CogAT 7 scores across races will be tested for each battery. The results of each battery, including demographic distribution, will be reported for analysis.
When answering research question 4 (Is there a difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?) a factorial ANOVA test will be conducted to compare main effects of race upon results at the 90th percentile and 95th percentile.
Limitations and Delimitations
Since this study is done within a school district, it is limited to that school district. This study can be replicated through the use of similar methodology, but is not intended to be generalized to a larger population. The limitations are defined as restrictions beyond the researcher’s control, while the delimitations are limits the research has placed on the study (Baltimore County Public Schools, 2017). Limitations to this study include: (a) this study is not intended to be generalizable to a greater population, as it was done within District and (b) this study is not seeking to determine causation, only relation. The following are delimitations of the study: (a) data is narrowed to 2nd graders administered the CogAT 7 during the 2015-2016, 2016-2017, and 2017-2018 school years, (b) students with invalid scores in one or more batteries
36


will be eliminated from the data set, and (c) students whose parents opted out of testing will not be included in the data set.
Ethical Considerations
Ethical considerations by the researcher include the following responsibilities: (a) to the participants, (b) sponsors, including the University of Colorado Denver and District, (c) the field of research, and (d) the research community (Govil, 2013). In the following sections, the researcher relates the plan for ethical considerations, as they pertain to this study.
Participants
In regards to the participants, the researcher will work ethically and with integrity with the data. The data is reported as de-identified, so the researcher will not have access to the student name. It will be disaggregated by race, as reported by the family to PowerSchool, the database in District. Besides the individual battery scores and the racial designation of the student, no other identifying information will be given to the researcher.
Informed consent is not required for this study, as it is de-identified and the sample size is large. There is no impact on workload of participants, as it is extant data. Privacy for participants is ensured through the de-identifying nature of the data. No psychological harm will be caused to participants, since the data is de-identified and the sample is large.
Sponsors
In order to best ensure ethical considerations for sponsors, the researcher is operationalizing all key terms in use during the study. For both University of Colorado Denver and District, the researcher is following proper protocol in obtaining permissions and signatures before conducting any research. The findings from this study are not intended to be prescriptive for District, rather this study is intended to give options for further decision-making. The
37


findings will be shared from data interpretations that have justifiable rational. Also, limitations for the study are included in the proposal to avoid ethical confusion. The nature of words used in the study will maintain professionalism and will seek to establish best practices within the district.
Field of Research
This study seeks to add to the current body of research on culturally and linguistically diverse gifted learners. The underrepresentation of gifted minorities is clear in the literature. What is not clear, however, is how to go about screening and referring our CLD learners for gifted evaluation. This study seeks to build on other researchers’ work and add to the field by providing one district’s methods for equitably working to refer and identify CLD gifted learners. Research Community
For the research community, the researcher will provide a technical research report, which will include methodology and findings. In order to maintain the integrity of educational research, highest standards will be used when reporting and conducting research. And finally, the researcher will use the highest integrity to gather, report on, and disseminate the findings of the research.
38


CHAPTER IV
RESULTS
This chapter reports the data analyses used for this study, including descriptive statistics, relative frequency, t-tests, ANOVA, and MANOVA testing. After conducting the MANOVA testing, analysis for goodness-of-fit through use of Chi-square testing was conducted to determine where the significance was situated. This testing was done to compare the number of gifted referrals of each race to the number that statistically should be referred at the 90th and 95th percentiles. The Chi-square testing was also used to compare the number of gifted referrals to the demographic make-up of the district. In the end, Chi-square testing is discussed as situated in the 90th and 95th percentiles.
The researcher obtained permissions from the district to include three years of data in this study. The following identifiers were used for this study: CogAT ID, administration year, student demographic, Verbal battery percentile, Quantitative battery percentile, and Nonverbal battery percentile. Student demographic designations were taken from PowerSchool, where parents and families determine demographic identity. It should also be noted that the designation labels for demographic groups are taken from the labels used in PowerSchool. The data was checked for missing data. If a student had a score for any of the three batteries, it was included in the data set. Missing data included any student that was opted out for testing or had invalid scores for all three batteries. Descriptive statistics and normality testing is presented prior to describing the findings to the research questions. Each research question is presented followed by test results. The results are described and interpreted for the context of this study. The research questions discussed are:
39


1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7?
2. What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups?
3. What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups?
4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?
Prior to disseminating results, it is important to remind the reader how gifted referral is obtained in District. A cognitive test, CogAT7, is used as a universal screener, given to all second-grade students in the district. The full battery version is used with Verbal, Quantitative, and Nonverbal components. A score of 90th percentile or above, using national age norms, refers a student for further academic testing. A screening at 90th percentile is intended to cast a wider net for gifted potential. Students still are required to meet the 95th percentile mark for Colorado state gifted identification. The results shared hereafter are from CogAT, used as an entry point to gifted identification.
Descriptive Statistics
In this section, descriptive statistics are presented, including valid and missing cases. Procedures for checking valid and missing data are described. Then, normality is tested using skewness and kurtosis. Central tendencies for battery and race are described to check for outlying data and participation by race. Finally, distributions across administration years and batteries are explored to check for reliability across testing years. In the end, a determination of validity for data usage is given for this study.
40


Valid and Missing Cases
Table 7 gives the overall data set statistics, including valid and missing cases for each of the batteries (Verbal, Quantitative, and Nonverbal). It also lists the percentage of valid and missing cases. The sample population from 2016, 2017, 2018 data began with 11,648 students taking the CogAT 7 test. The valid scores for each battery are listed along with cases missing. Table 7
Observations by Battery 2016, 2017, and 2018 Combined
Battery N Valid Percent N Cases Missins Percent Total
Verbal 10972 94.2 676 5.8 11648
Quantitative 10659 91.5 989 8.5 11648
Nonverbal 11312 97.1 336 2.9 11648
Cases missing include scores from not enough test questions answered for a percentile rank. Any valid battery score from the 11,648 students will be used for gifted referral determination. Only students with three invalid battery scores will be excluded from the data. Since all student samples have at least one valid score, all 11,648 student scores will be included in this study. When analyzing by battery, only valid test scores from that battery will be evaluated. The missing cases will be excluded from the data set.
Testing for Normalcy
Based on normal distribution having skewness of ± 1 and kurtosis of ± 3, all CogAT batteries (Verbal, Quantitative, and Nonverbal) are within the threshold of normal distribution. Skewness and kurtosis of each CogAT battery are given in Table 8. Since the data is within the threshold, the data from each testing year is valid for this study.
41


Table 8
Testing for Normalcy by Battery 2016, 2017, and 2018 Combined
Battery N Mean SD Skewness Kurtosis Normal Distribution
Verbal 10972 57.18 26.277 -.324 -.922 Yes
Quantitative 10659 67.13 23.444 -.749 -.273 Yes
Nonverbal 11312 63.31 24.628 -.524 -.679 Yes
Central Tendency by Race and Battery
Before reporting out regarding data from research questions, the researcher conducted descriptive statistics and t-tests to check for variance for each battery of the CogAT 7; including Verbal, Quantitative, and Nonverbal. In this section, central tendencies are reported by race and individual battery. The researcher is looking for any deviations or large discrepancies between number of participants, mean, and/or standard deviation; as possible variations could impact final findings and skew conclusions drawn. When computing percentile ranking means, only test scores (l-99th percentiles) were accepted as valid scores. Any score at zero was excluded from the data set, as this is not a valid score. Table 9 gives the number of valid scores, means, and standard deviations for each of the three batteries organized by race.
Consistently, quantitative mean results are higher for each demographic group. Trends between battery means are similar across racial groups. However, Asian and White have consistently higher means across batteries than their Black, Hispanic, and American Indian counterparts. This is consistent with findings from Ford (2011) and Miller (2004). Standard deviations are relatively consistent across racial groups, with the exception of a couple outliers, American Indian Verbal Battery and Asian Quantitative Battery. The American Indian Verbal Battery shows a greater variance of scores when compared to other racial groups. The Asian Quantitative Battery shows a smaller variance of scores when compared to other racial groups.
42


Table 9
Descriptive Statistics, All CogAT 7 Battery Scores by Race, including 2016, 2017, 2018 Data
Race CogAT Battery n M SD
American Indian Verbal 60 51.72 29.07
Quantitative 57 61.58 26.19
Nonverbal 62 58.85 25.28
Asian Verbal 942 64.12 24.22
Quantitative 946 76.81 20.67
Nonverbal 971 72.08 22.16
Black Verbal 1036 46.70 25.16
Quantitative 991 55.98 24.59
Nonverbal 1088 49.11 24.02
Hispanic Verbal 2114 46.57 25.16
Quantitative 2036 58.52 23.91
Nonverbal 2215 54.86 24.92
Multi-Racial Verbal 878 57.70 25.90
Quantitative 843 65.79 23.76
Nonverbal 899 62.32 24.42
White Verbal 5918 61.72 25.27
Quantitative 5762 70.78 21.73
Nonverbal 6053 67.77 23.05
Comparisons Across Administration Years Within Batteries
In order to verify the data set was reliable across administration years, comparisons were
made of all scores within each battery across administration years, checking for variance. First,
descriptive statistics, including number of observations, mean, and standard deviation, are
reported for years 2016, 2017, and 2018. After that, independent-sample t-tests are reported
comparing administration year results within each battery.
Verbal battery. Table 10 reports the Verbal battery across administration years (2016, 2017, 2018). Then, results for t-tests are given in Table 11. Independent-samples t-tests were
run to determine if there were differences in Verbal battery between administration years 2016 vs
43


2017 vs 2018. For administration years 2016 us 2017 an independent-samples t-test was run. There were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2016 vs 2017, as assessed by Levene’s test for equality of variances (p = .291). For the Verbal battery, there was no statistically significant difference for 2016 (M= 58.32, SD = 25.88) and 2017 (M= 57.74, SD = 26.28), where (M = .584, 95% Cl [-.597, 1.766], *(7488) = .969,p = .332, d= .02.
A Welch t-test was run to determine if there were differences in Verbal battery scores between administration years 2016 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s test for equality of variances (p = .034). There was a significant difference between 2016 (M= 58.32, SD = 25.88) and 2018 (M= 55.35, SD = 26.61), where (M = 2.97, 95% Cl [1.76, 4.18], *(7168) = 4.809, p = <001, g= .11.
An independent-samples t-test was run to determine if there were differences in Verbal battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot. There was homogeneity of variances, as assessed by Levene’s test for equality of variances (p = .282). There was a significant difference between 2017 (M= 57.74, SD = 26.28) and 2018 (M= 55.35, SD = 26.61), where (M= 2.385, 95% Cl [1.162, 3.607], *(7199) = 3.825,p = <001, d= .09.
44


Table 10
Comparison of CogAT Scores for Administration Years (2016, 2017, 2018), Verbal Battery
Administration Year n Mean SD
2016 3771 58.32 25.88
2017 3719 57.74 26.28
2018 3482 55.35 26.61
Table 11
Comparison of Verbal Battery CogAT Scores for Administration Years, T-Tests
Years Compared df P Significance
2016 vs 2017 7488 .332 No
2016 vs 2018 7251 <.001 Yes
2017 vs 2018 7199 <.001 Yes
As the t-tests report, there is a statistically significance in two out of three comparison years for the Verbal battery. These results show that this test could have different outcomes, depending on the years gathered. This should be noted for further studies, since findings will depend on years pulled and are conditional on the population of the district. When reviewing the means, the mean for 2018 was significantly lower than the two previous years (2016 and 2017), leading to a significant difference in comparison years.
Quantitative battery. Table 12 gives a reporting for the Quantitative battery across administration years (2016, 2017, 2018). Then, results for t-tests are given in Table 13. Independent-samples t-tests were run to determine if there were differences in Quantitative battery between administration years 2016 vs 2017 us 2018. For administration years 2016 us 2017, there were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2016 vs 2017, as assessed by
45


Levene’s test for equality of variances (p = .786). For the Quantitative battery, there was no statistically significant difference for 2016 (M= 66.95, SD = 23.43) and 2017 (M= 67.13, SD = 23.64), where (M= -.185, 95% Cl [-1.264, .894], t(7309) = -.336,p = .737, d= .008.
An independent-samples t-test was run to determine if there were differences in Quantitative battery scores between administration years 2016 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2016 vs 2018, as assessed by Levene’s test for equality of variances (p = .601). There was no significant difference between 2016 (M= 66.95, SD = 23.43) and 2018 (M= 67.32, SD = 23.247), where (M= -.373, 95% Cl [-1.465, .720], t(7028) = -.669,p= .601, d= .02.
An independent-samples t-test was run to determine if there were differences in Quantitative battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2017 vs 2018, as assessed by Levene’s test for equality of variances (p = .435). There was no significant difference between 2017 (M= 67.13, SD = 23.64) and 2018 (M= 67.32, SD = 23.247), where (M= -.188, 95% Cl [-1.29, .914], t(6975) = -334,p = .738, d= .008.
As reported in Table 13, no statistically significant difference was found when comparing administration years, lending to reliability between administration years. When reviewing the means for each administration year, they were close across years, ranging from 66.95-67.32. Results from this data are going to be consistent in this population between administration years.
46


Table 12
Comparison of CogAT Scores for Administration Years (2016, 2017, 2018), Quantitative Battery
Administration Year n Mean SD
2016 3682 66.95 23.43
2017 3629 67.13 23.64
2018 3348 67.32 23.25
Table 13
Comparison of Quantitative Battery CogAT Scores for Administration Years
Years Compared df P Significant
2016 vs 2017 7309 .737 No
2016 vs 2018 7028 .504 No
2017 vs 2018 6975 .738 No
Nonverbal battery. Table 14 gives a reporting for the Nonverbal battery across administration years (2016, 2017, 2018). Then, results for t-tests are given in Table 15. Independent-samples t-tests were run to determine if there were differences in Nonverbal battery between administration years 2016 vs 2017 us 2018. For administration years 2016 us 2017, there were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2016 vs 2017, as assessed by Levene’s test for equality of variances (p = .697). For the Nonverbal battery, there was a statistically significant difference for 2016 (M= 62.34, SD = 24.765) and 2017 (M= 63.8, SD =24.72), where (M=-1.46, 95% Cl [-2.568, -.352], t(7668) =-2.584,p = .01, d= .06.
An independent-samples t-test was run to determine if there were differences in Nonverbal battery scores between administration years 2016 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each
47


administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2016 vs 2018, as assessed by Levene’s test for equality of variances (p = .219). There was a significant difference between 2016 (M= 62.34, SD = 24.765) and 2018 (M= 63.86, SD = 24.361), where (M= -1.521, 95% Cl [-2.633, -.410], *(7514) = -2.683,p = .007, d= .06.
An independent-samples t-test was run to determine if there were differences in Nonverbal battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q-Q Plot, and there was homogeneity of variances between 2017 vs 2018, as assessed by Levene’s test for equality of variances (p = .405). There was no significant difference between 2017 (M= 63.8, SD =24.72) and 2018 (M= 63.86, SD = 24.361), where (M= -.061, 95% Cl [-1.177, 1.055], *(7436) = -.108,/) = .914, d= .002.
Table 14
Comparison of CogAT Scores for Administration Years (2016, 2017, 2018), Nonverbal Battery
Administration Year N Mean SD
2016 3874 62.34 24.76
2017 3796 63.80 24.72
2018 3642 63.86 24.36
Table 15
Comparison of Nonverbal Battery CogAT Scores for Administration Years
Years Compared df P Significance
2016 vs 2017 7668 .010 Yes
2016 vs 2018 7514 .007 Yes
2017 vs 2018 7436 .914 No
48


The data shows statistically significant differences in two out of three comparisons.
When reviewing the means between administration years, 2016 had a significantly lower mean score than 2017 and 2018, leading to the significant difference found in the t-tests. Again, as with the Verbal battery, results for this test are going to vary from year to year. This variance should be taken into account as others look to replicate this study.
Summary
The descriptive statistics show that Verbal and Nonverbal varied within this district between administration years. Quantitative stayed consistent and no significant difference was found between administration years. This study was not intended to generalize across districts or states, rather give a reporting of one school district. Since the observation sample for this study is large, we will move forward with the research questions. Due to the large difference of representative demographic groups, ranging from 60 to over 6,000, a weighting adjustment was factored prior to performing any of the research question testing. Although it is a large sample size, it is truly inclusive of all students within the district. Since the scope of this study does not extend beyond this district, the weighting adjustment is sufficient.
Research Question 1:
Does race affect test scores from the Cognitive Abilities Test, CogAT 7?
In addressing this question, the researcher separated each battery by racial group, as designated by PowerSchool. Each battery (Verbal, Quantitative, and Nonverbal) was tested separately. ANOVA testing was used, with a weighted adjustment, due to the population discrepancy between racial groups. The following sections explain the results found for each battery.
49


Verbal Battery
A one-way Welch ANOVA was conducted to determine if CogAT Verbal battery scores were different between racial groups. Participants were classified into demographic groups, including: American Indian (n = 60), Asian (n = 942), Black (n = 1036), Hispanic (n = 2114), Multi-Racial (n = 878), and White (n = 5918). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Verbal test scores were statistically significantly different between racial groups, Welch’s F(5, 589.49) = 160.69,p < .001. Means and standard deviations are provided in Table 16. The group means for Verbal battery were statistically significantly different (p<001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does affect Verbal battery test scores.
Table 16
Population Descriptive, Verbal Battery, Used for ANOVA Testing
Race N Mean SD
American Indian 60 51.72 29.07
Asian 942 64.12 24.22
Black 1036 46.70 25.16
Hispanic 2114 46.57 25.81
Multi 878 57.70 25.89
White 5918 61.72 25.28
Games-Howell post hoc analysis (Table 17) revealed that statistically significant differences existed between each racial comparison, except between Black*Hispanic and White* Asian. No significant difference between American Indian and other racial groups, excluding Asian, was found.
50


Table 17
Games-Howell Post Hoc Comparing Verbal CogAT Scores Across Race
Racial Comparison p-value Significant Different
American Indian*Asian .023 * Yes
American Indian*Black .780 No
American Indian*Hispanic .752 No
American Indian*Multi .631 No
American Indian*White .100 No
Asian*Black < .001** Yes
Asian*Hispanic < .001** Yes
Asian*Multi < .001** Yes
Asian*White .056 No
Black*Hispanic 1.000 No
Black*Multi < .001** Yes
Black*White < .001** Yes
Hispanic*Multi < .001** Yes
Hispanic*White < .001** Yes
Multi* White < .001** Yes
*: p-value <0.05 **: p-value <0.01
Quantitative Battery
A one-way Welch ANOVA was conducted to determine if CogAT Quantitative battery scores were different between racial groups. Participants were classified into demographic groups, including: American Indian (n = 57), Asian (n = 946), Black (n = 991), Hispanic (n = 2036), Multi-Racial (n = 843), and White (n = 5762). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Quantitative test scores were statistically significantly different between racial groups, Welch’s F{5, 560.27) = 165.887,/) < .001. Means and standard deviations are provided in Table 18.
51


Table 18
Population Descriptive, Quantitative Battery, Usedfor ANOVA Testing
Race N Mean SD
American Indian 57 61.58 26.19
Asian 946 76.81 20.67
Black 991 55.98 24.59
Hispanic 2036 58.52 23.91
Multi 843 65.79 23.76
White 5762 70.78 21.73
The group means for Quantitative battery were statistically significantly different (p<001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does affect Quantitative battery test scores. Table 19 lists the post hoc testing results.
Table 19
Games-HowellPost Hoc Comparing Quantitative CogATScores across Race
Racial Comparison P -value Significant Difference
American Indian*Asian .001** Yes
American Indian*Black .619 No
American Indian*Hispanic .952 No
American Indian*Multi .844 No
American Indian*White .103 No
Asian*Black < .001** Yes
Asian*Hispanic < .001** Yes
Asian*Multi < .001** Yes
Asian*White < .001** Yes
Black*Hispanic .079 No
Black*Multi < .001** Yes
Black*White < .001** Yes
Hispanic*Multi < .001** Yes
Hispanic*White < .001** Yes
Multi* White < .001** Yes
*: p-value <0.05 **: p-value <0.01
52


Games-Howell post hoc analysis (Table 19) revealed that statistically significant differences existed between all racial group comparisons, except between Black*Hispanic. Also, statistically significant difference was not found when comparing American Indian to other racial groups, excluding Asian.
Nonverbal Battery
A one-way Welch ANOVA was conducted to determine if CogAT Nonverbal battery scores were different between racial groups. Participants were classified into demographic groups, including: American Indian (n = 57), Asian (n = 946), Black (n = 991), Hispanic (n = 2036), Multi-Racial (n = 843), and White (n = 5762). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Nonverbal test scores were statistically significantly different between racial groups, Welch’sF(5, 609.784) = 201.603,p< .001. Means and standard deviations are provided in Table 20. The group means for Nonverbal battery were statistically significantly different (p<001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does affect Nonverbal battery test scores.
Table 20
Population Descriptive, Nonverbal Battery, Used for ANOVA Testing
Race N Mean SD
American Indian 57 58.85 25.277
Asian 946 72.08 22.164
Black 991 49.11 24.024
Hispanic 2036 54.86 24.925
Multi 843 62.32 24.422
White 5762 67.77 23.048
53


Games-Howell post hoc analysis (Table 21) revealed that statistically significant differences existed between all racial group comparisons; however, statistically significant difference was not found when comparing American Indian to other racial groups, excluding Asian and Black.
Table 21
Games-Howell Post Hoc Comparing Nonverbal CogAT Scores across Race
Racial Comparison P- â– value Significant Difference
American Indian*Asian .002* Yes
American Indian*Black .047* Yes
American Indian*Hispanic .821 No
American Indian*Multi .900 No
American Indian*White .077 No
Asian*Black < .001** Yes
Asian*Hispanic < .001** Yes
Asian*Multi < .001** Yes
Asian*White < .001** Yes
Black*Hispanic < .001** Yes
Black*Multi < .001** Yes
Black*White < .001** Yes
Hispanic*Multi < .001** Yes
Hispanic*White < .001** Yes
Multi* White < .001** Yes
*: p-value <0.05 **: p-value <0.01
Answering Research Question 1:
Does race affect test scores from the Cognitive Abilities Test, CogAT 7
In answering Research Question 1, data indicates that yes, race does affect test scores. Overall, each ANOVA test comparing race to an individual battery (Verbal, Quantitative, and Nonverbal) resulted in a p-value lower than the alpha value of 0.05; thus, rejecting the null hypothesis. In each case, there was a significant difference in test results between race. This led to post hoc testing, using Games-Howell to test individual comparisons between racial groups to
54


identify significant comparison differences. In these post hoc tests, consistently there was a difference between racial groups, with the exclusion of American Indian compared with Asian and once with Black. It should be noted that American Indian showed no statistically significant difference, except with Asian, across batteries. Also, American Indian population size was between 57-62 observations. This was an outlier when compared to other racial populations, ranging from 843-6053. Therefore, these results should be taken with caution, as the number of cases drastically differed.
Research Question 2: What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups?
It is clear from the literature that scores for Black and Hispanic groups are consistently lower (Ford, 2013; Plucker & Peters, 2016). What the researcher is trying to determine is do the scores significantly affect the referral rates between racial groups, thus making the referral rates inequitable. In this section, each battery is separated and compared using ANOVA testing, across racial groups. Table 22 gives the frequency of 95th percentile referral by demographic. Also, the percentage referred compared to racial population is reported. For gifted referrals to be equitable, the percentage of referrals should closely align with the demographic make-up for the school district. Table 23 gives the overall referral rate for each demographic as a percentage of all gifted referrals across the district. The data was compared to the overall gifted referral and subsequently compared to overall demographic make-up. Once again, these percentages should closely align if referrals are equitable across demographic groups. These frequencies were then used in the ANOVA testing to check for equitability across racial group.
55


Table 22
Relative Frequency of Gifted Referral (95th Percentile) by Race for each CogAT Battery
Race CogAT Battery N Frequency Percentage of racial population
American Indian Verbal 62 2 3.1
Quantitative 55 7 10.8
Nonverbal 59 4 6.3
Asian Verbal 961 72 7.5
Quantitative 964 184 19.1
Nonverbal 978 145 14.8
Black Verbal 1086 15 1.4
Quantitative 1055 23 2.2
Nonverbal 1118 13 1.2
Hispanic Verbal 2192 56 2.6
Quantitative 2139 70 3.3
Nonverbal 2250 74 3.3
Multi-Racial Verbal 917 51 5.6
Quantitative 894 65 7.3
Nonverbal 920 68 7.4
White Verbal 6046 403 6.7
Quantitative 5973 583 9.8
Nonverbal 6117 567 9.3
Table 23
Relative Frequency of Gifted Referral (95th Percentile) by Race
Race N Frequency % of Racial Population % of Gifted Referrals from 95th Percentile % of Whole Student Population
American Indian 60 10 16.7 0.6 0.6
Asian 943 261 26.5 15.9 8.7
Black 1017 42 3.7 2.6 9.4
Hispanic 2069 147 6.4 8.9 19.1
Multi-Racial 873 125 14.3 7.6 8.1
White 5850 1059 18.1 64.4 54.1
56


A one-way Welch ANOVA was conducted to determine if gifted referral at the 95th percentile was significantly different across racial groups. Participants were classified into demographic groups, including: American Indian (n = 60), Asian (n = 943), Black (n = 1017), Hispanic (n = 2069), Multi-Racial (n = 873), and White (n = 5850). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Table 24 relates population descriptives used for the ANOVA testing. 95th percentile gifted referrals were statistically significantly different between racial groups, Welch’s F(5, 592.85) = 99,73, p < .001; therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does create a significant difference in 95th percentile gifted referral. Games-Howell post hoc analysis (Table 25) revealed that statistically significant differences existed between all racial group comparisons; however, statistically significant difference was not found when comparing American Indian to other racial groups.
Table 24
Population Descriptors, 95th Percentile Gifted Referrals, Used for ANOVA Testing
Race n Mean SD
American Indian 60 .17 .376
Asian 943 .28 .448
Black 1017 .04 .199
Hispanic 2069 .07 .257
Multi 873 .14 .350
White 5850 .18 .385
57


Table 25
Games-Howell Post Hoc Comparing 95th Percentile Gifted Referral across Race
Racial Comparison p-value Significant Difference
American Indian*Asian .263 No
American Indian*Black .122 No
American Indian*Hispanic .378 No
American Indian*Multi .997 No
American Indian*White 1.000 No
Asian*Black < .001** Yes
Asian*Hispanic < .001** Yes
Asian*Multi < .001** Yes
Asian*White < .001** Yes
Black*Hispanic .006 * Yes
Black*Multi < .001** Yes
Black*White < .001** Yes
Hispanic*Multi < .001** Yes
Hispanic*White < .001** Yes
Multi* White < .040 * Yes
*: p-value <0.05 **: p-value <0.01
All demographic group comparisons, excluding American Indian, revealed significant differences. Further testing is necessary to determine the degree to which this difference occurs. However, what can be said is that race does affect gifted referral using the CogAT7 at the 95th percentile.
Research Question 3: What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups?
Part of the district effort to increase diversity within gifted education included referring students at the 90th percentile and above for further testing. This research question tested the significance of difference between racial groups at the 90th percentile, to see if changing from 95th percentile to 90th percentile reduced the referral disparity between racial groups.
58


Table 26 gives the frequency of 90th percentile referral by demographic. It also lists the percentage referred when compared to the racial population. The ANOVA testing indicated this difference was significant.
Table 26
Relative Frequency of Gifted Referral (90th Percentile) by Race for each CogAT Battery
Race CogAT Battery N Frequency Percentage of racial population
American Indian Verbal 64 7 10.9
Quantitative 62 8 12.9
Nonverbal 63 6 9.5
Asian Verbal 961 132 13.7
Quantitative 964 294 30.5
Nonverbal 978 249 25.5
Black Verbal 1086 34 3.1
Quantitative 1055 59 5.6
Nonverbal 1118 43 3.8
Hispanic Verbal 2192 99 4.5
Quantitative 2139 145 6.8
Nonverbal 2250 148 6.6
Multi-Racial Verbal 917 95 10.4
Quantitative 894 137 15.3
Nonverbal 920 116 12.6
White Verbal 6046 797 13.2
Quantitative 5973 1161 19.4
Nonverbal 6117 1083 17.7
Table 27 gives the overall referral for each demographic and compared it to the overall gifted referral and then to overall demographic make-up. In order for the data to be equitable across racial groups, the referral percentage and the demographic make-up should closely resemble each other.
59


Table 27
Relative Frequency of 90th Percentile Gifted Referral by Race
Race N Frequency % of Racial Population Referred % of Total Gifted Referrals % of Whole Student Population
American Indian 60 14 23.3 .5 .6
Asian 943 385 40.8 13.5 8.7
Black 1017 97 9.5 3.4 9.4
Hispanic 2069 253 12.2 8.9 19.1
Multi-Racial 873 212 24.3 7.5 8.1
White 5850 1884 32.2 66.2 54.1
A one-way Welch ANOVA was conducted to determine if gifted referral at the 90th percentile was significantly different across racial groups. Participants were classified into demographic groups, including: American Indian (n = 60), Asian (n = 943), Black (n = 1016), Hispanic (n = 2069), Multi-Racial (n = 873), and White (n = 5850). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q-Q Plot. Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Table 28 relates population descriptives used for the ANOVA testing. 90th percentile gifted referrals were statistically significantly different between racial groups, Welch’s F{5, 592.57) = 155.523,p< .001; therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does create a significant difference in 90th percentile gifted referral.
Games-Howell post hoc analysis (Table 29) revealed that statistically significant differences existed between all racial group comparisons, except with the following comparisons American Indian*Black, American Indian*Hispanic, American Indian*Multi, American Indian*White, and Black*Hispanic.
60


Table 28
Population Descriptors, 90th Percentile Gifted Referrals, Used for ANOVA Testing
Race N Mean SD
American Indian 60 .23 All
Asian 943 .41 .492
Black 1017 .10 .295
Hispanic 2069 .12 .328
Multi 873 .24 .429
White 5850 .32 .467
Table 29
Games-Howell Post Hoc Comparing 90th Percentile Gifted Referral across Race
Racial Comparison P- â– value Significant Difference
American Indian*Asian .037* Yes
American Indian*Black .154 No
American Indian*Hispanic .354 No
American Indian*Multi 1 .000 No
American Indian*White .601 No
Asian*Black < .001** Yes
Asian*Hispanic < .001** Yes
Asian*Multi < .001** Yes
Asian*White < .001** Yes
Black*Hispanic .234 No
Black*Multi < .001** Yes
Black*White < .001** Yes
Hispanic*Multi < .001** Yes
Hispanic*White < .001** Yes
Multi* White < .001** Yes
*: p-value <0.05 **: p-value <0.01
61


Research Question 4:
What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?
Using a stratified random sampling, a MANOVA was conducted to determine if gifted referral at the 90th and 95th percentiles were significantly different across racial groups. The stratified random sampling was applied, to provide equal representation for each racial group. Since the American Indian demographic had a significantly lower number of observations and found that it was not significantly different with other demographics, excluding Asian, this racial group was excluded from this question. Participants were classified into demographic groups, including: Asian (n = 500), Black (n = 500), Hispanic (n = 500), Multi-Racial (n = 500), and White (n = 500). Homogeneity of variances was violated, as assessed by Levene’s Test of Homogeneity of Variance (p = <001). Means and standard deviations are provided in Table 30. There was a statistically significant difference between racial groups for 90th percentile and 95th percentile referrals, F(8, 4990) = 26.658,p = <.001; Pillai’s Trace = .082; partial rf = .041; thus rejecting the null hypothesis and accepting the alternative hypothesis. There is a significant difference between 90th and 95th percentile referrals between race, using the CogAT 7.
Table 30
Number, Mean, and Standard Deviation of Stratified Random Sampling, One-way MANOVA
Race n 90th Mean Percentile SD 95th Mean Percentile SD
Asian 500 .40 .490 .28 .451
Black 500 .10 .300 .03 .176
Hispanic 500 .13 .334 .08 .268
Multi 500 .23 .423 .13 .339
White 500 .32 .468 .18 .385
62


In order to determine rates of referral and determine where the significance was situated in the One-way MANOVA, follow-up Chi-square goodness-of-fit testing was conducted. The results of the follow-up testing are reported in the next sections.
95th Percentile Referral by Race, Chi-Square Goodness-of-Fit
In this section, the data is organized by racial group and referrals at the 95th percentile. The complete population set was used in this analysis, since the researcher desired to evaluate total number of referrals for each demographic group. Relative referrals expected by population and observed referrals are given. Chi-square goodness-of-fit tests are applied to determine if each racial group had the expected five percent of its population represented. The results of these Chi-square tests are in Table 31 below.
Table 31
Gifted Referral by Race at the 95th Percentile, Chi-Square Goodness-of-Fit
Race Relative Expected by Population Observed p-value Significant Difference
American Indian 3 10 < .001** Yes +2.35
Asian 47 261 < .001** Yes +4.55
Black 57 43 .051 No -0.25
Hispanic 115 148 .002 * Yes +0.29
Multi-Racial 47 126 < .001** Yes + 1.68
White 309 1072 < .001** Yes +2.47
*: p-value <0.05 **: p-value <0.01
Individual Chi-square goodness-of-fit tests were conducted for each racial group to determine whether the participants in this study had similar proportions at the 90th percentile. The expected frequency for each demographic was five percent of the total population, as this was testing for the 95th percentile and above. The Chi-square goodness-of-fit test indicated that
63


the following demographics had significantly more referrals than expected: American Indian (X2(l) = 14.757, p = <001); Asian (x2(l) = 1020.971, p = <001); Hispanic (x2(l) = 10.003, p = .002); Multi-Racial (x2(l) = 140.429, p = <001); and White (x2(l) = 1978.562, p = <001). CogAT7 referred triple the number of American Indian than expected, five times the number of Asian, slightly more Hispanic, a little more than double the expected number of Multi-Racial and more than triple the number of White. When testing for Black, the chi-square goodness-of-fit test indicated that the CogAT7 test referred within the limits of five percent of the population: Black (X2(l) = 3.803, p = .051). Even though the test is pulling at least five percent of each racial population, the proportions are skewed severely when discussing the percentage of overall gifted referral disaggregated by race. For example, although CogAT7 is screening around five percent of Black population, it is simultaneously pulling significantly more percentages of American Indian, Asian, Hispanic, Multi-Racial, and White. Hispanic is only slightly higher at slightly more than expected; whereas, for Asian and White populations expected referrals were significantly higher at five times for Asian and triple for White. This large discrepancy leads the researcher to examine if screening at the 90th percentile alleviates or exacerbates this discrepancy.
90th Percentile Referral by Race, Chi-Square Goodness-of-Fit
In this section, the data is disaggregated by race and referrals at the 90th percentile are used. The complete population set was used in this analysis, since the researcher wanted to evaluate total number of referrals for each demographic group. Relative referrals expected by population and observed referrals are given and Chi-square goodness-of-fit tests are applied to determine if each racial group had the expected ten percent of the population represented. The results of these Chi-square tests are in Table 32 below.
64


Table 32
Gifted Referral by Race at the 90th Percentile, Chi-Square Goodness-of-Fit
Race Expected Observed p-value Significant Difference
American Indian 7 14 .002* Yes +1.00
Asian 94 385 < .001** Yes +3.10
Black 115 101 .175 No -0.12
Hispanic 230 259 .043* Yes +0.13
Multi-Racial 94 216 < .001** Yes +1.30
White 618 1910 < .001** Yes +2.09
*: p-value <0.05 **: p-value <0.01
Individual Chi-square goodness-of-fit tests were conducted for each racial group to determine whether the racial populations in this study had similar proportions at the 90th percentile. The expected frequency for each demographic was ten percent of the total population, as this was testing for the 90th percentile and above. The Chi-square goodness-of-fit tests indicated that the following demographics had significantly more referrals than expected: American Indian (x2(l) = 9.615, p = .002); Asian (x2(l) = 995.717, p = <.001); Hispanic (x2(l) = 4.093, p = .043); Multi-Racial (x2(l) = 176.887, p = <001); and White (x2(l) = 2993.603, p = <001). CogAT7 referred twice the number of American Indian than expected, quadruple the number of Asian, slightly more Hispanic, double the expected number of Multi-Racial and triple the number of White. When testing Black, the Chi-square goodness-of-fit test indicated that the CogAT7 test referred within the limits of ten percent of the population: Black (x2(l) = 1.843, p = .175). Even though the test is pulling at least ten percent of each racial population, the proportions are skewed severely when discussing the percentage of overall gifted referral disaggregated by race. For example, although CogAT7 is screening around ten percent of Black population, it is simultaneously pulling significantly more percentages of American Indian,
65


Asian, Hispanic, Multi-Racial, and White. Hispanic is only slightly higher at slightly more than expected; whereas, for Asian and White populations expected referrals were significantly higher at almost quadruple times for Asian and over triple for White. This large discrepancy led the researcher to examine the overall gifted referral rate disaggregated by race.
Overall Gifted Referral Rate Disaggregated by Race
In the previous sections, goodness-of-fit was reported for the 95th and 90th percentiles. It was found that observed gifted referrals and expected referrals for student groups, excluding Black, were significantly different. The researcher then wanted to determine if the proportions of overall gifted referral were similar to the overall student population. The goal of gifted education is to refer an equitable number of students from each demographic, based on the overall student racial population. So, for example, if fifty percent of the given population is White, there should be around fifty percent of the gifted referrals coming from the White demographic. It was determined that Chi-square goodness-of-fit tests would be conducted, as the overall percentages of gifted referral would be compared to the overall student population.
Chi-square goodness-of-fit tests were conducted to determine whether the students referred for further gifted testing were typical of the overall student population. The expected frequencies came from the overall student population (see Table 33). The Chi-square goodness-of-fit test for the 95th percentile indicated that the percentage of gifted referrals was statistically significantly different from the proportions found in the overall student population (%2(4) = 17.527, p = .002). The Chi-square goodness-of-fit test for the 90th percentile indicated that the percentage of gifted referrals was statistically significantly different from the proportions found in the overall student population (%2(4) = 15.449, p = .004). These results indicated that, even with moving to the 90th percentile for gifted referral, there continued to be a statistically
66


significant inequitable distribution of gifted referral across demographics, highly favoring White and Asian populations.
Table 33
Percent of Gifted Referral by Race at the 90th and 95th Percentiles
90th Percentile Total Gifted Referrals 95th Percentile Total Gifted Referrals % Overall Student
Race Percent n Percent n Population
American Indian 0.5 14 0.6 10 0.6
Asian 13.5 385 15.9 261 8.7
Black 3.4 101 2.6 43 9.4
Hispanic 8.9 259 8.9 148 19.1
Multi-Racial 7.5 216 7.6 126 8.1
White 66.2 1910 64.4 1072 54.1
It should be noted that for Black, this change to 90th percentile did significantly increase the number of Black students being referred for gifted evaluation. However, this also came at the cost of highly over-referring both White and Asian populations. Referring at the 90th percentile was not advantageous for Hispanic, since it did not significantly increase the number of Hispanic gifted referrals in comparison to the overall gifted referral, at 8.9% of the total gifted referrals at both the 90th and 95th percentiles. Alternative methods of screening and referrals should be sought to more equitably refer students from varying demographics. Options for these methods are discussed in the next chapter, as well as recommendations for future studies.
Summary
The results of this study indicate (a) CogAT 7 test scores are affected by race, (b) 95th percentile gifted referrals were significantly different across race, (c) 90th percentile gifted referrals were significantly different across race, (d) when comparing 90th and 95th percentile referrals, there was a significant difference, and (e) gifted referrals at the 90th and 95th percentiles were significantly not equitable across race. Although screening for the 90th percentile does
67


increase the number of referrals per demographic, there is still a vast over-referral of Asian and White and under-referral of Black and Hispanic populations. So, although one can argue that moving to the 90th percentile statistically increases the number of referrals for each demographic, it does not dramatically of gifted referrals. Therefore, alternate referral methods and follow-up studies are recommended.
68


CHAPTER V
DISCUSSION
In this chapter, the study is summarized along with findings of each research question. Conclusions are drawn and related back to the current body of literature introduced in Chapter 2. Implications and future research is outlined as potential future studies, namely: (a) development of equity goals, (b) second screening process, and (c) talent enrichment programs and/or strategies for CLD learners. Finally, a summary and final thoughts are drawn as the researcher concludes this study.
Summary of the Study
Controversy abounds in education around racial topics, including achievement gap, opportunity gap, educational debt, and the excellence gap. These gaps, or significant differences in achievement, resources, and/or opportunity, create inequitable access to gifted education. This study was developed to determine whether current practices in screening students at the 90th percentile, compared to the traditional 95th percentile, would help to mitigate the racial discrepancies and inequities currently prevailing in gifted screening practices. This referral at the 90th percentile was not meant to re-establish the 95th percentile identification practices and/or requirements, but for referral purposes. Further testing would be conducted after the referral and current CDE practices and policies would be used for gifted identification criteria.
Data were collected from a Colorado school district, where equity work is paramount.
The district assessment department provided the researcher with three years’ worth of CogAT 7 data, used as a universal screener for gifted referral. The full nine-subtest assessment were administered to every second grader in the district. Any student scoring 90th percentile or above
69


on any of the three batteries were referred for additional achievement testing, which is used to create a body of evidence for gifted identification.
The following research questions framed this study:
1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7?
2. What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups?
3. What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups?
4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile?
The results of this study indicated (a) CogAT 7 test scores are affected by race, (b) 95th percentile gifted referrals were significantly different across race, (c) 90th percentile gifted referrals were significantly different across race, (d) when comparing 90th and 95th percentile referrals, there was a significant difference, and (e) gifted referrals at the 90th and 95th percentiles were significantly not equitable across race. Therefore, alternate referral methods and follow-up studies are recommended.
Findings
Each of the results identified in the last section are discussed here. They are organized by (a) race and test score, (b) 95th percentile referrals, (c) 90th percentile referrals, and (d) comparison between 90th and 95th percentile referrals.
Race and Test Scores
This study found a statistically significant difference in test scores between racial groups. When answering Research Question 1, “Does race affect test scores from the Cognitive Abilities
70


Test, CogAT 7,” this study indicated that race does affect CogAT 7 test scores, with significant underrepresentation of Black and Hispanic populations when compared with White and Asian counterparts. This was consistent with findings regarding the achievement and excellence gaps across the nation (Ford, 2013; Miller, 2004; Olszewski-Kubilius & Thomson, 2010; Plucker & Peters, 2016). In the literature, Lohman (2005), creator of CogAT 7, states there are two types of giftedness: high-accomplishment (achievement oriented) and high-potential (aptitude oriented). He also states, “predictors of achievement in reading, mathematics, social studies, and science are the same for White, Black, Hispanic, and Asian American students” (p. 344). Although the researcher agrees with Lohman that there are two ways to look at gifted, achievement and aptitude, the researcher disagrees that this aptitude can be found in one test, as evidenced by the discrepancy in gifted referral by race from the CogAT 7. Also, from this study the researcher found that the predictors of achievement are significantly different across racial groups, in contrast with Lohman (2005) that predictors are the same across demographic groups. Lohman (2005) states that current predictors are based on past performance, to which the researcher agrees. However, it is generalized that racial groups have similar performance, to which this study demonstrates that there is statistically significant difference when comparing racial groups. The findings of this study are similar to numerous studies acknowledging the underrepresentation of gifted minoritized groups, Black and Hispanic (Feldman, 2003; McBee, 2006; Naglieri & Ford, 2015; Plucker, Burroughs, & Song, 2010; Worrell, 2014).
95th Percentile Referrals
When answering Research Question 2, “What is the difference in gifted referral, based on CogAT 7, at the 95th percentile and above, between racial groups,” the study shows that even with the adjustments to CogAT 7 (Lohman, 2011), it is not equitably referring all racial groups at
71


the 95th percentile. Although the top 5th percentile of all racial groups were referred, thus resulting in referring at least five percent of each population, the inequitable distribution of referral is significant. For example, Black students are referred at the threshold of five percent, while Asian students were referred at three times that rate. Likewise, White students were referred at twice the rate of their Black counterparts. Thus, referrals at the 95th percentile are inequitable. Lohman, himself states that using CogAT 7, “does not eliminate group differences” (Lohman, n.d.). This study shows that not only does it not eliminate group differences, these differences are statistically significant.
Another approach is using a nonverbal screening process, such as with the Naglieri Nonverbal Ability Test (NNAT). Naglieri, creator of NNAT, and Ford (2015) discuss the need for a nonverbal testing screener that more equitably refers students from Black and Hispanic populations. Giessman, Gambrell, and Stebbins (2013) conducted a study to identify the difference between referrals using the CogAT and the NNAT. They found that the Nonverbal CogAT 6 score identified as many minoritized students as the NNAT 2. However, one of the limitations to this study was that the sample groups were different; one including second graders, the other including kindergarten through second grade. Another important limitation is that the same students did not take both tests. Therefore, comparing the two samples should be taken with caution, as the difference or lack of difference could be due to the samples, not necessarily the tests. A follow-up study screening with a nonverbal test compared with the CogAT 7, where students take both tests, would aid in determining if the nonverbal test more equitably refers Black and Hispanic students.
72


90th Percentile Referrals
Answering Research Question 3, “What is the difference in gifted referral, based on CogAT 7, at the 90th percentile and above, between racial groups,” it was found that moving to the 90th percentile did little to influence the overall proportionality for gifted referral across racial groups. Table 34 shows the number of students referred at the 90th percentile compared to the 95th percentile.
Table 34
Relative Frequency of 90th and 95th Percentile Gifted Referral by Race
90th Percentile Total Gifted Referrals 95th Percentile Total Gifted Referrals % Overall Student
Race Percent N Percent N Population
American Indian 0.5 14 0.6 10 0.6
Asian 13.5 385 15.9 261 8.7
Black 3.4 101 2.6 43 9.4
Hispanic 8.9 259 8.9 148 19.1
Multi-Racial 7.5 216 7.6 126 8.1
White 66.2 1910 64.4 1072 54.1
Although frequencies change for Black students (from 43 to 101), over doubling the number of Black students referred, when compared to White students (from 1072 to 1910) and Asian students (from 261 to 385), there is some change to the proportionality of overall gifted referral; however, this is not significant enough to support 90th percentile referrals, as shown in Table 35. It increased Black representation from 2.7% overall representation to 3.4% at the 90th percentile referral. However, it also increases White representation from 64.4% to 66.2%. The effects of this change are discussed in the next research question. For Hispanic students, there was an increase in frequency from 148 to 259 gifted referrals; however, when compared to White and Asian populations, this does nothing to increase the overall gifted referral proportionality when compared to other racial groups, remaining at 8.9% in both 95th and 90th percentile referrals. As
73


the Hispanic student population is the fastest growing minority group in the state of Colorado (CDE, 2018), CLD gifted referral needs to be a high priority. The results of this study conclude that moving to the 90th percentile referral alone does little to reduce the overall disproportionality of racial discrepancies in gifted referral.
Table 35
Percent of Gifted Referral by Race at the 90th and 95th Percentiles
Relative Percent of Gifted Referrals % of Overall
90th Percentile 95th Percentile Student Population
American Indian 0.5 0.6 0.6
Asian 13.5 15.9 8.7
Black 3.4 2.6 9.4
Hispanic 8.9 8.9 19.1
Multi-Racial 7.5 7.6 8.1
White 66.2 64.4 54.1
Comparison Between 90th and 95th Percentile Referrals
Answering Research Question 4, “What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90th percentile to the 95th percentile,” the results show that there was mixed review of benefit and that the results have to be considered in the context of the whole findings across demographic groups. For instance, it was noted in the results section that for Black students, the change to 90th percentile referral did significantly increase the number of Black students, from 43 to 101 (see Table 34), being referred for gifted evaluation. However, this also came at the high cost of a significant increase of gifted referrals in both White and Asian populations (see Table 34). Referring at the 90th percentile was advantageous for Hispanic students, as it the frequency of referral increased from 148 to 259; however, it did not significantly increase the proportionality of Hispanic referrals in comparison to overall gifted referrals, at 8.9% of the total referrals at both the 90th and 95th percentiles.
74


Conclusions
The conclusions for this study are not uncommon in the field of gifted education, namely: (a) race affects test scores, (b) the need for inclusive referral practices is evident, and (c) underrepresentation of ethnically diverse gifted students exists and is prevalent. This section addresses each of these conclusions and relate them back to literature introduced in Chapter 2: Literature Review.
Race as a significant contributor to gifted referral adds to the body of research on racial inequities in gifted education (Cornell, Delcourt, Goldberg, & Bland, 1995; Erwin & Worrell, 2012; Grantham & Ford, 2003). What is unclear is the cause. There are possibilities, including: (a) testing bias, (b) using aptitude vs current ability, (c) poverty as a factor, and (d) language as a factor. Testing biases in intelligence tests has been scrutinized for the lack of norming across different minority groups. Black and Latino children consistently fall, on average, 10-15 points lower than middle-class White children (Ford, 2013b). This is not because the Black and Latino students are any less smart, but because tests were normed across a majority of White children (Ford, 2013b). Joseph and Ford (2006) state, “Nondiscriminatory assessment is concerned with fairness in all aspects of evaluating individuals” (p. 44). Some suggestions for equitable processes include: (a) least biased instruments, (b) avoiding confirmatory bias, and (c) ensuring policies and procedures are fair or nondiscriminatory (Joseph & Ford, 2006).
To become inclusive in gifted referral and identification practices, it is important to remember that enacting policies that are completely unbiased and equitable is unrealistic (Joseph & Ford, 2006; Ortiz, 2007). Ford (2012) states, “no field of education is free, or will be free, of dilemmas of difference” (p. 392). However, it is clear that research is needed regarding disproportionalities of gifted referral and equitable practices. Through research on least biased
75


instruments, using aptitude vs current ability, dispelling myths and biases around CLD gifted learners, considering poverty, factoring in language development, and changing processes to be more inclusive, state and district units can work together to close excellence gaps.
Table 36 reports Colorado student population trends disaggregated by race. Hispanic student population is the fastest growing demographic in Colorado, with well over double the next demographic group, Multi-Racial. Yet, it is the most underserved populations in Colorado gifted education, where American Indian (3.6%), Asian (9.7%), Black (2.7%), Hispanic (3.4%), Multi-Racial (7.1%), and White (8.4%) are the percentage of public students enrolled in a gifted program by race (National Center for Educational Statistics, 2016). For proportionality across race, there should be approximately 5% of any given population in gifted programming. This inequitable trend is not isolated in Colorado; it is a nation-wide dilemma.
Table 36
Colorado State Student Population Trends 2010-2017 (CDE, 2018)
Race 2010-2011 School Year 2016-2017 Difference
American Indian 7452 6511 -941
Asian 24493 28309 3816
Black 40537 41478 941
Hispanic 266098 303573 37475
Multi-Racial 23565 36388 12823
White 479327 486537 7210
The literature reviewed and this study demonstrate that underrepresentation of ethnically diverse gifted students needs to be a matter of great consideration in gifted education. Inequity of gifted referral processes will not be resolved quickly; it is woven into current systemic practices (Ford, 2012; Sullivan, 2011). Whatever decisions are for equitable referral and
76


identification for gifted education, researchers must begin to look differently; alternative entry points for inclusion of minoritized groups must be considered. Without these changes, systems will continue to over identify White and Asian children and under identify Black and Hispanic children.
Implications and Future Research
In the next sections, recommendations are addressed and introduced as follow-up studies to better reduce the inequity of gifted underrepresentation, namely: (a) development of equity goals, (b) second screening process, and (c) talent enrichment programs and/or strategies for CLD learners. Equity goals will focus attention of gifted education to equity-driven practice across referrals, curricula, and programming. Changing screening practices to be more inclusive will decrease the underrepresentation of minoritized groups. The development of talent enrichment programs for high-ability CLD students, utilizing a combination of CLD and gifted strategies, will aid in identifying and retaining CLD gifted learners in programs. All of these equity-driven strategies and further studies will aid in mitigating the disproportionality of minoritized populations in gifted education.
Equity Goals
One equity-driven method in gifted education is to introduce equity goals (Ford,
2013). Ford proposes a formula for equitable identification practices for underrepresented populations, namely American Indian, Hispanic, and Black students; applying the Office of Civil Rights (OCR) Civil Rights Data Collection (CRDC), to the current district gifted identification, to determine equitability of demographic distributions. The formula is as follows: Percentage of Black/Hispanic students in district (P) x 20% = B; Equity goal (E) = P-B. Using this equity formula, current District data distribution is applied in Table 37. The equity goal encourages the
77


district equity goal for gifted identification should move from the current 17% of American Indian, Hispanic, and Black populations to 22.6%, to more equitably represent the student population. Multi-racial groups are equitably represented in current gifted identification and programming. White and Asian populations continue to be overrepresented. These equity goals can be accomplished through equity-driven screening practices, focused on identifying our Black and Hispanic students, and inclusiveness in gifted enrichment programming.
Table 37
District Enrollment Compared to Gifted Identification, Disaggregated by Race
Ethnic Group 2016 2017 2018
American % District Population Indian/ % District Gifted Education Black^^ % Underrepresentation 31.07 18 42 31.43 17 46 31.57 17 46
% Equity Goal/Target for Gifted Equity Goal = Equity Goal = Equity Goal =
Representation (minimal 22.6 22.6 22.6
discrepancy using 20% formula) Increase from Increase from Increase from
18% to 22.6% 17% to 22.6% 17% to 22.6%
Multi-Racial % District Population 6.22 6.63 6.91
% District Gifted Education 6 6 6
% Underrepresentation 3.5 9.5 13.2
% Equity Goal/T arget for Gifted Equity Goal = Equity Goal = Equity Goal =
Representation (minimal 4.4 4.4 4.4
discrepancy using 20% formula) Minimal equity Minimal equity Minimal equity
goal of 4.4% goal of 4.4% goal of 4.4%
has been met. has been met. has been met.
White/Asian % District Population 62.71 61.95 61.52
% District Gifted Education 76 77 77
% Underrepresentation 0 0 0
% Equity Goal/T arget for Gifted O verrepre sented O verrepre sented O verrepre sented
Representation (minimal
discrepancy using 20% formula)
(Ford, 2013; Office of Assessment and Evaluation, District, 2018)
78


Second Screening as an Equity-Driven Alternative
It is evident from the literature that there is not a consensus on which testing measures to use. This is due to various studies with conflicting data (Lakin & Lohman, 2011; Naglieri & Ford, 2005). The practice of using one test or measure will continue to omit highly-able student groups, regardless of which measure is applied (Plucker & Peters, 2016). One test or measure is not enough (Ford, 2013b; Frasier, 1991). No one instrument should be used to exclude, nor include students in a program (Felder, et al., 2015; Ford, 2013b). Research proposes screening in multiple ways to capture highly-able students from various backgrounds and with various strengths (Ford, 2013b; Plucker & Peters, 2016). In order to do this, the researcher recommends a double screening. The second screening would be a focused screen through Title I schools, as Title I schools are federally designated as requiring additional support. These schools would be given a second screening test that is suggested to more equitably screen for diverse populations (Ford & Naglieri, 2003). Any student scoring in the 95th percentile or above on either screener would be referred for additional testing. Title I school data could be used in a follow-up study to determine if giving the second screener will indeed more equitably refer students, reducing disproportionality between racial groups.
Currently in District, students screened between the 90th and 94th percentiles are then given achievement testing. After this achievement testing, even with a 95th percentile or above in an academic content area, a student can only be placed in a talent pool designation until there is more evidence for gifted determination, per current Colorado Department of Education guidelines (CDE, 2016). As the guidelines state, a student must have three data points for any given academic content area in the 95th percentile to be recommended for gifted identification. With a score between the 90th to 94th percentile, a student may be recommended for talent pool
79


designation. Full identification cannot be given, unless a student is administered more testing beyond the one achievement test. As minoritized groups do not perform as high on achievement testing, there is controversy over whether this testing is biased against these groups, namely Black and Hispanic groups (Plucker & Peters, 2016). So, based on the evidence in this current study, lowering referral to the 90th percentile to increase equitable representation of minoritized groups, does little for Hispanic students. Although the Black referral rate increase was significant, the underrepresentation dilemma was still significant. When comparing how many White and Asian students are also referred at the 90th percentile, there were triple, quadruple, and more referred for gifted evaluation in these populations. This inequity continued when screening at the 90th percentile. As the data revealed, even at the 90th percentile, White and Asian students were still vastly over-referred, while Black and Hispanic students were vastly under-referred.
The researcher proposes more strategic use of the money spent on screening. Instead of screening for the 90th percentile for further testing, the researcher proposes the following as an equity-driven practice for gifted referral. The researcher suggests screening in a focused manner, using a combination of cognitive tests. CogAT (Verbal, Quantitative, and Nonverbal) would still be given as a universal screener across the district. This screener would be utilized, even with the inequity, because it draws out verbal and quantitative thinkers. Further academic testing, to build a body of evidence, would be recommended for any student scoring at the 95th percentile or higher. Students scoring 90th to 94th percentiles are watched and given access to opportunities to enrich their learning through advanced courses and/or enrichment. These students are referred later if their regular academic testing demonstrates strength areas. This would be on a case-by-case basis. An additional nonverbal test, suggested as a more equitable measure (Ford & Naglieri, 2003), will be administered at Title I schools.
80


For Title I schools, the screening practices would be slightly different. In addition to the universal screener given to all second graders across the district, Title I students would be given an additional nonverbal screener, the NNAT3, as this is suggested to more equitably screen for diverse populations (Ford & Naglieri, 2003). Any student at Title I schools scoring 95th percentile on either the CogAT7 or the NNAT3 would receive further academic testing. Title I schools are the focus for additional screening, as these schools are federally identified as having high populations of students in poverty (U.S. Department of Education, 2015) and potentially higher CLD populations (Mantosis, 2013).
Through a focused approach to gifted screening practices, using an equity-driven second screening, systemic inequities can be reduced. More research needs to be gathered to find pathways that demonstrate equitable referral practices. While the paperwork and identification processes catch-up, much can be done to develop the talents of all students, especially high-ability CLD learners.
Talent Enrichment Programs for High-Ability CLD Students
In addition to focused funding for cognitive and academic testing, the researcher recommends further use of and research for talent enrichment programs for high-ability CLD students. The researcher recommends collaboration with school districts that have published work around equity work for high-ability CLD learners. High-ability CLD learners are defined here as students scoring in the top quartile, namely the 75th percentile and above. There is overwhelming research suggesting that opportunities need to be given to students of color in order to bridge the opportunity and achievement gaps (Plucker & Peters, 2016). The researcher suggests a study where best practice CLD and gifted strategies are studied and the common overlapping strategies used to develop or determine curricula and/or strategies for high-ability
81


CLD learners. These strategies would be studied with high-ability CLD learners to test effectiveness in closing the achievement gap. The effectiveness of strategies and/or curricula would be studied using norm-referenced assessments.
Through developing high-ability CLD learner talents, the effects of opportunity and achievement gaps will be lessened. Students will have access to advanced coursework and receive scaffolded supports necessary for their success. In the end, while equitable referral and identification processes are adapted there will be access for current high-ability CLD learners to rigorous and advanced curricula.
Summary
Through providing focused use of resources and funds, more CLD students can be identified as exceptional and receive gifted services and opportunities. In this way, systems can move to increasing CLD populations in gifted programming, while striving for more equitable representation from White and Asian populations. It is clear that further research is necessary to decrease the inequity of systemic gifted screeners for referrals. Data in this field are needed to determine whether systemic changes are increasing the equitable representation of all racial groups in gifted programming. Then, further, there is need for rigorous advanced programming for and the retention of CLD students, through the use of CLD and gifted strategies together.
Final Thoughts
Inequity of gifted referral processes will not be resolved quickly; it is woven into current systemic practices (Ford, 2012; Sullivan, 2011). This study shows that even with moving to the 90th percentile as a referral marker, there are still statistically significant inequities in racial representation in gifted referral. Equity-driven practices such as (a) equity goals, (b) second screening processes, and (c) talent enrichment programs and/or strategies for CLD learners can
82


help mitigate the current issues surrounding inequity in gifted referral and thus programming. Further research is necessary to learn more regarding CLD gifted learners, for as Donna Ford (2013a) stated, “the more we know about others, the less we make up” (p. 66). As further study is done on CLD gifted referrals and programming, the field will move closer to utilizing practices that lead to more equitable representation in gifted populations.
83


REFERENCES
Anderson, S., Medrich, E., & Fowler, D. (2007). Which achievement gap? Phi Delta Kappan, 88, 547-550.
Baltimore County Public Schools. (2017). Planning the methodology: Limitations and delimitations. Retrieved from BCPS Independent Research Seminar: https://www.bcps.org/offices/lis/researchcourse/develop writing methodology limitatio ns.html
Beisser, S. R. (2008). Unintended consequences of the no child left behind mandates on gifted students. Forum on Public Policy: A Journal of the Oxford Round Table, 1-13.
Borland, J. H. (2003). The death of giftedness: Gifted education without gifted children. Rethinking Gifted Education, 10, 105.
Borland, J. H. (2009). Myth 2: The gifted constitute 3% to 5% of the population. Moreover, giftedness equals high IQ, which is a stable measure of aptitude: Spinal tap psychometrics in gifted education. Gifted Child Quarterly, 55(4), 236-238.
Boyles, D., Carusi, T., & Attick, D. (2009). Historical and critical interpretations of social
justice. In W. Ayers, T. M. Quinn, & D. Stovall (Eds.), Handbook of Social Justice in Education (pp. 30-42). New York: Routledge.
Business Wire. (2013, November). Houghton Mifflin Harcourt's Cognitive Abilities Test (CogAT) evolves to meet the needs of English language learners. Retrieved from Business Wire: http://www.businesswire.com
CDE: Colorado Department of Education. (2018). Colorado education facts andfigures.
Retrieved from https://www.cde.state.co.us/communications/coeducationfactsandfigures
CDE: Office of Gifted Education (2016). Gifted identification. Colorado Department of
Education, Office of Gifted Education. Denver: CDE. Retrieved from CDE Colorado Department of Education: https://www.cde.state.co.us/gt/identification
District (2017). Definition and identification. Retrieved from Advanced Academic Services: web address not available.
Colangelo, N., Assouline, S. G., & Gross, M. U. (2004). A nation deceived: How schools hold back America's brightest students. The Templeton National Report on Acceleration. International Center for Gifted Education and Talent Development.
Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J., Mood, A. M., Weinfeld, F. D., & York, R. L. (1966). Equality of educational opportunity. U.S. Department of Health, Education, and Welfare. Washington: National Center for Educational Statistics.
84


Collins, C., & Yeskel, F. (2013). The dangerous consequence of growing inequality. In M. Adams, W. J. Blumenfeld, C. Castaneda, H. Hackman, M. L. Peters, & X. Zuniga, Readings for Diversity and Social Justice (3rd ed., pp. 157-164). New York: Routledge.
Colorado Department of Education (2017). CMAS - Mathematics, English Language Arts, Science and Social Studies Data and Results. Retrieved from Colorado Department of Education: https://www.cde.state.co.us/assessment/cmas-dataandresults
Cornell, D. G., Delcourt, M. A., Goldberg, M. D., & Bland, L. C. (1995). Achievement and self-concept of minority students in elementary school gifted programs. Talents and Gifts, 75(2), 189-209.
Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed methods approaches (5th ed.). Los Angeles: Sage.
Egalite, A. J., Fusarelli, L. D., & Fusarelli, B. C. (2017). Will decentralization affect
educational inequity? The Every Student Succeeds Act. Educational Administration Quarterly, 53(5), 757-781.
Encyclopedia Britannica. (2016, March). National Defense Education Act (NDEA). (T. C. Hunt, Editor) Retrieved from Encyclopedia Britannica: https://www.britannica.com/topic/National-Defense-Education-Act
Erwin, J. O., & Worrell, F. C. (2012). Assessment practices and the underrepresentation of minority students in gifted and talented education. Journal of Psychoeducational Assessment, 30(1), 74-87.
Exceptional Student Services Unit. (2016, March). ECEA rules: Rules for the administration of the exceptional children's educational act. Retrieved from CDE: Colorado Department of Education: Retrieved from
https://www.cde.state.co.us/sites/default/files/documents/spedlaw/download/ecearules-
march2013.pdf
Felder, M. T., Taradash, G. D., Antoine, E., Ricci, M. C., Stemple, M., & Byamugisha, M. (2015). Increasing diversity in gifted education: Research-based strategies for identification and program service. Waco, Texas: Prufrock Press Inc.
Feldman, D. H. (2003). A developmental, evolutionary perspective on giftedness. In J. Borland (Ed.), Rethinking gifted education, 9-33. New York: Teachers College Press.
Ford, D. Y. (1998). The underrepresentation of minority students in gifted education: Problems and promises in recruitment and retention. The Journal of Special Education, 32(1), 4-14.
Ford, D. Y. (2011). Multicultural gifted education. Waco, Texas: Prufrock Press Inc.
85


Full Text

PAGE 1

INCLUSIVE PATHWAYS TO GIFTED EDUCATION: EXAMINING GIFTED REFERRAL PROCESSES by MELISSA DAYLE DURTSCHI B.S., Brigham Young University, 2004 M.A ., Adams State University, 2016 A dissertation submitted to the Faculty of the Graduate School of the Uni versity of Colorado in partial fulfillment of the requirements for the degree of Doctor of Education Leadership in Educational Equity 2019

PAGE 2

ii ! 2019 MELISSA DAYLE DURTSCHI ALL RIGHTS RESERVED

PAGE 3

iii This dissertation for the Doctor of Education degree by Melissa Dayle Durtschi has been approved for the Leadership for Educational Equity by Connie L. Fulmer, Chair Rod ney L. Blunck James Christense n Date: M ay 18 , 2019

PAGE 4

iv Durtschi, Melissa Dayle ( EdD , Leadershi p in Educational Equity ) Inclusive Pathways to Gifted Education: Examining Gifted Referral Processes Dissertation directed by Professor Connie L. Fulmer ABSTRACT This study explored the affect race had on gifted referrals at the 90 th and 95 th percentiles, based on the Cognitive Abilities Text 7 (CogAT 7). Achievement, Opportunity and Excellence Gaps are defined and put into historical context including systems of power and privilege. Giftedness is defined. The history of Gifted Education is explored throu gh a Culturally and Lingu istically Diverse (CLD) lens. Research questions explored race and its a ffect on the Cognitive Abilities Test, CogAT 7 , and the significance of difference between racial groups at the 90 th and 95 th percentiles. The results of thi s study indicate : (a) CogAT 7 test scores are affected by race, (b) 95 th percentile gifted referrals were significantly different across race, (c) 90 th percentile gifted referrals were significantly different across race, (d) when comparing 90 th and 95 th p ercentile referrals, there was a significant difference, and (e) gifted referrals at the 90 th and 95 th percentiles were significantly not equitable across race. Recommendations for further study include the development of district equity goals, second scr eening process es, and talent enrichment programs and/or strategies for CLD learners . The form and content of this abstract are approved. I recommend its publication. Approved: Connie L. Fulmer

PAGE 5

v DEDICATION For David and Mom. If I had a degree to hand out, I would give it to my husband, David, for most supportive partner. You have stood by my side through the late nights, all nighters, melt downs, watched the kids, and loved me through this. Without your loving support, this would not be possibl e. For Mother, who encourages me and teaches me that I can be and do anything.

PAGE 6

vi ACKNOWLEDGEMENTS I want to acknowledge the efforts of my support network through this process. It is because of the following people that I was able to complete this wor k. My Dissertation Chair, Dr. Connie L. Fulmer. Without Dr. Fulmer's guidance, I would have been lost in the vastness of "All but Dissertation" land. She helped guide and direct me, while holding me accountable to deadlines I set for myself. Thank yo u Dr. Fulmer. Thank you for your expertise and wisdom having been down this road before. My Father, Editor Extraordinaire. Having written detailed reports from his days in the military, my father knows what it takes to be concise and clear in written commu nication. With his writing expertise , I am a better writer for it. Thank you Dad for your efforts to make my drafts the best they could be. Graduate Assistant and Sister from Another Mother, Dara Marin Prais. What can I say Dara? You are the other half of my brain. You have always understood me and for some reason things are always better after talking with you. Thank you for your time and dedication to mentor me through this process. You listened, guided, nudged, and generally taught me to be the best version of myself possible. My fellow Doctoral Candidates. I could not have done this without you. Sarah Famularo, who studied with me at Kneaders and listened to the hardships of each week along the way. Jessica Slattery, who always help ed me gain perspective and keep my eyes focused on my dreams. Jeremy Koselak, who kept me grounded in strong principles of school systems. Rob Thelen, who would send me funny texts to keep me light hearted. And Jon Ail, who showed me that through trial and hardships, you can emerge victorious. Thank you to each of you. Having others go through this experience with me helped forge friendships I hope to continue well beyond this Doctoral program. My mentor, Ashley Gehrke. She is an amazing example to me of the kind of leader I hope to be someday. She has taught me to put the "Gehrke Touch" on situations and weave it into conversations with others. Her sincere desire to serve others and collaborate resonates with me and I will be forever grateful for her guidance . Her leadership is instrumental in shaping who I aspire to be as a leader.

PAGE 7

vii TABLE OF CONTENTS CHAPTER I. ! INTRODUCTION ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É. . 1 Background ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ . 2 Achie vement Gap, Opportunity Gap, and Excellence Gap ÉÉÉÉ ÉÉÉÉ 4 The Problem of Practice ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ. 6 District Perspective ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É É 7 Conceptual Framework ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É.. 10 Research Questions ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 11 Research Assumptions ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉ É 12 Significance of Study ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É. 12 Limitations and Delimitations ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 13 Summary ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 13 II. ! LITERATURE REVIEW ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. 15 Systems of Power and Privilege ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 15 School Systems and Power and Privilege ÉÉÉÉÉÉÉÉÉÉÉ. 15 Historical Context ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. 16 More Recent Policies ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. 18 Definitions of Giftedness ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 19 Culturally and Linguistically Diverse Gifted Education ÉÉÉÉÉÉÉÉÉ 21 Gifted Identification with a Cultura lly and Linguistically Diverse Len s ÉÉ. 23 Colorado Department of Education ÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉ 24 District ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉ.. 25

PAGE 8

viii Cognitive Abilities Test, CogAT ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 25 Summary ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 28 III. ! METHODOLOGY ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 29 Research Design ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 29 Site Selection ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 30 Subjects ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 30 Variables ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 30 Conceptual Framework ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 31 Cognitive Abilities Test ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 32 R acial Groups É ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 32 Gifted Referral ÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉ.. 32 Percentiles ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 32 Research Questions ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 33 Instrument, CogAT 7 ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 34 Data Collection and Storage ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . . 34 Data Analysis ÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 35 Limitations and Delimitations ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 36 Ethical Considerations ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 37 Participants ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 37 Sponsors ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 37 Field of Research ÉÉÉÉÉÉÉÉÉÉÉ ÉÉ ÉÉÉÉÉÉÉÉ 38 Research Community ÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉ 38

PAGE 9

ix IV. ! RESULTS ... ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É .. 39 Descriptive Statistics ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É . 40 Valid and Missing Cases ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉ 41 Testing for Normal cy ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ 41 Central Tendency by Race and Battery ÉÉÉÉÉÉÉÉÉÉÉ É 42 Comparisons Across Administration Years Within Batteries ÉÉ ÉÉ 43 Verbal Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ 43 Quantitative Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É... 45 Nonverbal Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ.. 47 Summary ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉ 49 Re search Question 1: Does race affect test scores from the Cognitive Abilities Test, CogAT 7? ÉÉÉÉÉÉÉÉÉÉÉÉÉ . 49 Verbal Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ 50 Quantitative Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É.. 51 Nonverbal Battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. 53 Answering Research Question 1: Does race affect test scores from the Cognitive Abilities Test, CogAT 7 ? ÉÉÉÉÉÉÉÉÉÉÉÉ ... 54 Research Question 2: What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups? ÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉ.. 55 Research Question 3: What is the difference in gifted referral, based on CogAT 7, at the 90 th percentile and above, between racial groups? ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. 58

PAGE 10

x Research Question 4: What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile? ÉÉÉÉÉÉÉÉÉÉÉÉ 62 95 th P ercentile Referral by Race, Chi Square Goodness of Fit ÉÉÉ 63 90 th P ercentile Referral by Race, Chi Square Goodness of Fit ÉÉÉ 64 Overall Gifted Referral Rate Disaggregated by Race ÉÉÉÉÉÉ É 66 Summary ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 67 V. ! DISCUSSION ÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 69 Summary of StudyÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 69 FindingsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 70 Race and Test ScoresÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉ ÉÉÉ. 70 95 th Percentile ReferralsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 71 90 th Percentile ReferralsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 73 Comparison Between 90 th and 95 th Percentile ReferralsÉÉÉÉÉÉ 74 ConclusionsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 75 Implications and Future ResearchÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉ . 77 Equity GoalsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. . 77 Second Screening as an Equity Driven Alternative ÉÉÉÉÉÉÉ . 79 Talent Enrichment Programs for High Ability CLD Students ÉÉÉ . 81 Sum mary ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ... 82 Final Th oughts ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .. 82 REFERENCES ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ... 84

PAGE 11

xi APPENDIX A. University of Colorado Approval of Research LetterÉÉÉÉÉÉÉÉÉ... 92 B. District Research Approval LetterÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ... 93

PAGE 12

xii LIST OF TABLES TABLE 1. CMAS proficiency levels 2016 2017, English language arts, g rade 3 CMAS proficiency levels 2016 2017, mathematics, g rade 3ÉÉÉÉÉÉÉÉ . 3 2. CMAS % exceeded expectations 2016 2017, English language a rts and m athematics, Grade 3 ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉ É 6 3. Disaggrega ted student e nrollment, District K 12, 2015 20 18ÉÉÉÉÉÉÉÉ 9 4. District enrollment c ompared to gifted i dentification, disaggregated by ethnicity ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉ .. 10 5. Chances of being p oor in America ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .. 2 1 6. Opera tionalizing v ariables ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .. 31 7. Observations by b attery 2016, 2017, and 2018 c ombined ÉÉÉÉÉÉÉÉÉ .. 41 8. Testing for normalacy by b attery 2016, 2017, and 2018 c ombined ÉÉÉÉÉ É 42 9. Descriptive statistics, all CogAT 7 battery scores by race, including 2016, 2017, 2018 data ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 43 10. Comparison of CogAT scores for administration years (2016, 2017, 2018), verbal battery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ .. 45 11. Comparison of verbal b attery CogAT s cores for a dmin istration y ears , t testsÉ ... 45 12. Comparison of CogAT s cores for a dministration years (2016, 2017, 2018), quantitative b attery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 47 13. Comparison of q uantitative b attery CogAT s cores for administration y ears ÉÉ .. 47 14. Comp arison of CogAT s cores for a dministration y ears (2016, 2017, 2018), nonverbal b attery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ É 48

PAGE 13

xiii 15. Comparison of nonverbal b attery CogAT s cores for a dministration y ears ÉÉÉ 48 16. Population d escriptives , Verbal battery, used for AN OVA t esti ngÉÉÉÉÉÉ 50 17. Games Howell post h oc comparing verbal CogAT scores across r ace É ÉÉÉ .. 51 18. Population d escriptives, Quantitative battery, used for ANOVA t esting ÉÉÉ É 52 19. Games Howell post hoc c omparing q uantitative CogAT scores across r ac e ÉÉ. 5 2 20. Population d escriptives, Nonverbal battery, used for ANOVA t esting ÉÉÉÉ .. 53 21. Games Howell post hoc c ompar ing nonverbal CogAT scores across r ace ÉÉÉ 54 22. R elative frequency of gifted r eferral (95 th p ercentile) by r ace for each CogAT b attery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 56 23. Relat ive frequency of gifted r eferral (95 th p ercentile) by r ace ÉÉÉÉÉÉÉ É 56 24. Population Descriptives, 95 th Percentile Gifted Referrals, Used for ANOVA Testing ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 57 25. Ga mes Howell post hoc c omparing 95 th p er centile gifted r eferral across r ace É .. 58 26. Relat ive frequency of gifted r eferral (90 th p ercentile) by r ace for each CogAT b attery ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 59 27. Relative f requency of 90 th p ercentile gifted r eferral by r ace ÉÉÉÉÉÉÉÉ .. 60 28. Population d escriptives, 90 th percentile gifted r eferrals, used for ANOVA t esting ÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 61 29. Games Howell post hoc c omparing 90 th percentile gifted r eferral across r ace É .. 61 30 . Number, mean , and standard deviation of stratified random s ampling, o ne way MANOVA ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉ 62 31. Gifted r eferral by r ace at the 95 th p ercentile, Chi Square Goodness of Fit ÉÉÉ 63 32. Gifted r eferral by r ace at the 90 th p ercentile, Chi Square Good ness of Fit ÉÉÉ 65

PAGE 14

xiv 33. Percent of gifted r eferral by r ace at the 90 th and 95 th p ercentiles ÉÉÉÉÉÉÉ 67 34. Relative f requency of 90 th and 95 th percentile gifted referral by r a ceÉÉÉÉ É 73 35. Percent of gifted referral by r ace at the 90 th and 95 th p ercenti les ÉÉÉÉÉÉÉ 74 36. Colorado s tate student population t rends 2010 2017 ÉÉÉÉÉÉÉÉÉÉÉ. 76 37. District enrollment compared to gifted i dentification, disaggregated by r ace ÉÉÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. 78

PAGE 15

xv LIST OF FIGURES FIGURE 1. Inclusive excellence adaptive framework ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 8 2. Conceptual framework including comparison between 90 th and 95 th percentile referralsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 11 3. Referral processes as outlined by C olorado Department of EducationÉÉÉÉÉ 24 4. Conc eptual framework ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ . 31

PAGE 16

1 CHAPTER I INTRODUCTION As a Gifted Specialist, who has worked in varying ethnic and socio economic environments, I am deeply concerned about gifted opportunity inequities across demographic groups. I see fina ncially limited parents struggling to make a living; wanting the best for their children, yet unsure how to provide that opportunity. There are children yearning to learn, yet struggling to "catch up" t o their same age, White peers. After receiving my M asters in Culturally and Linguistically Diverse (CLD) education, I realized, although important, implementing CLD strategies was not enough. School s ystems need to change. A ccess to gifted education needs to systematically shift to provide equitable oppo rtunities in gifted education. This study focused on recent screening practices for gifted referral with in an urban to suburban school district in Colorado. For the purposes of this study, the school district name has been de identified and given a pseud onym of District. Discussion within D istrict abounded regarding the underrepresentation of diverse po pulations in gifted education. With recent changes at the Colorado Department of Education (CDE) in respect to gifted ident ification (CDE, 2016), there ha d been discussion around screening and referral processes, particula rly for diverse populations. The scope of this current study was to determine if equitable referral representation existed across racial groups from the Cognitive Abilities Test (CogAT) , Form 7 (Lohman, 2012), a universal screener used in D istrict . In this study , universal screener refers to a given test being administered across a group of students, in this case all second graders. It is not to be confused with the CogAT screening test, which includes three subtests. The test used was the full nine subtest version of CogAT 7. Since the sch ool district used 90 th percentile cutoffs for referral purposes, the researcher was

PAGE 17

2 interested in comparing the 90 th percentile cutoff with the 95 th p ercentile cutoff; which, if any, was a better representation of the student body. Background Despite the Civil Rights movement 's contribution to school desegregation, persistent inequities in the United States school system continue to exist (Ford, 2011 ). Prejudice is usually based on very limited perspective (Tatum, 2013). Prejudice is defined here as , "preconceived judgments toward people or a person because of race, ethnicity, nationality, religion, socioeconomic status, gender, age, disability, and other stereotypes" (Ford, 2013a, p. 63). The phrase minoritized is used in this study to denote, "people who endure mistreatment, and face prejudices that are enforced upon them because of situations outside of their control" (Odyssey, 2016). Minoritize d populations here are Black, Hi spanic , and American Indian populations. Within school systems, limiting perspective plays out with minoritized groups ' over identification for special education and underrepresentation in gifted programs. This discrepancy is related to the achievement gap , a stat istical analysis of performance based assessments determining how well a student does in school, usually between minorit ized and/or low socio economic students and their White and/or Asian peers (NEA, 2017). The co ntinued underrepresentation of minoritized groups in gifted education and the overrepresentation in special education, "strongly indicates systemic problems of inequity, prejudice, and marginalization within the education system" (Sullivan, 2011). In her book, Multicultural Gifted Education , Donna Ford (2011) cites statistics in the Condition of Education 2010 report (NCES, 2010) stating that while minorit ized populations continue to increase, there has not been a significant c hange in gifted identificatio n. White students are continuing to be over identified, while minori tized groups are underrepresented.

PAGE 18

3 Data from the 2016 2017 school year, Table 1, from CDE (2017) indicate student proficiency levels in grade three for English Language Arts (ELA) and M athematics. Table 1 CMAS Proficiency Levels 2016 2017, English Language Arts, Grade 3 Demographic # Valid Scores Mean Scale Score Standard Deviation % Did Not Yet Meet Expectations % Partially Met Expectations % Approaching Expectations % Met E xpectations % Exceeded Expectations American Indian 418 720 39 32.3 23.0 22.2 21.5 1.0 Asian 1,950 751 41 12.7 11.9 21.9 46.4 7.1 Black 3,031 722 38 31.6 21.1 22.5 23.6 1.2 Hispanic 21,215 723 37 29.0 22.8 23.8 2 3.2 1.1 White 33,907 748 38 11.5 14.2 23.9 45.8 4.6 Multi Racial 2,906 745 39 13.2 16.1 25.1 41.5 4.2 CMAS Proficiency Levels 2016 2017, Mathematics, Grade 3 Demographic # Valid Scores Mean Scale Score Standard Deviation % Did Not Yet Meet Expectations % Partially Met Expectations % Approaching Expectations % Met Expectations % Exceeded Expectations American Indian 422 724 37 26.5 25.8 24.2 17.5 5.9 Asian 1,981 757 39 7.7 12.1 19.4 40.3 20. 5 Black 3,034 723 35 26.7 24.1 26.3 20.1 2.8 Hispanic 22,954 725 34 23.7 25.7 26.4 21.3 2.9 White 33,931 749 35 8.3 14.8 26.2 38.4 12.4 Multi Racial 2,915 744 37 11.8 16.6 26.9 34.1 10.5

PAGE 19

4 American Indian, Black, and Hispanic students score significantly lower than their Asian and White counterparts. This trend in lower performance, known as the achievement gap, for ethnically diverse students is seen across the nation, not just in Colorado (Ford, 2013 b ; Miller, 2004; Olszewski Kubilius & Thomson, 2010; Plucker & Peters, 2016). With these trends, the need for equitable practices has become a necessity (Ford, 2012). Achievement Gap, Opportunity Gap, and Excellence Gap The phrase achievement gap can be controversia l, as it can denote more than just a lack of performance of minorit ized groups (Perry, Steele, & Hilliard, 2003). Hillard (2003) explains, "Framing the problem this way [as achievement gap] is itself problematic. Importantly, it establishes European aver age achievement as the universal norm" (p. 137). While the researcher advocates for change in using the term achievement gap , its use is well documented in the literature and therefore, for the purposes of this study, will be used when referring to the di screpancies in academic performance. In addition, the researcher uses the term achievement gap to indicate the discrepancy of achievement scores without placing a value on various demographic groups. The discrepancies in proficiency levels begin well befo re the third grade (Ford, 2011; Miller, 2004). They exist in opportunities outside of school, to which, low socio economic youth may not have access (Miller, 2004) . This is referred to as the opportunity gap . The use of the term opportunity gap refers t o "the unequal or inequitable distribution of resources and opportunities Ñ while achievement gap refers to outputs Ñ the unequal or inequitable distribution of educational results and benefits" (Great Schools Partnership, 2013, p. 1). It is the, "cumulative differences in access to key educational resources" (Darling Hammond, 2013, p.

PAGE 20

5 77). When denoting the cause to this achievement gap, the researcher will use the phrase opportunity gap . The opportunity gap can also be thought of as an educational debt , whe rein groups of people have been denied access to quality education for extended periods of time (Ladson Billings, 2013). Ladson Billings discussed this educational debt as one of moral importance in a nation where groups of people have been oppressed and then released. For example, Ladson Billings cited Preside nt Lyndon B. Johnson's policy on Affirmative A ction, stating, "it is unfair to keep people shackled for centuries, unshackle them, and then expect them to compete against those who have never known such restrictions" (p. 17). Therefore, there is a moral obligation for those making educational structures and systems to look through the cultural lens of equity. Rothstein (2013) stated that in order to address the issues facing school systems regardin g the opportunity gap , school improvement and efforts that focus on socioeconomic inequalities need to be combined . One other gap that needs to be addressed here is the excellence gap , those gaps in high performance between demographic groups on cognitive and academic testing (Plucker & Peters, 2016). Not only do discrepancies exist within basic proficiencies, but also in advanced scores (Plucker & Peters, 2016). Take, for exampl e, the data shared regarding CDE (2017) third grade results for ELA and Mathe matics (see Table 2 ). Table 2 indicates , especially in mathematics, a clear excel lence gap discrepancy; from 2.8 to 5.9 percent for Black and Hispanic , and American Indian p opulations to 12.4 and 20.5 percent in White and Asian populations respectively. The gaps in achievement are not localized to achieving proficiency; they are pervasive, even in gifted populations. This systemic discrepancy leads to further gaps when , because of excellence gaps, minoritized groups are left out of gifted

PAGE 21

6 program ming . M ajority groups end up achieving higher, as minoritized groups fall further behind. Data suggests that the United States' educational syste m perpetuate s the gaps through current screening processes for gifted referral and evaluation (Ford, 2011). Table 2 C MAS % Exceeded Expectations 2016 2017, English Language Arts and Mathematics, Grade 3 Third Grade English Language Arts Third Grade Mathematics Demographic % Exceeded Expectations Demographic % Exceeded Expectations American Indian 1.0 Ame rican Indian 5.9 Asian 7.1 Asian 20.5 Black 1.2 Black 2.8 Hispanic 1.1 Hispanic 2.9 White 4.6 White 12.4 Multi Racial 4.2 Multi Racial 10.5 In a study, focused on social class standing in relation to educational acc ess, Anyon (1980) relates observations and themes that emerged from a study done with students attending various schools with varying social classes. He relates that the lower the class standing at a given school, the more rote memorization and menial tas ks were observed. They related this to preparing the students for a workforce that would keep them in their current social class. As one moved up in social class as a school, there were more resources available and accessed by students. There was also a marked increase in creativity and expectations of teachers. In conclusion, this article relates higher class to higher expectations of students, and therefore, higher performance and autonomy. The researcher argues that inequitable access is at the hear t of underrepresentation in this case and generally in gifted education. The Problem of Practice All the changes in standards, such as Common Core, and re evaluations of laws ( i . e ., Elementary and Secondary Education Act and No Child Left Behind) have do ne little to effect

PAGE 22

7 change in the achievement gap (Plucker & Peters, 2016). The Every Student Succeeds Act (ESSA), authorized by President Obama in 2015, authorized funds for the Javits Gifted and Talented Students Education Program. It also allowed Titl e I funds to be used for programming needs of gifted students (Plucker & Peters, 2016). In order to continue moving toward more equitable gifted representation, access to gifted programming needs to be addressed (Plucker & Peters, 2016). One of the ways to do this is through gifted referr al, which leads to i dentification. Without change s to gifted referral, gifted programming continues to be overrepresented by White and Asian students; wh ile Black, Hispanic, and American Indian populations continue to be underserved. District Perspective The school district , as a whole, is striving for excellence for all its students. The Office of Inclusive Excellence , states its purpose is to, "serve as a cultural liaison between district and its diverse community, w orking to promote racial, ethnic, linguistic, religious and cultural understandings . . ." (Office of Inclusive Excellence, 2018, p. 2). In recent years, there has been a systematic push for inclusive excellence through required professional development, Beyond Diversity, for each staff member. This professional development increases staff's racial self awareness, thus heightening sensitivity for diverse populations . Part of the mission of the Office of Inclusive Excellence is to promote the follo wing f ramework (see Figure 1). District is fulfilling its goals of Racial Consciousness and Cultural Competence through professional development , such as Beyond Diversity, and training with Dr. Stembridge , a multicultural researcher. C apacity of the district i s being built to encompass a more culturally responsive environment .

PAGE 23

8 Figure 1. Inclusive excellence adaptive framework . Along with the work around Beyond Diversity, District has also partnered with Dr. Yemi Stembridge, who provides "technical assis tance for school improvement with a specific focus on equity" (My Reflection Matters, 2018, p. 1). This work is tracked in various formats, including Twitter, where Dr. Stembridge (2017) related, "Racial consciousness is the foundation of equity work beca use our country has deeply problematic and unresolved issues with race" (Retrieved from https://twitter.com/DrYemiS). The inclusive excellence work happening courageo usly in District needs to expand in efforts for more inclusive programming within gifted education. One of the goals from the Office of Inclusive Excellence includes, "promoting student access to rigorous learning opportunities" (Office of Inclusive Excellence, 2018, p. 8). One of the varied ways to fulfill this goal would be to evaluate cu rrent pathways to gifted education. The ultimate goal would be to provide additional access to gifted programming for our diverse gifted learners, through updating referral systems with an equity lens. Table 3 details the trends

PAGE 24

9 in cu rrent student enroll ment in District . Student enrollment is disaggregated by ethnic groups; grou ped by minoritized (American Indian , Black, and Hispanic) and majority (White and Asian). Asian is included in the majority designation, even though considered a minority, becaus e they are traditionally overrepresented in gifted populations (Plucker & Peters, 2016; Worrell, 2014). Table 3 Disagg regated Student Enrollment, District K 12, 2015 2018 Ethnic Groups 2015 2016 2016 2017 2017 2018 American Indian /Black/His panic # of students in ethnic groups 16,051 16,240 16,752 Total # of students enrolled 51,663 51,678 53,059 Percent of ethnic group compared to total population 31.07 31.43 31.57 Multi Racial # of students in ethnic group 3,216 3,424 3,665 Tot al # of students enrolled 51,663 51,678 53,059 Percent of ethnic group compared to total population 6.22 6.63 6.91 White/Asian # of students in ethnic group 32,396 32,014 32,642 Total # of students enrolled 51,663 51,678 53,059 Percent of ethn ic group compared to total population 62.71 61.95 61.52 (Offic e of Assessment and Evaluation , 2018) As evidenced by comparison data shown in Table 4 , current and historical gifted population s do not equitably represent the District student populat ion; White and Asian groups are disproportionately overrepresented, while Black, Hispanic, and American Indian groups are underrepresented (Ford, 2013 b ; Office of Assessment and Evaluation, District, 2018) . Current research that identifie s the reasons for this inequity is pervasive across the nation (Borland, 2003; Borland, 2009; Felder et al., 2015; Ford, 1998; Ford, 2011; Ford, Moore, & Scott, 2011; Ford, 2013 b ; Whiting & Ford, 2009; Worrell, 2014).

PAGE 25

10 Table 4 District Enrollment Compared to Gifted Identi fication, Disaggregated by Ethnic ity Ethnic Group 2016 2017 2018 American Indian / Hispanic/ Black % District Population 31.07 31.43 31.57 % District Gifted Education 18 17 17 % Underrepresentation 42 46 46 Multi Racial % District Populat ion 6.22 6.63 6.91 % District Gifted Education 6 6 6 % Underrepresentation 3.5 9.5 13.2 White/Asian % District Population 62.71 61.95 61.52 % District Gifted Education 76 77 77 % Underrepresentation 0 0 0 From these data, it is evident that exc ellence gaps exist and are similar to Colorado state and national data on underrepresentation of diverse populations. Through this study, the researcher intends to identify equitable referral of diverse populations. Conceptual Framework District uses the Cognitive Abilities Test 7 ( CogAT 7 ) as a cognitive measure of student ability. The researcher was interested in d etermining whether race affected gifted referr al in District , based on the use of CogAT 7. Figure 2 represents the conceptual framework of th is study. The conceptu al framework elements include: CogAT, racial groups, gifted referral, and percentiles . These are o perationalized in Chapter III. It is sufficient to include the overview of the conceptual framework here.

PAGE 26

11 Figure 2. Concep tual framework including comparison between 90 th and 95 th percentile referrals . Research Questions The research questions explored in this study include: 1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7? H 0 = Race does not affect test scores from the Cognitive Abilities Test. H 1 = Race does affect test scores from the Cognitive Abilities Test. 2. What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups? H 0 = There is no d ifference in gifted referral between racial groups at the 95 th percentile and above. H 1 = There is a difference in gifted referral between racial groups at the 95 th percentile and above. 3. What is the difference in gifted referral, based on CogAT 7, at t he 90 th percentile and above, between racial groups?

PAGE 27

12 H 0 = There is no difference in gifted referral between racial groups at the 90 th percentile and above. H 1 = There is a difference in gifted referral between racial groups at the 90 th percentile and above . 4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th p ercentile? H 0 = There is no significant difference in gifted referral between racial groups when using the 90 t h p ercentile compared to the 95 th percentile. H 1 = There is a significant difference in gifted referral be tween racial groups when using the 90 th percentile compared to the 95 th percentile. Research Assumptions The research assumptions of this study inclu de: (a) the instrument, CogAT 7, will elicit valid and reliable scores, (b) the data collected will be from second g rade screening processes in District , and (c) screening for gifted referral elicits more equitable referrals than teacher referral alone. Si gnificance of Study This study was created to compare referral processes across demographics using the Cognitive Abilities Test, Form 7 (Lohman, 2012), a universal scre ener used in District . In the end, the researcher compare d screening processes at the 9 0 th percentile and above compared to that the 95 th percentile and above to identify any difference in representative referral demographics for gifted evaluation. The study was intended to be used when making decisions around using percentile cut offs for gifted referral.

PAGE 28

13 Limitations and Delimitations Since this stu dy was conducted within a school district, it is limited to that school district. This study can be replicated through the us e of similar methodology, but was not intended to be generalized to a larger population. The limitations are defined as the restrictions not within the researcher's control, while the delimitations are limits the research has placed on the study (Baltimore County Public Schools, 2017). Li mitations to this study include (a) this study is not intended to be generalizable to a greater popula tion, as it was done within District and (b) this study is not seeking to determine causation, only relation. However, this study might be useful as a framework for other district equit ability studies. The following are delimitations of the study (a) data is narrowed to 2nd graders administered the CogAT 7 during the 2015 2016, 2016 2017, and 2017 2018 school years, (b) students with invalid scores in one or more batteries will be elimi nated from the data set, and (c) students whose parents opted out of testing will not be included in the data set. Summary This chapter discussed underrepresentation of minority groups namely: American Indian , Hispanic, and Black in current gifted program s across the nation, state, and district. The constructs of achievement gap, opportunity gap, and excellence gap are discussed and operationalized. Achievement gap is defined as the discrepancy between minority (American Indian, Hispanic, and Black) and their majority counterparts (White and Asian). Opportunity gap is defined as the discrepancy of opportunities available to minority groups compared to their majority counterparts. Finally, the excellence gap is defined as the discrepancy of achievement scores between minority and majo rity groups. These gaps lead to underrepresentation in gifted populations.

PAGE 29

14 The study was situated in District, an urban to suburban school district in the western United States . The equity work in District is described, in cluding collaborative efforts with Dr. Yemi Stembridge. The discrepancy of demographic representation is presented. A conceptual framework for comparing the current gifted referral screener, CogAT 7, at the 90 th percentile and the 95 th percentile is esta blished. A comparison model is presented for the study focus. The significance of the study is defined as establishing equitable gifted referral prac tices that are inclusive to diverse, traditionally underserved populations. Research questions, assumpti ons, limitations, and delimitations are outlined and described.

PAGE 30

15 CHAPTER II LITERATURE REVIEW In this review, recent literature is explored on giftedness through a culturally and linguistically diverse lens. In order to set context for this diversity lens , the history of systems of power and privilege is also related. Four major themes emerged from the literature review and form context and meaning around the current study. These themes include: (a) definitions of giftedness, (b) culturally and linguisti cally diverse (CLD) gifted education, (c) gifted identification with a CLD lens, and (d) Cognitive Abilities Test (CogAT). Systems of Power and Privilege Power and privilege exist in every part of our world. Since we are social beings and interact with ot hers around us, power and privilege are inherent in society (Gallagher, 2003). P ower and privilege are defined as (a) power being the ability to choose the norm or what normal looks and acts like and (b) privilege as being part of the dominant group that chooses, regardless of whether or not personally participating in the decision (Johnson, 2006). These two ideas work together to create stratification and difference in social sys tems. In this section, the system of school is defined through historical policies, structure, and norms as they relate to racial identity. Then, institutionalization of privilege and oppression is discussed and its implications on society at large and specifically within a school system for gifted learners. School Systems and Power and Privilege T he NAGC 2012 2013 State of the Nation reports that half of the states are citing more inclusive definitions of giftedness, only five included culturally diverse populations, five included students from low socioeconomic (SES), three in cluded English language learners (ELLs), and only two included those with disabilities (NAGC, 2013). Clearly, with only a few

PAGE 31

16 states reporting conditions for minority groups, there is need for diverse reform in gifted education. If there are not definiti ons that are inclusive, how are systems searching for marginalized groups to ensure inclusion in gifted programs? Historical Context The fight for social justice has been around for centuries, beginning with the ideas of Socrates ' and Aristotle 's ideas of justice. T hey searched for the meaning of justice and how to define it (Boyles, Carusi, & Attick, 2009). For the purpose of this study, it is enough to state that over the centuries definitions of social justice have been redefined. Yet, equity remains at the ecenter of it all. This literature review is focused on the development of social justice over the last fifty years, specifically on social justice in gifted education. The legal end of racial segregation within school systems is evident in the lan dmark case of Brown vs. the Board of Education in 1954. This case marked a step forward in systemic change. H owever, with it came a restructuring that resulted in great controversy and push back (Boyles, et al., 2009). Even though schools were integrated , how integration was to be implemented differed across the country. Although this was one step for racially diverse groups of students, it was not until the 1970s that federal mandates for special education were implemented . Prior to the 1970s, states w ere allowed significant freedom regarding enrollment , often marginalizing disabled and racial groups (Martin, Martin, & Terman, 1996). This resulted in refusal of services and misplacement of student s (Martin, et al., 1996). In addition to the racial d iscrepancies in services for marginalized individuals, it was not until the 1950s that national advocacy organizations for gifted learners began to have a place on the educational stage. The National Association for Gifted Children (NAGC) was established in 1954. S hortly after, in 1958, the National Defense Education Act was passed, the first of federal

PAGE 32

17 policies in relation to gifted education. This enacted policy to provide funding to improve school systems and promote secondary education (Encyclopedia Britannica, 2016). The Equality of Educational Opportunity report (Coleman et al., 1966) brought to light the wide spread achievement gaps between racial groups. This seminal work re e mphasized the vast achievement and opportunity gaps in racial performa nce on standardized tests, that only widened as the grade level increased. The 1970s brought reform s encouraging states to establish a working definition of giftedness and the inclusion of the Office of Gifted and Talented within the US Office of Educati on. In 1983, a pivotal report was submitted to the United States Secretary of Education, T. H. Bell, called A Nation at Risk (National Commission on Excellence in Education, 1983). In this report, the authors call out the lack of high standards for child ren across the United States. They also state that the vast majority of gifted students were not reaching their identified potential. The underperformance and laissez faire attitude found in school systems, if unchecked, would lead to ec onomic disaster ( National Commission on Excellence in Education, 1983). This report led to higher federally mandated stan dards, especially in high schools. It brought much needed reform to school system s , including research based p rograms for studying giftedness. By the turn of the century, much had been done to estab lish definitions for giftedness. H ow ever, programming was a problem as e ach state, district, and even school interpreted programming options in a wide spectrum of services. Another seminal report came in 2 004, with A Nation Deceived: How Schools Hold Back America's Brightest Students. Within this report, the authors conclude that because America is set on age level grades, gifted students will be continually marginalize d because they cannot accelerate at t heir own pace (Colangelo, Assouline, & Gross, 2004).

PAGE 33

18 More Recent Policies More recent policies include the No Child Left Behind Act of 2002 and t he Every Student Succeeds Act of 2015. The No Child Left Behind Act (NCLB) showed that the achievement gap bet ween racial groups continues (Jennings & Rentner, 2006). Although the intention of NCLB was to highlight the progress and gaps in student learning, it became increasingly difficult for schools to fulfil the prescriptive requirements. Gifted education als o suffered, as national efforts and state funding focused on students below proficiency (Beisser, 2008; Jolly & Makel, 2010). In 2012, the Obama administration began granting leniency on NCLB requirements in exchange for plans for state designed plans for equity work and high standards for all (Malin, Bragg, & Hackmann, 2017). With the enactment of Every Student Succeeds Act (ESSA) in 2015, there was greater focus on equity work focused on disadvantaged and high need students (Egalite, Fusarelli, & Fusar elli, 2017). It brought about a call for high standards preparing all students for college and career readiness (Malin, Bragg, & Hackmann, 2017). With ESSA, came funding for the Javits Gifted and Talented Students Education Act, which is the only federal ly funded gifted program (NAGC, 2018). The Javits fund was originally established in 1988, as part of the Elementary and Secondary Education Act, and is dedicated to providing research regarding gifted students (Senate, 1987). With the reauthorization of the Javits Fund, through the ESSA and the current work done through the NAGC, there is great advocacy in the United States currently for progression in serving underrepresented populations, namely Black and Hispanic groups. This advocacy and work now nee ds to become more localized to school systems.

PAGE 34

19 Definitions of Giftedness Now, with increasingly diverse populations, definitions for gifted education have become outdated. Before, definitions of giftedness were based on myths regarding what giftedness i s and upon theory lacking in research (Missett & McCormick, 2014; Sternberg, Jarvin, & Grigorenko, 2011; Worrell, 2014). Having a solid research based practice for gifted referral is essential to a well functioning program, as it provides access for gifte d curriculum (Missett & McCormick, 2014). The debate over what giftedness means is vast and has been going on for decades . There is the IQ based conception of giftedness, with its roots in Louis Terman's (1926) study of giftedness , where Terman studied students with IQs of 140 and above. This study affected the way people viewed giftedness, even persisting into the present (Missett & McCormick, 2014; Subotnik, Kassan, Summers, & Wasser, 1993). This concept of giftedness, based on IQ , has been criticize d as marginalizing racial groups, as well as leaving out creativity and twice exceptionality (Missett & McCormick, 2014; Reis & Renzulli, 2011). In recent years, the National Associ ation for Gifted Children (2016) reported that A common view that gifte d students do not need specialized services contribute to a vast disparity of programs and services across states and often within states, leaving many high ability students without the supports they need to achieve at high levels, which is a disser vic e to them and to the nation . ( p. 1) In an interview with M. Rene Islas, the Executive Director of NAGC and S. Dulong Langley, a NAGC board member getting her doctorate in underrepresented gifted populations, Langley stated, "Without federal mandates, state s vary by how they define giftedness, their criteria for identifying students who would benefit from gifted services, and their program services

PAGE 35

20 available" (Langley, 2015 , Interview ). Another research report conducted by Mccain and Pfieffer (2012) reporte d that even though there are half the states reporting specialized identification for underrepresented populations, "these procedures are not identified" (p. 78). Without identification procedures being explicit, there is too much room for interpretation of the state laws and how each location identifies students. This, in turn, leads to problems with ensuring best practices when screening for gifted referral ( Missett & McCormick, 2014) . One persistent cut off score that is constent to gifted identifica tion is the 95 th percentile (Mccain & Pfeiffer, 2012). This is systemically used as the qualifying percentile for many states. This 95 th percentile holds to the IQ based ideology , discussed previously, and continues to be congruent with the most research based theory around giftedness (Lohman & Korb, 2006; Sternberg et al., 2011). In the last thirty years, theories of giftedness have expanded from an IQ only based pedagogy to include: task commitment, motivation, creativity, multiple intelligences, talen t development vs. natural ability, and practical intelligence (Ford & Thomas, 1997; Mccain & Pfeiffer, 2012; Reis & Renzulli, 2011; Stern berg et al., 2011; Subotnik et al., 1993). Viewing giftedness not just as a s et ability level, but also as pote ntial for achievement, is important as issues of underrepresentation are addressed . This, however, requires more work and tim e to create protocols in identification of and programming for Culturally and Linguistically Diverse (CLD) gifted learners. Throu gh Colorado Department of Education (CDE), the members of the Exceptional Student Services Unit (2016) state that, according to compliance with the Exceptional Children's Educational Act, identification for gifted ability is only one of the ways a child ca n be identified. They can also be identified through aptitude, which they define as, "exceptional capability or potential in an y academic content areas" (p. 99). This aptitude identification can happen with either demonstration of, "advanced level on

PAGE 36

21 per formance assessments or 95 th percentile and above on standardized achievement tests" (Exceptional Student Services Unit, 2016, p. 100). There is room for alternate CLD student identification, through aptitude vs . ability identification; however, the metho dology used is unclear. The issue is not whether alternative assessments or identification processes for underrepresented populations should be used ; that much is clear (Naglieri & Ford, 2005). What is not clear is what those processes should look like a nd how to embed them in current practices. Culturally and Linguistically Diverse Gifted Education Within the literature, the issue surrounding underrepresentation of gifted minority groups, namely Black, Hispanic/Latino, and American Indian, is well docume nted (Plucker, Burroughs, & Song, 2010; Worrell, 2014). For the context of this study, gifted underrepresentation is defined as the disproportionality of ethnically diverse students (Black, Hispanic/Latino, and American Indian ) in gifted education when co mpared to the same population in relation to the whole (National Association for Gifted Children, 2008; Worrell, 2014). In other words, the gifted population should similarly match the demographics of the whole student body. Race is not the only factor when speaking of cultural and linguistic diversity in gifted education. There is a great deal of overlap with race and class standing, as evidenced by the data in Table 5 on the likelihood of being poor in America (Mantosis, 2013). Table 5 Chances of Bein g Poor in America Chance of being poor Parents in home Race 1:10 2 White 1:5 1, female White 1:5 2 Hispanic 1:3 1, female Hispanic 1:4 2 Black 1:3 1, female Black

PAGE 37

22 Not only is there marginalization of racial groups, but this marginalization makes it difficult for working class families to change their status. Mantosis (2013) explains that, "Class standing has a significant impact on chances for educational achievement" (p. 155). Most school systems use subjective referral systems to identify gifted students, which systems work from middle class values and norms (Felder, Taradash, Antoine, Ricci, Stemple, & Byamugisha, 2015). There is not a lot of consideration to account for poverty. Many working class families face dilemmas, such as (a) more hour s at work, (b) latchkey children, and (c) increased health problems due to work stress and pressures leading to increased health car costs (Collins & Yeskel, 2013). These problems lead to less time and energy to help students with homework, to explain and problem solve, and to be involved in a child's education. Whereas, upper middle class and upper class families have the resources available to hire tutors, attend parent nights and school nights, and generally appear more involved in their child's educat ion. Working families can also have the added barrier of navigating two languages: their native language and English. This barrier creates difficulty in advocating for their child, making connections with school employees who only speak English, and navi gating the school system. With the added stress of trying to navigate a new language system as well as a school system that is foreign, English Language Acquisition (ELA) families can struggle to get the appropriate education for their child. This is not because they don't want the education or lack of advocacy , but because a system is scant on resources also does not speak their native language (Hos, 2016). In addition to economic barriers, there can be deficit mindset in school staff . "Deficit thinkin g (e.g., stereotypes, biases, low and negative expectations) compromises teacher referrals, as well as how nomination forms and checklists are completed, the test and instruments selected, the specific cut off score selected, and the ultimate placement cri teria and decision" (Ford,

PAGE 38

23 2013 b , p. 10 ). As educators begin to examine personal biases and become more asset focused, they will become better at referring students for special education services. They will appreciate the nuances of gifted tendencies mor e and teachers, "so many of whom are wedded to stereotypical and traditional notions of gifted grounded in high test scores and good behavior," (Ford, 2013 b , p. 8 ) will become more aware of differing characteristics of diverse giftedness. As teachers are trained and become more familiar with gifted characteristics of students that are (a) CLD, (b) disabled, and (c) from low socioeconomic status (SES), they will be better prepared to engage in more equitable referral systems (Felder, et al., 2015). Gifted Identification with a Culturally and Linguistically Diverse Lens The battle for equitable representation in gifted education has been a long one. Frasier (1991) states that a major challenge to the underrepresentation within gifted populations is the lac k of identification of diverse populations. She goes on to state that referral only based models preclude students from diverse populations. Frasier advocates for data from multiple sources in order to identify gifted minoritized students. This social dilemma of underrepresentation in gifted programming is rooted in the systems of power and privilege discussed earlier. In an interview conducted with Dr. Mary M. Frasier, founder of the Torrance Center for Creative Studies, Dr. Frasier states, You are de aling with a very sensitive social problem [underrepresentation of minority students in gifted education]. There is no one who would tell you, ÔWell, the reason that these kids aren't in the program is because I am prejudiced. I discriminate. I am biase d in my opinion about giftedness'Épeople just won't say that . ( Grantham, 2002, p. 51) Whether intentional or unintentional, biases and systemic views of minoritized groups have excluded and continue to exclude access for many talented CLD students. For t he current study,

PAGE 39

24 the researcher is interested in looking at the systemic processes in place for equitable access to gifted identification and thus, access to programming. The next sections outline how the State Department of Education and District are st riving to minimize excellence gaps for inclusive excellence. Colorado Department of Education As part of Colorado Department of Education (C DE), the Office of Gifted Education developed a Gifted Identification manual to be used throughout the state to guid e administrative units at district and school levels. In this manual, it outlines ( see Figure 3) that there should be varied methods of referral, including universal screening measures (Medina, 2016). Figure 3. Referral processes as outlined by Colora do Department of Education. Through the use of multipl e entries for gifted referral, C DE is ensuring more inclusive pathways to gifted referral and identification, leading to more inclusion in gifted programming.

PAGE 40

25 District Within District, universal screening processes have been established at varying grade levels, including second grade (Office of Advanced Academics and Gifted Services, 2011). In second grade, the universal screener is the CogAT 7, a nationally norm referenced cognitive test. From this test, students scoring at the 90 th percentile or above are screened for further evaluation (Office of Advanced Academic and Gifted Se rvices, 2018). The researcher wa s interested in determining if screening at the 90 th percentile versus the 95 th perce ntile overcomes some of the biases discussed earlier, namely the achievement gap, opportunity gap, and the excellence gap. Thus, yielding a more diverse gifted screened population , better matching District student population. Cognitive Abilities Test, Cog AT The debate over equitable practices for identifying CLD gifted learners is long. Controversy exists between two major writers of cognitive tests; Lohman, author of CogAT, and Naglieri, author of NNAT. Lohman and Naglieri debate over best practices for identifying CLD gifted learners. The researcher examines some of the controversy here, but focuses efforts on describing CogAT and it s utilization in context to District , as that is where the study is taking place and District uses the CogAT as a gifted screener in second grade. Lohman (2005) explains that there are two types of giftedness: high accomplishment and high potential. Highly accomplished students are ones that are already performing high on standardized tests or academic tests. High potenti al students are ones that are not yet demonstrating their unique abilities in a particular domain. Lohman (2005) adds that these groups of students possess differing programming needs. He cites that using nonverbal tests for identifying English Language Learners for gifted programming should use more than "figural

PAGE 41

26 reasoning abilities " (Lohman & Gambrell, 2012, p. 41). He advocates for verbal and quantitative reasoning batteries over a nonverba l only test (Lohman & Gambrell, 2012). His argument is that t here is unequivocal data demonstrating that t he " predictors of achievement in reading, mathematics, social studies, and science are the same for White, Black, Hispanic, and Asian American students" (Lohman, 2005, p. 344). This is vastly different from the findings of Naglieri and Ford (2005, 2015). Naglieri and Ford (2015) have long written about the discrepancies in gifted representation of minority populations, especially for Black and Hispanic students. Data from the Office for Civil Rights is clear in the underrepresentation of Black and Hispanic students in gifted education. Naglieri and Ford (2015) cite data from the Office for Civil Rights stating that, "Black students are the most underrepresented racial group, followed by Hispanic students" (p. 2 34). They state that, "On a daily basis, educators struggle with finding the most effective ways to both identify and serve gifted students who are not reaching their potential, as measured by tests or as perceived by teachers, counselors, or parents" (Na glieri & Ford, 2005, p. 34). Others have also reported regarding these gifted discrepancies of ethnically diverse students (Feldman, 2003; Fraiser, Hunsaker, Lee, Finley, Garcia, & Martin, 1995; McBee, 2006). Lohman, Korb, and Lakin (2008) counter Naglier i and Ford (2005) when they conducted a study comparing cognitive measures including the Raven, NNAT, and CogAT. Their assumptions include that, "one cannot assume that nonverbal tests level the playing field for children who come from different cultures or who have had different educational opportunities" (p. 293). The authors state that nonverbal tests help to provide useful information, but caution against use of nationally normed data that are normed for different populations, saying that this

PAGE 42

27 can be misleading. Although, later Lohman and Gambrell (2012) advocate use of local norms for increasing gifted referral from ethnically diverse populations. As part of the ever developing and fine tuning process, the CogAT test has been adapted recently to bett er meet the needs of ELLs. At the annual 2013 NAGC conference, Jim Nicholson, President of HMH Riverside, announced changes to the structuring of CogAT, including less emphasis on the language components that can hinder second language learners. Nicholso n stated, "The reduced language emphasis on CogAT increases access to enrichment programs for students from a variety of backgrounds" (Business Wire, 2013, p. 1). The researcher could not find published studies regarding the effects of this change on stud ent populations being referred for gifted evaluation. There is great controversy in the gifted field around what test and data should be used for CLD learners ( Lakin & Lohman, 2011; Naglieri & Ford, 2005). What can be agreed on is that no one test or on e method is going to guarantee that all students needing gifted referral will receive the referral (Greenfield, 1997; Helms, 1992; Naglieri & Ford, 2005). Given this, there are ways to improve referral practices to make selection more equitable. The rese archer also agrees that, "Regardless of their linguistic and cultural background, any student who demonstrates a need for more demanding curricula should be challenged" (Naglieri & Ford, 2005, p. 34). The question is, how do systems effectively screen for the most equitable outcome? This study was developed to evaluate current practices for racial equitability and to determine if equitable numbers of students can be referred for gifted evaluation at the 90 th percentile compared to the 95 th percentile .

PAGE 43

28 Sum mary This chapter reviewed current literature regarding the four major themes of this study including: (a) definitions of giftedness, (b) culturally and linguistically diverse (CLD) gifted education, (c) gifted identification with a CLD lens, and (d) Cogni tive Abilities Test (CogAT). It framed this work in literature and historical context of systems of power and privilege. This study focused on determining if equitable representation for gifted referral can be reached if the percentile cut off at the 95 t h percentile is dropped to the 90 th percentile . In this way, the researcher is looking to affirm current District practices or call for a change in policy. The battle for equitable gifted practices continues on the national, as well as local level . Str iving to find most equitable practices for gifted referral in District can help other districts in their efforts to do the same.

PAGE 44

2 9 CHAPTER III METHODOLOGY This chapter describes and outlines the methodology used in this study. The conceptual framew ork is reviewed from Chapter 1 . The variables a re operationalized. The method discussed include s a non experimental, comparative study. Research design elements, including: site selection, subjects, and variables are described and operationalized. Afte r that, data collection processes are ou tlined. Data analyses, including descriptive statistics, relative frequency, t tests, ANOVA, MANOVA, and Chi Square are used for this study. Finally, CogAT 7, the instrument being researched, is described with curr ent reliability and validity. This is a non experimental, comparative study; examining the difference between racial group CogAT 7 scores. It is also a relational study; examining the relationship between racial group test scores and gifted r eferral. Ult imately, the researcher examine d the relationship between racial designation and gifted referral at the 90 th percentile and 95 th percentile , through the use of the CogAT 7 as a screener. Since this study is not manipulating variables and its p urpose is to compare variables (race and CogAT 7) to test the hypothesis, a non experimental, comparative study is appropriate (Johnson & Christensen, 2012). The focus of this study also lends itself to descriptive and relationship, which is part of the n on experimental design, as outlined by Johnson and Christensen (2012). The researcher's purpose is to check for equitable referral representation across racial groups. Research Design Design elements of this study include site selection, subjects, a nd variables. The site selection was from a large Denver Metro school district. Subjects were selected from the current

PAGE 45

30 screening processes of the Office of Advanced Academic and Gifted Services in District. Finally, variables are operationalized in ter ms of this study. Site Selection Site selection for this study was convenient, as the researcher worked in District as an Advanced Academic Specialist (formerly known as Gifted and Talented). Since the purpose of this study is to examine current prac tices for Gifted referral, the researcher desired to examine processes supporting referral of diverse populations by examining current practices. Subjects The target population is 2 nd graders in District, because they take the CogAT 7 screener. The sampli ng approach is convenient, because it is within the school district where the researcher is currently employed. The data is disaggregated by race. All CogAT 7 scores, within the 2 nd grade District population, will be pulled from Powerschool and used for testing the statistical significance of racial distribution. No other identifying information was gathered to protect the anonymity of students. Variables The variables compared and used in this study include d : CogAT 7, categorical racial groups , and gifted referral . The CogAT 7 contains the three batteries: Verbal, Quantitative, and NonVerbal. The categorical data came from the District data, which is Powerschool. Finally, the gifted referral data was from the referrals at the 90 th and 95 th perce ntiles. These were measured using the procedures as described in Table 6.

PAGE 46

31 Table 6 Operationalizing Variables Variable How it will be measured CogAT 7 independent continuous CogAT 7 has three batteries: (a) Verbal, (b) Quantitative, and (c) NonVer bal. Each of the batteries will be used separately in the screening process. A student scoring 95 th percentile or higher on any of the batteries will be referred for gifted evaluation and further testing. Racial Groups independent categorical Using P owerschool, each racial group will be categorized and grouped together for statistical analysis. The CogAT 7 will be disaggregated into racial groups for purposes of determining the percentage of students referred for gifted evaluation. These results wil l then be compared for statistical significance of scores. Gifted Referral dependent categorical A student scoring 95 th percentile or higher on any of the batteries will be referred for gifted evaluation and further testing. In this study, the 90 th perce ntile cutoff will also be explored for gifted referral. Conceptual Framework District uses CogAT 7 as a cognitive measur e of student ability. The researcher is interested in determining whether race affects gifted referral in District , based on the use of CogAT 7. Figure 4 visually represents the conceptual framework of this study. Figure 4. Conceptual framework .

PAGE 47

32 Cognitive Abilities Test CogAT 7, "measures students' learned reasoning abilities in the three areas most linked to academic succes s in school: Verbal, Quantitative and Nonverbal" (Houghton Mifflin Harcourt, 2017 , p.7). It is currently used as a universal screener, at the end of 2 nd grade, for gifted referral. A universal screener is a test given to a whole grade level, across a dis trict, to identify students that n eed to be referred (C DE: Office of Gifted Education, 2016). If a child performs at the 95 th percentile or above, on any given battery, they are referred for additional testing (District, 2017). Racial Groups Ethnic and racial categorical data is taken from PowerSchool, a District student database, and can be sorted accordingly. Racial populations including: White, Black, Hispanic, Asian, American Indian , and Multi Racial will be included in this study. Gifted Referral Within D istrict , the first step in identifying a student for giftedness is a referral and evaluation ( District , 2017). This referral is taken from various sources, one being CogAT 7. Other sources for gifted referral include: parents, teachers, or othe r professionals working with the student i . e. , doctors, psychologists, etc . ( District , 2017). Percentiles Given national norms, the CogAT 7 uses age based national norms and grade based national norms (Houghton Mifflin Harcourt, 2017 ) . District uses the age based national norms at the 95 th percentile and above for referral and identification (District, 2017). This evaluation at the 95 th percentile , using national norms, and above meets the C DE guidelines for gifted

PAGE 48

33 e valuation (C DE, 2016). This study compared screening at the 9 0 th percentile or 95 th percentile or above is more equitable for minoritized populations . Research Questions The research questions explored in this study include: 1. Does race affect test scores from the Cognitive Abilities Test, CogAT 7? H 0 = Race does not affect test scores from the Cognitive Abilities Test. H 1 = Race does affect test scores from the Cognitive Abilities Test. 2. What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups? H 0 = There is no difference in gifted referral between racial groups at the 95 th percentile and above. H 1 = There is a difference in gifted referral between racial groups at the 95 th percentile and above. 3. What is the difference i n gifted referral, based on CogAT 7, at the 90 th percentile and above, between racial groups? H 0 = There is no difference in gifted referral between racial groups at the 90 th percentile and above. H 1 = There is a difference in gifted referral between racia l groups at the 90 th percentile and above. 4. What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th p ercentile? H 0 = There is no significant difference in gifted referra l between racial groups when using the 90 th p ercentile compared to the 95 th percentile.

PAGE 49

34 H 1 = There is a significant difference in gifted referral be tween racial groups when using the 90 th percentile compared to the 95 th percentile. Instrument CogAT 7 T he CogAT 7 was chosen as the study instrument to be analyzed, using district demographic distributions, because it is used in District as a universal screener for gifted referral. The following describes CogAT 7 sampling and norming processes, along with reliability and validity. CogAT 7 has a substantial standardization sample of 65,350 K 12 students, normed w ith achievement tests such as Iowa Tests of Basic Skills and Iowa T ests of Educational Development . Reliability for the sample was high, with split half correlations between .80 and .92 for Verbal, Quantitative, and Nonverbal batteries. Global scores had split half reliability values between .91 and .97 (Lohman, 2012; Warne, 2015). The validity of CogAT 7 was tested through concurrent validity studi es with the Naglieri Nonverba l Ability Test (NNAT 2) and Iowa assessments (Lohman, 2012). The correlation with the NNAT was done in a study sample of 149 second grade students, with all three CogAT batteries correlating (r= .51) with the NNAT. The correl ation scores with the Iowa assessments yielded the following ranges for differing grade levels (r = .42 .83 for Verbal, r = .30 .79 for Quantitative, r = .32 .69 for Nonverbal, and r = .40 .85 for the total CogAT 7 battery (Lohman, 2012; Warne, 2015). Wit h the large standardization sample, high correlations, and concurrent validity studies, the CogAT 7 is a reliable and valid measure. Data Collection and Storage Sampling was of all second gra ders taking the CogAT 7 ; thus, resulting in a reliable study, as it was representative of the target population. A sample size of approximately 9 ,000

PAGE 50

35 ( some 3,000 for each of three years) helped reduce the margin of error. Students that are opted out of district assessments were excluded from this study. Also, student s who recently moved in to District and who ha ve not taken the CogAT 7 were exclud ed from the study. Validity was ensured through use of CogAT 7, a valid measure of cognitive ability. Through the use of PowerSchool, students were grouped using the family selected racial designation, providing valid racial groupings. Data for this study, including three years of CogAT data (2015 2016, 2016 2017, 2017 2018) was collected from the Office of Assessment and Evaluation within District. The data had the f ollowing identifiers: (a) racial designation, (b) Verbal battery score, (c) Quantitative battery score, and (d) Nonverb al battery score. The data was grouped by school, but no other identifying infor mation was given to the researcher. The data was stored electronically on a District secured computer, ensuring the securit y of the data. When the data was downloaded, it will be on a District secured computer. The data went through Exc el and SPSS for data processing. These E xcel and SPSS files were stored on the District secured computer. The data was shared in raw form only with District members assigned to this study or for University of Colorado staff assigned to this study as a committee member or in the role of methodology coach. Data Analysis Three y ears of data ( school years: 2015 2016, 2016 2017, and 2017 2018) was collected. This provided a representative sample set of over 9,000 students (approximately 3,000 students per year), ensuring reliability of findings. Prior to any analysis testing of t he data, descriptive statistics and t tests will be run, checking for variance of the samples . The t test data will be compared to ensure similar sample distributions prior to the ANOVA testing. Relative frequency will also be given of gifted referrals b y race .

PAGE 51

36 For research questions: ( a ) Does race affect test scores from the Cognitive Abilities Test, CogAT 7, ( b ) What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups, and ( c ) What is the diff erence in gifted referral, based on CogAT 7, at the 90 th percentile and above, between racial groups , an ANOVA test will be conducted. Confidence level is set at 95%. Descriptive statistics will first be conducted focusing on CogAT 7 scores in each batte ry (Verbal, Quantitative, and Nonverbal) , across the fol lowing races: American Indian , Black, Hispanic, Asian, White, and Multi Racial . Statistical assumptions of CogAT 7 scores across races will be tested for each battery. The results of each battery, including demographic distribution, will be reported for analysis . When answe ring research question 4 (I s there a difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile? ) a factorial ANOV A test will be conducted to compare main effects of race upon results at the 90 th percentile and 95 th percentile . Limitations and Delimitations Since this study is done within a school district, it is limited to that school district. This study can b e replicated through the use of similar methodology, but is not intended to be generalized to a larger population. The limitations are defined as restrictions beyond the researcher's control, while the delimitations are limits the research has placed on t he study (Baltimore County Public Schools, 2017). Limitations to this study include: (a) this study is not intended to be generalizable to a greater population, as it was done within District and (b) this study is not seeking to determine causation, only relation. The following are delimitations of the study: (a) data is narrowed to 2nd graders administered the CogAT 7 during the 2015 2016, 2016 2017, and 2017 2018 school years, (b) students with invalid scores in one or more batteries

PAGE 52

37 will be eliminated from the data set, and (c) students whose parents opted out of testing will not be included in the data set. Ethical Considerations Ethical considerations by the researcher include the following responsibilities: (a) to the participants, (b) sponsors, including the University of Colorado Denver and District, (c) the field of research, and (d) the research community (Govil, 2013). In the following section s, the researcher relates the plan for ethical considerations, as they pertain to this study. Parti cipants In regards to the participants, the researcher will work ethically and with integrity with the data. The data is reported as de identified, so the researcher will not have access to the student name. It will be disaggregated by race, as reported by the family to PowerSchool, the database in District. Besides the individual battery scores and the racial designation of the student, no other identifying information will be given to the researcher . Informed consent is not required for this study, a s it is de identified and the sample size is large. There is no impact on workload of participants, as it is extant data. Privacy for participants is ensured through the de identifying nature of the data. No psychological harm will be caused to particip ants, since the data is de identified and the sample is large. Sponsors In order to best ensure ethical considerations for sponsors, the researcher is operationalizing all key terms in use during the study. For both University of Colorado Denver an d District, the researcher is following proper protocol in obtaining permissions and signatures before conducting any research. The findings from this study are not intended to be prescriptive for District, rather this study is intended to give options fo r further decision making. The

PAGE 53

38 findings will be shared from data interpretations that have justifiable rational. Also, limitations for the study are included in the proposal to avoid ethical confusion. The nature of words used in the study will maintain professionalism and will seek to establish best practices within the district. Field of Research This study seeks to add to the current body of research on culturally and linguistically diverse gifted learners. The underrepresentation of gifted mino rities is clear in the literature. What is not clear, however, is how to go about screening and referring our CLD learners for gifted evaluation. This study seeks to build on other researchers' work and add to the field by providing one district's method s for equitably working to refer and identify CLD gifted learners . Research Community For the research community, the researcher will provide a technical research report, which will include methodology and findings. In order to maintain the integrity of educational research, highest standards will be used when reporting and conducting research. And finally, the researcher will use the highest integrity to gather, report on , and disseminate the findings of the research.

PAGE 54

39 CHAPTER IV RESULTS This chapt er reports the d ata analyses used for this study , including descriptive statistics, relative frequency, t tests, ANOVA, and MANOVA testing . After conducting the MANOVA testing, analysis for goodness of fit through use of Chi s quare testing was conducted t o determine where the significance was situated . This testing was done to compare the number of gifted referrals of each race to the number that statistically should be referred at the 90 th and 95 th percentiles. The Chi square testing was also used to co mpare the number of gifted referrals to the demogr aphic make up of the district. In the end, Chi square testing is discussed as situated in the 90 th and 95 th percentiles. The researcher obtained permissions from the district to include three years of data in this study. The following identifiers were used for this study: CogAT ID, administration year, student demographic, Verbal battery percentile, Q uantitative battery percentile, and N onverbal battery percentile. Student demographic designations were ta ken from PowerSchool, where parents and families determine demographic identity. It should also be noted that the designation labels for demographic groups are taken from the labels used in PowerSchool. The data was checked for missing data. If a studen t had a score for any of the three batteries, it was included in the data set . Missing data included any student that was opted out for testing or had invalid scores for all three batteries. Descriptive statistics and normality testing is presented prior to describing the findings to the research questions. Each research question is presented followed by test results. The results are described and interpreted for the context of this study. The research questions discus s ed are:

PAGE 55

40 1. ! Does race affect test sc ores from the Cognitive Abilities Test, CogAT 7? 2. ! What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups? 3. ! What is the difference in gifted referral, based on CogAT 7, at the 90 th percentile and a bove, between racial groups? 4. ! What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile? Prior to disseminating results, it is important to remind the reader how g ifted referral is obtained in District. A cognitive test, CogAT7, is used as a universal screener, given to all second grade students in the district. The full battery version is used with Verbal, Quantitative, and Nonverbal components. A score of 90 th percentile or above, using national age norms, refers a student for further academic testing. A screening at 90 th percentile is intended to cast a wider net for gifted potential. Students still are required to meet the 95 th percentile mark for Colorado s tate gifted identification. The results shared hereafter are from CogAT, used as an entry point to gifted identification. Descriptive Statistics In this section, descriptive statistics are presented, including valid and missing cases. Procedures for che cking valid and missing data are described. Then, normality is tested using skewness and kurtosis. Central tendencies for battery and race are described to check for outlying data and participation by race. Finally, distributions across administration y ears and batteries are explored to check for reliability across testing years. In the end, a determination of validity for data usage is given for this study.

PAGE 56

41 Valid and Missing Cases Table 7 gives the overall data set statistics, including valid and mi ssing cases for each of the batteries (Verbal, Quantitative, and Nonverbal). It also lists the percentage of valid and missing cases. The sample population from 2016, 2017, 2018 data began with 11,648 students taking the CogAT 7 test. The valid scores f or each battery are listed along with cases missing. Table 7 Observations by Battery 2016, 2017, and 2018 Combined Valid Cases Missing Battery N Percent N Percent Total Verbal 10972 94.2 676 5.8 11648 Quantitative 10659 91.5 989 8.5 11648 No nverbal 11312 97.1 336 2.9 11648 Cases missing include scores from not enough test questions answered for a percentile rank. Any valid battery score from the 11,648 students will be used for gifted referral determination. Only students with three inval id battery scores will be excluded from the data. Since all student samples have at least one valid score, all 11,648 student scores will be included in this study. When analyzing by battery, only valid test scores from that battery will be evaluated. T he missing cases will be excluded from the data set. Testing for Normalcy Based on normal distribution having skewness of ± 1 and kurtosis of ± 3, all CogAT batteries (Verbal, Quantitative, and Nonverbal) are within the threshold of normal distribution . Skewness and kurtosis of each CogAT battery are given in Table 8. Since the data is within the threshold , the data from each testin g year is valid for this study.

PAGE 57

42 Table 8 Testing for Normalcy by Battery 2016, 2017, and 2018 Combined Battery N Mean SD Skewness Kurtosis Normal Distribution Verbal 10972 57.18 26.277 . 324 . 922 Yes Quantitative 10659 67.13 23.444 . 749 . 273 Yes Nonverbal 11312 63.31 24.628 . 524 . 679 Yes Central Tendenc y b y Race and Battery Before reporting out regarding data from research questions, the researcher conducted descriptive statistics and t tests to check for variance for each battery of the CogAT 7; including Verbal, Quantitative, and Nonverbal. In this section, central tendencies are reported by race and indivi dual battery. The researcher is looking for any deviations or large discrepancies between number of participants, mean, and/or standard deviation; as possible variations could impact final findings and skew conclusions drawn. When computing percentile ra nking means, only test scores (1 99 th percentiles) were accepted as valid scores. Any score at zero was excluded from the data set , as this is not a valid score . Table 9 gives the number of valid scores, means, and standard deviations for each of the thr ee batteries organized by race. Consistently, quantitative mean results are higher for each demographic group. Trends between battery means are similar across racial groups. However, Asian and White have consistently higher means across batteries than th eir Black, Hispanic, and American Indian counterparts. This is consistent with findings from Ford (2011) and Miller (2004) . Standard deviations are relatively consistent across racial groups, with the exception of a couple outliers, American Indian Verba l Battery and Asian Quantitative Battery. The American Indian Verbal Battery shows a greater variance of scores when compared to other racial groups. The Asian Quantitative Battery shows a smaller variance of scores when compared to other racial groups.

PAGE 58

43 Table 9 Descriptive Statistics, All CogAT 7 Battery Scores by Race , including 2016, 2017, 2018 D ata Race CogAT Battery n M SD American Indian Verbal Quantitative Nonverbal 60 57 62 51.72 61.58 58.85 29.07 26.19 25.28 Asian Verbal Quantitative Nonverbal 9 42 946 971 64.12 76.81 72.08 24.22 20.67 22.16 Black Verbal Quantitative Nonverbal 1036 991 1088 46.70 55.98 49.11 25.16 24.59 24.02 Hispanic Verbal Quantitative Nonverbal 2114 2036 2215 46.57 58.52 54.86 25.16 23.91 24.92 Multi Racial Verbal Q uantitative Nonverbal 878 843 899 57.70 65.79 62.32 25.90 23.76 24.42 White Verbal Quantitative Nonverbal 5918 5762 6053 61.72 70.78 67.77 25.27 21.73 23.05 Comparisons Across Administration Year s Within Batteries In order to verify the data se t w as reliable across administration years, comparisons were made of all scores within each battery across administration years, checking for variance. First, descriptive statistics, including number of observations, mean, and standard deviation, are repo rted for years 2016, 2017, and 2018. After that, independent sample t tests are reported comparing administration year results within each battery. Verbal battery. Table 10 reports the Verbal battery across administration years (2016, 2017, 2018). Then , results for t tests are given in Table 11 . Independent samples t tests were run to determine if there were differences in Verbal battery between administration years 2016 vs

PAGE 59

44 2017 vs 2018 . For administration years 2016 vs 2017 an independent samples t t est was run. T here were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was homogeneity of variances betw een 2016 vs 2017 , as assessed by Levene's test for equality of variances ( p = .291). For the Verbal battery, there was no statistically significant difference for 2016 ( M = 58.32, SD = 25.88) and 2017 ( M = 57.74 , SD = 26.28 ), where ( M = .584, 95% CI [ .59 7, 1.766], t (7488) = .969, p = .332, d = .02. A Welch t test was run to determine if there were differences in Verbal battery scores between administration years 2016 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot. Homogeneity of variances was violated, as assessed by Levene's test for equality of variances ( p = .0 34 ). There was a significant di fference between 2016 ( M = 58.32, SD = 25.88) and 2018 ( M = 55.35 , SD = 26.61 ), where ( M = 2.97 , 95% CI [ 1.76, 4.18 ], t (7168 ) = 4.809 , p = <.001, g = .11 . An independent samples t test was run to determine if there were differences in Verbal battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Verbal battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot. There was h omogeneity of variances, as assessed by Levene's test for equality of variances ( p = .282 ). There was a sig nificant difference between 2017 ( M = 57.74, SD = 26.28) and 2018 ( M = 55.35, SD = 26.61 ) , where ( M = 2.385 , 95% CI [ 1.162, 3.607 ], t (7199 ) = 3.825 , p = <.001, d = .09 .

PAGE 60

45 Table 10 Comparison of CogAT Scores for Administration Years ( 2016, 2017, 2018 ) , Verbal Battery Administration Year n Mean SD 2016 3771 58.32 25.88 2017 3719 57.74 26.28 2018 3482 55.35 26.61 Table 11 Comparison of Verbal B attery CogAT Scores for Administration Years , T Tests Year s Compared df p Significance 2016 vs 2017 7488 . 332 No 2016 vs 2018 7251 <.001 Yes 2017 vs 2018 7199 < .0 0 1 Yes As the t test s report , there is a statistically significance in two out of three comparison years for the Verbal battery. These results show that this test could have different outcomes, depending on the years gathered. This should be noted for further studies, since findings will depend on years pulled and are conditional on the po pulation of the district. When reviewing the means, the mean for 2018 was significantly lower than the two previous yea rs (2016 and 2017), leading to a significant difference in comparison years. Quantitative battery. Table 12 gives a reporting for the Quantitative battery across administration years (2016, 2017, 2018). Then, results for t tests are given in Table 13 . Independent samples t tests were run to determine if there were differences in Quantitative battery between administration years 2016 v s 2017 vs 2018 . For administration years 2016 vs 2017, t here were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q P lot, and there was homogeneity of variances between 2016 vs 2017, as assessed by

PAGE 61

46 Levene's test for equality of variances ( p = .786 ). For the Quantitative battery, there was no statistically significant difference for 2016 ( M = 66.95 , SD = 23.43 ) and 2017 ( M = 67.13 , SD = 23.64 ), where ( M = .185, 95% CI [ 1.264, .894 ], t ( 7309) = .336 , p = . 737 , d = . 008 . An i ndependent samples t test was run to determine if there were differences in Quantitative battery scores between administration years 2016 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was homogeneity of variances between 201 6 vs 2018, as assessed by Levene's test for equality of variances ( p = .601). There was no significant difference between 2016 ( M = 66.95, SD = 23.43) and 2018 ( M = 67.32 , SD = 23.247 ), where ( M = .373 , 95% CI [ 1.465, .720 ], t ( 7028 ) = .669 , p = .601 , d = .02 . An independent samples t test was run to determine if there were differences in Quantitative battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative batter y scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was homog eneity of variances between 2017 vs 2018, as assessed by Levene's test for equality of variances ( p = .435). There was no sign ificant difference between 2017 ( M = 67.13, SD = 23.64) and 2018 ( M = 67.32, SD = 23.247), where ( M = .188 , 95% CI [ 1.29, .914 ], t ( 6975 ) = .334 , p = .738 , d = .008 . As reported in Table 13, no statistically significant difference was found when comparin g administration years, lending to reliability between administration years. When reviewing the means for each administration year, they were close across years, ranging from 66.95 67.32. Results from this data are going to be consistent in this populati on between administration years.

PAGE 62

47 Table 12 Comparison of CogAT Scores for Administration Years ( 2016, 2017, 2018 ) , Quantitative Battery Administration Year n Mean SD 2016 3682 66.95 23.43 2017 3629 67.13 23.64 2018 3348 67.32 23.25 Table 13 Comparis on of Quantitative Battery CogAT Scores for Administration Years Year s Compared df p Significant 2016 vs 2017 7309 .737 No 2016 vs 2018 7028 .504 No 2017 vs 2018 6975 .738 No Nonverbal battery. Table 14 gives a reporting for the Nonverbal batte ry across administra tion years (2016, 2017, 2018). Then, results for t tests are given in Table 15 . Independent samples t tests were run to determine if there were differences in Nonverbal battery between administration years 2016 vs 2017 vs 2018. For a dministration years 2016 vs 2017, there were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was hom ogeneity of variances between 2016 vs 2017, as assessed by Levene's test for equality of variances ( p = .697 ). For the Nonverbal battery, there was a statistically significant difference for 2016 ( M = 62.34 , SD = 24.765 ) and 2017 ( M = 63.8 , SD = 24.72 ), whe re ( M = 1.46 , 95% CI [ 2.568, .352 ], t ( 7668 ) = 2.584 , p = . 01 , d = .06 . An independent samples t test was run to determine if there were differences in Nonverbal battery scores between administration years 2016 vs 2018. There were no outliers in the da ta, as assessed by inspection of a boxplot. Quantitative battery scores for each

PAGE 63

48 administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was homogeneity of variances between 2016 vs 2018, as assessed by Levene' s test for equality of variances ( p = .219 ). There was a sig nificant difference between 2016 ( M = 62.34, SD = 24.765 ) and 2018 ( M = 63.86 , SD = 24.361 ), where ( M = 1.521 , 95% CI [ 2.633, .410 ], t ( 7514 ) = 2.683 , p = .007 , d = .06 . An independent sample s t test was run to determine if there were differences in Nonverbal battery scores between administration years 2017 vs 2018. There were no outliers in the data, as assessed by inspection of a boxplot. Quantitative battery scores for each administration year were normally distributed with fat tails, as assessed by the Q Q Plot, and there was homogeneity of variances between 2017 vs 2018, as assessed by Levene's test for equality of variances ( p = .405). There was no sig nificant difference between 2017 ( M =63.8, SD =24.72) and 2018 ( M = 63.86, SD = 24.361), where ( M = .061 , 95% CI [ 1.177, 1.055 ], t (7436 ) = .108, p = .914 , d = . 002 . Table 14 Comparison of CogAT Scores for Administration Years ( 2016, 2017, 2018 ) , Nonverbal Battery Administration Year N Mean SD 2016 3874 62.34 24.76 2017 3796 63.80 24.72 2018 3642 63.86 24.36 Table 15 Comparison of Nonverbal Battery CogAT Scores for Administration Years Year s Compared df p Sig nificance 2016 vs 2017 7668 .010 Yes 2016 vs 2018 7514 .007 Yes 201 7 vs 2018 7436 .914 No

PAGE 64

49 The data shows statistically significant differences in two out of three comparisons. When reviewing the means between administration years, 2016 had a significantly lower mean score than 2017 and 2018, leading to the significant difference found in the t test s . Again, as with the V erbal battery, results for this test are going to vary from year to year. This variance should be taken into account as others look to replicate this study. Summary The descriptive statistics show tha t Verbal and Nonverbal varied within this district between administration years. Quantitative stayed consistent and no significant difference was found between administration years. This study was not intended to generalize across districts or states, ra ther give a reporting of one school district. Since the observation sample for this study is large, we will move forward with the research questions . Due to the large difference of representative demographic groups, ranging from 60 to over 6,000, a weigh ting adjustment was factored prior to performing any of the research question testing. Although it is a large sample size, it is truly inclusive of all students within the district. Since the scope of this study does not extend beyond this district, the weighting adjustment is sufficient. R esearch Q uestion 1: Does race affect test scores from the Cognitive Abilities Test, CogAT 7 ? In addressing this question, the researcher separated each battery by racial group , as designated by PowerSchool . Each ba ttery (Verbal, Quantitative, and Nonverbal) was tested separately. ANOVA testing was used, with a weighted adjustment, due to the population discrepancy between racial groups. The following sections explain the results found for each battery.

PAGE 65

50 Verbal Bat tery A one way Welch ANOVA was conducted to determine if CogAT Verbal battery scores were different between racial groups . Participants were classified into demographic groups, including: American Indian (n = 60 ), Asian (n = 942), Black (n = 1036 ), Hispan ic (n = 2114 ) , Multi Racial (n = 878 ) , and White (n = 5918 ). There were no outliers , as assessed by inspection of boxplots. T h e data was normally distributed with fat tails, as assessed by the Q Q Plot . Homogeneity of variances was violated, as assessed by Levene's Test of Homogeneity of Variance (p = <.001 ). Verbal test scores were statistically significantly di fferent between racial groups, Welch's F (5, 589.49 ) = 160.69 , p < .001. Means and standard deviatio n s are provided in Table 16 . The group mea ns for Verbal battery were statistically significantly different (p<.001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does affect Verbal battery test scores. Table 16 Population Descriptive, Verbal Ba ttery, Used for ANOVA Testing Race N Mean SD American Indian 60 51.72 29.07 Asian 942 64.12 24.2 2 Black 1036 46.70 25. 16 Hispanic 2114 46.57 25. 81 Multi 878 57.70 25.89 White 5918 61.72 25. 28 Games Howell post hoc analysis (Table 17) revealed th at statistically significant differences existed between each racial comparison, except between Black*Hispanic and White*Asian. No significant difference between American Indian and other racial groups, excluding Asian, was found.

PAGE 66

51 Table 17 Games Howell Post Hoc Comparing Verbal CogAT Scores A cross Race Racial Comparison p value Significant Different American Indian*Asian .02 3 * Yes American Indian*Black .7 80 No American Indian*Hispanic .7 52 No American Indian*Multi .6 31 No American Indian*White . 100 No Asian*Black < .001 ** Yes Asian*Hispanic < .001 ** Yes Asian*Multi < .001 ** Yes Asian*White .056 No Black*Hispanic 1.000 No Black*Multi < .001 ** Yes Black*White < .001 ** Yes Hispanic*Multi < .001 ** Yes Hispanic*White < .001 ** Yes Multi*White < .001 ** Yes * : p value <0.05 **: p value <0.01 Quantitative B attery A one way Welch ANOVA was conducted to determine if CogAT Quantitative battery scores were different between racial groups. Participants were classified into demogr aphic groups, inclu ding: American Indian (n = 57 ), Asian ( n = 946 ), Black (n = 991 ), Hispanic (n = 2036 ), Multi Racial (n = 843 ), and White (n = 5762 ). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as asse ssed by the Q Q Plot . Homogeneity of variances was violated, as assessed by Levene's Test of Homogeneity of Variance (p = <.001 ). Quantitative test scores were statistically significantly different between racial groups, Welch's F (5, 560.27 ) = 165.887 , p < .001. Means and standard deviations are provided in Table 18 .

PAGE 67

52 Table 18 Population Descriptive , Quantitative Battery, Used for ANOVA Testing Race N Mean SD American Indian 57 61.58 26.19 Asian 946 76.81 20. 67 Black 991 55.98 24.5 9 Hispanic 2036 58.52 23.9 1 Multi 843 65.79 23.7 6 White 5762 70.78 21.73 The group means for Quantitative battery were statistically significantly different (p<.001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race doe s affect Quantitative battery test scores. Table 19 lists the post hoc testing results. Table 19 Games Howell Post Hoc Comparing Quantitative CogAT Scores across Race Racial Comparison p value Significant Difference American Indian *Asian .001 ** Yes Ame rican Indian *Black .619 No American Indian *Hispanic .95 2 No American Indian *Multi .8 44 No American Indian *White .10 3 No Asian*Black < .001** Yes Asian*Hispanic < .001** Yes Asian*Multi < .001** Yes Asian*White < .001** Yes Black*Hispanic .079 No B lack*Multi < .001** Yes Black*White < .001** Yes Hispanic*Multi < .001** Yes Hispanic*White < .001** Yes Multi*White < .001** Yes *: p value <0.05 **: p value <0.01

PAGE 68

53 Games Howell post hoc analysis (Table 19) revealed that statistically significant diff erences existed between all racial group comparisons, except between Black*Hispanic. Also, statistically significant difference was not found when comparing American Indian to other racial groups, excluding Asian. Nonverbal Battery A one way Welch ANOVA was conducted to determine if CogAT Nonverbal battery scores were different between racial groups. Participants were classified into demographic groups, including: American Indian (n = 57), Asian (n = 946 ), Black (n = 991 ), Hispanic (n = 2036 ), Multi Rac ial (n = 843 ), and White (n = 5762 ). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q Q Plot . Homogeneity of variances was violated, as assessed by Levene's Test of Hom ogeneity of Variance (p = <.001). Nonverbal test scores were statistically significantly different between racial groups, Welch's F (5, 609.784 ) = 201.603 , p < .001. Means and standard deviations are provided in Table 20 . The group means for Nonverbal ba ttery were statistically significantly different (p<.001) and, therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does affect Nonverbal battery test scores. Table 20 Population Descriptive , Nonverbal Battery, Use d for ANOVA Testing Race N Mean SD American Indian 57 58.85 25.277 Asian 946 72.08 22.164 Black 991 49.11 24.024 Hispanic 2036 54.86 24.92 5 Multi 843 62.32 24.4 22 White 5762 67.77 23.04 8

PAGE 69

54 Games Howell post hoc analysis (Table 21) revealed that s tatistically significant differences existed between all racial group comparisons; however, statistically significant difference was not found when comparing American Indian to other racial groups, excluding Asian and Black. Table 21 Games Howell Post Ho c Comparing Nonverbal CogAT Scores across Race Racial Comparison p value Significant Difference American Indian *Asian .002 * Yes American Indian *Black .04 7* Yes American Indian *Hispanic .8 21 No American Indian *Multi .900 No American Indian *White .07 7 No Asian*Black < .001 ** Yes Asian*Hispanic < .001 ** Yes Asian*Multi < .001 ** Yes Asian*White < .001 ** Yes Black*Hispanic < .001 ** Yes Black*Multi < .001 ** Yes Black*White < .001 ** Yes Hispanic*Multi < .001 ** Yes Hispanic*White < .001 ** Yes Multi* White < .001 ** Yes *: p value <0.05 **: p value <0.01 Answering R esearch Q uestion 1: Does race affect test scores from the Cognitive Abilities Test, CogAT 7 In answering Research Question 1 , data indicat es that yes, race does a ffect test scores. Overall , each ANOVA test comparing race to an individual battery (Verbal, Quantitative, and Nonverbal) resulted in a p value lower than the alpha value of 0.05; thus, rejecting the null hypothesis. In each case, there was a significant differen ce in test results between race . This led to post hoc testing, using Games Howell to test individual comparisons between racial groups to

PAGE 70

55 identify significant comparison differences. In these post hoc tests, consistently there was a difference between racial groups, with the exclusion of American Indian compared with Asian and once with Black . It should be noted that American Indian showed no statistically significant difference, except with Asian, across batteries. Also, American Indian population size was between 57 62 observations. This was an outlier when compared to other racial populations , ranging from 843 6053 . Therefore, these results should be taken with caution, as the number of cases drastically differed . R esearch Q uestion 2 : What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups ? It is clear from the literature that scores for Black and Hispanic groups are consistently lower (Ford, 2013; Plucker & Peters, 2016) . What the researcher is trying to determine is do the scores significantly affect the referral rates between racial groups , thus making the referral rates inequitable . In this section, each battery is separated and compared using ANOVA testing, across racial groups. Table 22 gives the f requency of 95 th percentile referral by demographic. Also, t he percentage referred compared to racial population is reported . For gifted referrals to be equitable, the percentage of referrals should closely align with the demographic make up for the scho ol district. Table 23 gives the overall referral rate for each demographic as a percentage of all gifted referrals a cross the district. The data was compared to the overall gifted referral and subsequently compared to overall demographic make up. Once a gain, these percentages should closely align if referrals are equitable across demographic groups. These frequencies were then used in the ANOVA testing to check for equitability across racial group.

PAGE 71

56 Table 22 Relative Frequency of Gifted Referral (95 th Percentile) by Race for each CogAT Battery Race CogAT Battery N Frequency Percentage of racial population American Indian Verbal Quanti ta tive Nonv erbal 62 5 5 59 2 7 4 3.1 10.8 6.3 Asian Verbal Quanti ta tive Nonv erbal 961 964 978 72 184 145 7.5 19.1 14.8 Black Verbal Quanti ta tive Nonv erbal 1086 1055 1118 15 23 13 1.4 2.2 1.2 Hispanic Verbal Quantit at ive Nonv erbal 2192 2139 2250 56 70 74 2.6 3.3 3.3 Multi Racial Verbal Quantit at ive Nonv erbal 917 894 920 51 65 68 5.6 7.3 7.4 White Verbal Quantit at ive No nv erbal 6046 5973 6117 403 583 567 6.7 9.8 9.3 Table 23 Relat ive Frequency of Gifted Referral (95 th Percentile) by Race Race N Frequency % of Racial Population % of Gifted Referrals from 95 th Percentile % of Whole Student Population American I ndian 60 10 16.7 0.6 0. 6 Asian 943 261 26.5 15.9 8.7 Black 1017 42 3.7 2.6 9.4 Hispanic 2069 147 6.4 8.9 19.1 Multi Racial 873 125 14.3 7.6 8.1 White 5850 1059 18.1 64.4 54.1

PAGE 72

57 A one way Welch ANOVA was conducted to determine if gifted referral at the 95 th percentile was significantly different across racial groups. Participants were classified into demographic groups, including: American Indian (n = 60), Asian (n = 943), Black (n = 1017 ), Hispanic (n = 2069), Multi Racial (n = 873), and White ( n = 5850). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tail s, as assessed by Q Q Plot . Homogeneity of variances was violated, as assessed by Levene's Test of Homogeneity of Variance (p = <.00 1). Table 24 relates population descriptives used for the ANOVA testing. 95 th percentile gifted referrals were statistically significantly different between racial groups, Welch's F (5, 592.85 ) = 99.73 , p < .001; therefore, we can reject the null hypothes is and can accept the alternative hypothesis. Race does create a significant difference in 95 th percentile gifted referral. Games Howell post hoc analysis (Table 25 ) revealed that statistically significant differences existed between all racial group com parisons; however, statistically significant difference was not found when comparing American Indian to other racial groups. Table 24 Population Descriptors , 95 th Percentile Gifted Referrals, Used for ANOVA Testing Race n Mean SD American Indian 60 . 17 .376 Asian 943 .28 .448 Black 1017 .04 .199 Hispanic 2069 .07 .257 Multi 873 .14 .350 White 5850 .18 .385

PAGE 73

58 Table 25 Games Howell Post Hoc Comparing 95 th Percentile Gifted Referral across Race Racial Comparison p value Sig nificant Difference American Indian *Asian .263 No American Indian *Black .1 22 No American Indian *Hispanic .3 78 No American Indian *Multi .997 No American Indian *White 1.000 No Asian*Black < .001 ** Yes Asian*Hispanic < .001 ** Yes Asian*Multi < .001 ** Yes Asian*White < .001 ** Yes Black*Hispanic .006 * Yes Black*Multi < .001 ** Yes Black*White < .001 ** Yes Hispanic*Multi < .001 ** Yes Hispanic*White < .001 ** Yes Multi*White < .040 * Yes *: p value <0.05 **: p value <0.01 All demographic group comparisons, excluding American Indian, revealed significant differences. Further testing is necessary to determine the degree to which this difference occurs. However, what can be said is that race does affect gifted referral using the CogAT7 at the 95 th percentile . R esearch Q uestion 3 : What is the difference in gifted referral, based on CogAT 7, at the 9 0 th percentile and above, between racial groups ? Part of the district effort to in crease diversity within gifted educa tion included referring students at the 90 th percentil e and above for further testing. This research question tested the significance of difference between racial groups at the 90 th percentile, to see if changing from 95 th percentile to 90 th percentile reduced the referral disparity between racial groups.

PAGE 74

59 Ta ble 26 gives the frequency of 90 th percentile referral by demographic. It also lists the percentage referred when compared to the racial population. The ANOVA testing indicated this difference was significant. Table 26 Relat ive Frequency of Gifted Refe rral (90 th Percentile) by Race for each CogAT Battery Race CogAT Battery N Frequency Percentage of racial population American Indian Verbal Quanti ta tive Nonv erbal 64 62 63 7 8 6 10.9 12.9 9.5 Asian Verbal Quanti ta tive Nonv erbal 961 964 978 132 294 249 1 3.7 30.5 25.5 Black Verbal Quantit at ive Nonv erbal 1086 1055 1118 34 59 43 3.1 5.6 3.8 Hispanic Verbal Quanti ta tive Nonv erbal 2192 2139 2250 99 145 148 4.5 6.8 6.6 Multi Racial Verbal Quantit at ive Nonv erbal 917 894 920 95 137 116 10.4 15.3 12.6 White Ve rbal Quantit at ive Nonv erbal 6046 5973 6117 797 1161 1083 13.2 19.4 17.7 Table 27 gives the overall referral for each demographic and compared it to the overall gifted referral and then to overall demographic make up. In order for the data to be equitabl e across racial groups, the referral percentage and the demographic make up should closely resemble each other.

PAGE 75

60 Table 27 Relative Frequency of 90 th Percentile Gifted Referral by Race Race N Frequency % of Racial Population Referred % of Total Gifted Re ferrals % of Whole Student Population American Indian 60 14 23. 3 .5 .6 Asian 943 385 40.8 13.5 8.7 Black 101 7 97 9.5 3.4 9.4 Hispanic 2069 253 12.2 8.9 19.1 Multi Racial 873 212 24.3 7.5 8.1 White 5850 1884 32.2 66.2 54.1 A one way Welch ANOV A was conducted to determine if gifted referral at the 90 th percentile was significantly different across racial groups. Participants were classified into demographic groups, in cluding: American Indian (n = 60 ), Asian (n = 943), Black (n = 1016), Hispanic (n = 2069), Multi Racial (n = 873), and White (n = 5850). There were no outliers, as assessed by inspection of boxplots. The data was normally distributed with fat tails, as assessed by the Q Q Plot . Homogeneity of variances was violated, as assessed b y Levene's Test of Homogeneity of Variance (p = <.001). Table 28 relates population descriptives used for the ANOVA testing. 90 th percentile gifted referrals were statistically significantly different between racial groups, Welch's F (5, 5 92.57 ) = 155.523 , p < .001; therefore, we can reject the null hypothesis and can accept the alternative hypothesis. Race does create a significant difference in 90 th percentile gifted referral . Games Howell post hoc analysis (Table 29) revealed that statistically signifi cant differences existed between all racial group comparisons, except with the following comparisons American Indian*Black, American Indian*Hispanic, American Indian*Multi, American Indian*White, and Black*Hispanic.

PAGE 76

61 Table 28 Population Descriptors , 90 th Percentile Gifted Referrals, Used for ANOVA Testing Race N Mean SD American Indian 60 .23 .427 Asian 943 .41 .492 Black 1017 .10 .29 5 Hispanic 2069 .12 .328 Multi 873 .24 .429 White 58 50 .32 .467 Table 29 Games Howell Post Hoc Comparing 90 th Per centile Gifted Referral across Race Racial Comparison p value Significant Difference American Indian *Asian .03 7* Yes American Indian *Black .1 54 No American Indian *Hispanic .3 54 No American Indian *Multi 1.000 No American Indian *White .601 No Asian*Bl ack < .001 ** Yes Asian*Hispanic < .001 ** Yes Asian*Multi < .001 ** Yes Asian*White < .00 1** Yes Black*Hispanic .234 No Black*Multi < .001 ** Yes Black*White < .001 ** Yes Hispanic*Multi < .001 ** Yes Hispanic*White < .001 ** Yes Multi*White < .001 ** Ye s *: p value <0.05 **: p value <0.01

PAGE 77

62 R esearch Q uestion 4: What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile? Using a stratified random sampling , a M ANOVA was conducted to determine if gifted referral at the 90 th and 95 th percentiles were significantly different across racial groups. The stratified random sampling was applied, to provide equal representation for each racial group. Since the Amer ican Indian demographic had a significantly lower number of observations and found that it was not significantly different with other demographics , excluding Asian, this racial group was excluded from this question. Participants were classified into demog raphic groups, including: Asian (n = 500), Black (n = 500), Hispanic (n = 500), Multi Racial (n = 500), and White (n = 500 ) . Homogeneity of variances was violated, as assessed by Levene's Test of Homogeneity of Variance (p = <.001). Means and standard d eviations are provided in Table 30 . There was a statistically significant difference between racial groups for 90 th percentile and 95 th percentile referrals , F (8, 4990) = 26.658, p = <.001; Pillai's Trace = .082; partial " 2 = .041 ; thus rejecting the nul l hypothesis and accepting the alternative hypothesis. There is a significant difference between 90 th and 95 th percentile referrals between race, using the CogAT 7. Table 30 Number, Mean, and Standard Deviation of Stratified Random Sampling , One way MANOV A 90 th Percentile 95 th Percentile Race n Mean SD Mean SD Asian 500 . 40 .49 0 . 28 . 451 Black 500 . 10 .300 . 03 .176 Hispanic 500 . 13 .334 . 08 .268 Multi 500 . 23 .423 . 13 .339 White 500 .3 2 .46 8 . 18 . 385

PAGE 78

63 In order to determine rates of referral and determine where the significance was situated in the One way MANOVA, follow up Chi square goodness of fit testing was conducted. The results of the follow up testing are reported in the next sections. 9 5 th Percentile Referral by Race , Chi Square Goodn ess of Fit In this section, the data is organized by racial group and referrals at the 9 5 th percentile. The complete population set was used in this analysis, since the researcher desired to evaluate total number of referrals for each demographic group. Relative referrals expected by population and observ ed referrals are given. C hi square goodness of fit test s are applied to deter mine if each racial group had the expected five percent of its population repr esented. The results of these C hi square tests are in Table 31 below. Table 31 Gifted Referral by Race at the 95 th Percentile , Chi Square Goodness of Fit Race Relative Expected by Population Observed p value Significant Difference American Indian 3 10 < .001** Yes +2.35 Asian 47 261 < .001** Yes +4 .55 Black 57 43 .051 No 0.25 Hispanic 115 148 .002 * Yes +0.29 Multi Racial 47 126 < .001 ** Yes +1.68 White 309 1072 < .001 ** Yes +2.47 *: p value <0.05 **: p value <0.01 Individual C hi square goodness of fit tests were conducted for each racial gro up to determine whether the pa rticipants in this study had similar proportion s at the 90 th percentile. The expected frequency for each demographic was five percent of the total population, as this was testing for the 95 th percentile and above. The C hi sq uare goodness of fit test indicated that

PAGE 79

64 the following demographics had significantly more referrals than expected: American Indian ( # 2 (1) = 14.757 , p = <.001 ); Asian ( # 2 (1) = 1020.971 , p = <.001); Hispanic ( # 2 (1) = 10.003 , p = . 002 ); Multi Racial ( # 2 (1) = 140.429 , p = <.001); and White ( # 2 (1) = 1978.562 , p = <.001). CogAT7 referred triple the number of American Indian than expected, f ive times the number of Asian, slightly more Hispanic, a little more than double the expected number of Multi R acial and more than triple the number of White. When testing for Black, the chi square goodness of fit test indicated that the CogAT7 test refer red within the limits of five percent of the population: Black ( # 2 (1) = 3.803 , p = . 051 ). Even though the test is pulling at least five percent of each racial population, the proportions are skewed severely when discussing the percentage of overall gifted referral disaggregated by race. For example, although CogAT7 i s screening around five percent of Black population, it is simultaneously pulling significantly more percentages of American Indian, Asian, His panic, Multi Racial, and White. Hispanic is only slightly higher at slightly more than expected; whereas, for As ian and White populations expected referrals were significantly higher at five times for Asian and triple for White. This large discrepancy leads the research er to examine if screening at the 90 th percentile alleviates or exacerbates this discrepancy. 90 t h Percentile Referral by Race , Chi Square Goodness of Fit In this section, the data is disaggregated by race and referrals at the 90 th percentile are used . The complete population set was used in this analysis, since the researcher wanted to evaluate tota l number of referrals for each demographic group. Relative referrals expected by population and observ ed referrals are given and C hi square goodness of fit test s are applied to deter mine if each racial group had the expected ten percent of the population repr esented. The results of these C hi square tests are in Table 32 below.

PAGE 80

65 Table 32 Gifted Referral by Race at the 90 th Percentile , Chi Square Goodness of Fit Race Expected Observed p value Significant Difference American Indian 7 14 .002* Yes +1.00 Asi an 94 385 < .001** Yes +3.10 Black 115 101 .175 No 0.12 Hispanic 230 259 .043* Yes +0.13 Multi Racial 94 216 < .001** Yes +1.30 White 618 1910 < .001** Yes +2.09 *: p value <0.05 **: p value <0.01 Individual C hi square goodness of fit tests were con ducted for each racial group to determine whether the racial populations in this study had similar proportion s at the 90 th percentile. The expected freque ncy for each demographic was ten percent of the total population, as this was testing for the 90 th pe rcentile and above. The C hi square goodness of fit test s indicated that the following demographics had significantly more referrals than expected: American Indian ( # 2 ( 1) = 9.615, p = .002 ) ; Asian ( # 2 ( 1) = 995.717, p = <.001 ) ; Hispanic ( # 2 (1) = 4.093, p = .043 ) ; Multi Racial ( # 2 (1) = 176.887, p = <.001 ) ; and White ( # 2 (1) = 2993.603, p = <.001 ) . CogAT7 referred twice the number of American Indian than expected, quadru ple the number of Asian, slightly more Hispanic, double the expected number of Multi R acial and triple the number of White. When testing Black, the C hi square goodness of fit test indicated that the CogAT7 test referred within the limits of ten percent of the population: Black ( # 2 (1) = 1.843, p = .175 ) . Even though the test is pulling at least ten percent of each racial population, the proportions are skewed severely when discussing the percentage of overall gifted r eferral disaggregated by race. For example, although CogAT7 is screening around ten percent of Black population, it is simultaneously pulling significantly more percentages of American Indian,

PAGE 81

66 Asian, Hispanic, Multi Racial, and White. Hispanic is only slightly higher at slightly more than expected; whereas, for Asia n and White populations expected referrals were significantly higher at almost quadruple times for Asian and over triple for Wh ite. This large discrepancy led the researcher to examine the overall gifted referral rate disaggregated by race. Overall Gifted Referral Rate Disaggregated by Race In the previous sections, goodness of fit was reported for the 95 th and 90 th percentiles. I t was found that observed gifted referrals and expected referrals for student groups , excluding Black, were significantly diffe rent. The researcher then wanted to determine if the proportions of overall gifted referral were similar to the overall student population. The goal of gifted education is to refer an equitable number of students from each demographic, based on the overal l student racial population. So, for example, if fifty percent of the given population is White, there should be around fifty percent of the gifted referrals coming from the White demographic. It was determined that C hi square goodness of fit test s would be conducted, as the overall percentages of gifted referral would be compared to the overall student population. Chi square goodness of fit tests were conducted to determine whether the students referred for further gifted testing were typical of the ove rall student population. The expected frequencies came from the overall student population (see Table 33). The Chi square goodness of fit test for the 95 th percentile indicated that the percentage of gifted referrals was statistically significantly diffe rent from the proportions found in the overall student population ( # 2 (4) = 17.527, p = .002). The Chi square goodness of fit test for the 90 th percentile indicated that the percentage of gifted referrals was statistically significantly different from the proportions found in the overall student population ( # 2 (4) = 15.449 , p = .004). These results indicated that, even with moving to the 90 th percentile for gifted referral, there continued to be a statistically

PAGE 82

67 significant inequitable distribution of gifted referral across demographics, highly favoring White and Asian popu lations. Table 33 Percent of Gifted Referral by Race at the 90 th and 95 th Percentiles 90 th Percentile Total Gifted Referrals 95 th Percentile Total Gifted Referrals % Overall Student Population Race Percent n Percent n American Indian 0.5 14 0.6 10 0.6 Asian 13.5 385 15.9 261 8.7 Black 3.4 101 2.6 43 9.4 Hispanic 8.9 259 8.9 148 19.1 Multi Racial 7.5 216 7.6 126 8.1 White 66.2 1910 64.4 1072 54.1 It should be noted that for Black, this change to 90 th percentile did significantly increase the num ber of Black students being referred for gifted evaluation. However, this also came at the cost of highly over referring both White and Asian populations. Referring at the 90 th percentile was not advantageous for Hispanic, since it did not significantly increase the number of Hispanic gifted referrals in comparison to the overall gifted referral, at 8.9% of the total gifted referrals at both the 90 th and 95 th percentiles. A lternative methods of screening and referrals should be sought to more equitably r efer students from varying demographics. Options for the se methods are discussed in the next chapter , as well as recommendations for future studies. Summary The results of this study indicate (a) CogAT 7 test scores are affected by race, (b) 95 th percenti le gifted referrals were significantly different across race, (c) 90 th percentile gifted referrals were significantly different across race, (d) when comparing 90 th and 95 th percentile referrals, there was a significant difference, and (e) gifted referrals at the 90 th and 95 th percentiles were significantly not equitable across race. Although screening for the 90 th percentile does

PAGE 83

68 increase the number of referrals per demographic, there is still a vast over referral of Asian and White and under r eferral of Black and Hispanic populations . So, although one can argue that moving to the 90 th percentile statistically increases the number of referrals for each demographic, it does not dramatically of gifted referrals. Therefore, alternate referral methods and fo llow up studies are recommended .

PAGE 84

69 CHAPTER V DISCUSSION In this chapter, the study is summarized along with findings of each research question. Conclusions are drawn and related back to the current body of literature introduced in Chapter 2. Implicati ons and future research is outlined as potential future studies, namely: (a) development of equity goals, (b) second screening process, and (c) talent enrichment programs and/or strategies for CLD learners. Finally, a summary and final thoughts are drawn as the researcher concludes this study. Summary of the Study Controversy abounds in education around racial topics, including achievement gap, opportunity gap, educational debt, and the excellence gap. These gaps, or significant differences in achieveme nt, resources, and/or opportunity, create inequitable access to gifted education. This study was developed to determine whether current practices in screening students at the 90 th percentile, compared to the traditional 95 th percentile, would help to miti gate the racial discrepancies and inequities currently prevailing in gifted screening practices. This referral at the 90 th percentile was not meant to re establish the 95 th percentile identification practices and/or requirements, but for referral purposes . Further testing would be conducted after the referral and current CDE practices and policies would be used for gifted identification criteria. Data were collected from a Colorado school district, where equity work is paramount. The district assessment department provided the researcher with three years' worth of CogAT 7 data, used as a universal screener for gifted referral . The full nine subtest assessment were administered to ever y second grader in the district. A ny student scoring 90 th percentile or above

PAGE 85

70 on any of the three batteries were referred for additional achievement testing, which is used to create a body of evidence for gifted identification. The following research questions framed this study: 1. ! Does race affect test scores from the Cogni tive Abilities Test, CogAT 7? 2. ! What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile and above, between racial groups? 3. ! What is the difference in gifted referral, based on CogAT 7, at the 90 th percentile and above, between racia l groups? 4. ! What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile? The results of this study indicate d (a) CogAT 7 test scores are affected by race, (b) 95 th per centile gifted referrals were significantly different across race, (c) 90 th percentile gifted referrals were significantly different across race, (d) when comparing 90 th and 95 th percentile referrals, there was a significant difference, and (e) gifted refe rrals at the 90 th and 95 th percentiles were significantly not equitable across race. Therefore, alternate referral methods and follow up studies are recommended. Findings Each of the results identified in the last section are discussed here. They are o rganized by (a) race and test score, (b) 95 th percentile referrals, (c) 90 th percentile referrals, and (d) comparison between 90 th and 95 th percentile referrals. Race and Test Scores This study found a statistically significant difference in test scores be tween racial groups. When answering Research Question 1 , "Does race affect test scores from the Cognitive Abilities

PAGE 86

71 Test, CogAT 7 ," this study indicated that race does affect CogAT 7 test scores , with significant underrepresentation of Black and Hispanic populations when compared with White and Asian counterparts . This was consistent with findings regarding the achievement and excellence gaps across the nation (Ford, 2013; Miller, 2004; Olszewski Kubilius & Thomson, 2010; Plucker & Peters, 2016). In the literature, Lohman (2 005), creator of CogAT 7, states there are two types of giftedness: high accomplishment (achievement oriented) and high potential (aptitude oriented). He also states , "predictors of achievement in reading, mathematics, social studies, and science are the same for White, Black, Hispanic, and Asian American students" (p. 344). Although the researcher agrees with Lohman that there are two ways to look at gifted , achievement and aptitude , the researcher disagrees that this aptitude can be found in one test , as evidenced by the discrepancy in gifted referral by race from the CogAT 7 . Also, from this study the researcher found that the predictors of achievement are significantly different across racial groups, in contrast with L ohman (2005) that predictors are the same across demographic groups. Lohman (2005) states that current predictors are based on past performance, to which the researcher agrees. However, it is generalized that racial groups have similar performance, to which this stu dy demonstrates that there is statistically significant difference when comparing racial groups . The findings of this study are similar to numerous studies acknowledging the underrepresentation of gifted minoritized groups, Black and Hispanic ( Feldman, 20 03; McBee, 2006; Naglieri & Ford, 2015; Plucker, Burroughs, & Song, 2010; Worrell, 2014). 95 th Percentile Referrals When answering Research Question 2, " What is the difference in gifted referral, based on CogAT 7, at the 95 th percentile a nd above, bet ween racial groups," the study shows that e ven with the adjustments to CogAT 7 (Lohman, 2011), it is not equitably referring all racial groups at

PAGE 87

72 the 95 th percentile . Although the top 5 th percentile of all racial groups were referred , thus re sulting in re ferring at least five percent of each population, the inequitable distribution of referral is significant. For example, Black students are referred at the threshold of five percent , while Asian students were referred at three times that rate . Likewise, W hite students were referred at twice the rate of their Black counterparts. Thus, referrals at the 95 th percentile are inequitable. Lohman, himself states that using CogAT 7, "does not eliminate group differences" (Lohman, n.d.). This study shows that no t only does it not eliminate group differences, these differences are statistically significant. Another approach is using a nonverbal screening process, such as with the Naglieri Nonverbal Ability Test ( NNAT ) . Naglieri, creator of NNAT, and Ford (2015) discuss the need for a nonverbal testing screener that more equitably refers students from Black and Hispanic populations. Giessman, Gambrell, and Stebbins (2013) conducted a study to identify the difference between referrals using the CogAT and the NNAT . They found that the Nonverbal CogAT 6 score identified as many minoritized students as the NNAT 2. However, one of the limitations to this study was that the sample groups were different; one including second graders, the other including kindergarten t hrough second grade. Another important limitation is that the same students did not take both tests. Therefore, comparing the two samples should be taken with caution , as the difference or lack of difference could be due to the samples, not necessarily t he tests . A follow up study screening with a nonverbal test compared with the CogAT 7 , where students take both tests , would aid in determining if the nonverbal test more equitably refers Black and Hispanic students .

PAGE 88

73 90 th Percentile Referrals Answeri ng Research Question 3, " What is the difference in gifted referral, based on CogAT 7, at the 90 th percentile and above, between racial groups ," it was found that moving to the 90 th percentile did little to influence the overall proportionality for gifted r eferral across racial groups. Table 34 shows the number of students referred at the 90 th percentile compared to the 95 th percentile. Table 34 Relative Frequency of 90 th and 95 th Percentile Gifted Referral by Race 90 th Percentile Total Gifted Referrals 95 th Percentile Total Gifted Referrals % Overall Student Population Race Percent N Percent N American Indian 0.5 14 0.6 10 0.6 Asian 13.5 385 15.9 261 8.7 Black 3.4 101 2.6 43 9.4 Hispanic 8.9 259 8.9 148 19.1 Multi Racial 7.5 216 7.6 126 8.1 Whit e 66.2 1910 64.4 1072 54.1 Although frequencies change for Black students (from 43 to 101 ), over doubling the number of Black students referred , when compared t o White students (from 1072 to 1910 ) and Asian students (from 261 to 385), there is some chang e to the proportionality of overall gifted referral; however, this is not significant enough to support 90 th percentile referrals, as shown in Table 35. It increased Black representation from 2.7% overall representation to 3.4% at the 90 th percentile refe rral. However, it also increases White representation from 64.4% to 66.2%. The effects of this change are discussed in the next research question. For Hispanic students, there was an increase in frequency from 148 to 259 gifted referrals; however, when compared to White and Asian populations, this does nothing to increase the overall gifted referral proportionality when compared to other racial groups, remaining at 8.9% in both 95 th and 90 th percentile referrals. As

PAGE 89

74 the Hispanic student population is th e fastest growing minority group in the state of Colorado (CDE, 2018), CLD gifted referral needs to be a high priority. The results of this study conclude that moving to the 90 th percentile referral alone does little to reduce the overall disproportionali ty of racial discrepancies in gifted referral. Table 35 Percent of Gifted Referral by Race at the 90 th and 95 th Percentiles Race Relative Percent of Gifted Referrals 90 th Percentile 95 th Percentile % of Overall Student Populati on American Indian 0.5 0.6 0. 6 Asian 13.5 15.9 8.7 Black 3.4 2.6 9.4 Hispanic 8.9 8.9 19.1 Multi Racial 7.5 7.6 8.1 White 66.2 64.4 54.1 Comparison Between 90 th and 95 th Percentile Referrals Answering Research Question 4, " What is the significance of difference of gifted referral between racial groups, based on the CogAT 7, from the 90 th percentile to the 95 th percentile ," the results show that there was mixed review of benefit and that the results have to be considered in the context of the whole findings across demographic groups. For instance, i t was noted in the results section that for Black students , the change to 90 th percentile referral did significantly increase the number of Black students , from 43 to 101 (see Table 34), being referred fo r gifted evaluation. However, this also came at the high cost of a significant increase of gifted referrals in both White and Asian populations (see T able 34) . Referring at the 90 th percentile was advantageous for Hispanic students, as it the frequency o f referral increased from 148 to 259; however, it did not significantly increase the proportionality of Hispanic referrals in comparison to overall gifted referral s , at 8.9% of the total referrals at both the 90 th and 95 th percentiles.

PAGE 90

75 Conclusions The conc lusions for this study are not uncommon in the field of gifted education , namely: (a) race affects test scores, (b) the need for inclusive referral practices is evident, and (c) underrepresentation of ethnically diverse gifted students exists and is preval ent . This section addresses each of these conclusions and relate them back to literature introduced in Chapter 2: Literature Review. Race as a significant contributor to gifted referral adds to the body of research on racial inequities in gifted education (Cornell, Delcourt, Goldberg, & Bland, 1995; Erwin & Worrell, 2012; Grantham & Ford, 2003). What is unclear is the cause. There are possibilities , including: (a) testing bias, (b) using aptitude vs current ability, (c) poverty as a factor, and (d) langu age as a factor. Testing bias es in intelligence tests has been scrutinized for the lack of norming across different minority groups. Black and Latino children consistently fall, on average, 10 15 points lower than middle class White children (Ford, 2013 b ). This is not because the Black and Latino students are any less smart, but because tests were normed across a majority of White children (Ford, 2013 b) . Joseph and Ford (2006) state, "Nondiscriminatory assessment is concerned with fairness in all aspect s of evaluating individuals" (p. 44). Some suggestions for equitable processes include: (a) least biased instruments, (b) avoiding confirmatory bias, and (c) ensuring policies and procedures are fair or nondiscriminatory (Joseph & Ford, 2006) ! To become i nclusive in gifted referral and identification practices, it is important to remember that enacting policies that are completely unbiased and equitable is unrealistic (Joseph & Ford, 2006; Ortiz, 2007). Ford (2012) states, "no field of education is free, or will be free, of dilemmas of difference " (p. 392). However, it is clear that research is needed regarding disproportionalities of gifted referral and equitable practices. " Through research on least bias ed

PAGE 91

76 instruments, using aptitude vs current ability, dispelling myths and biases around CLD gifted learners, considering poverty, factoring in language development, and changing processes to be more inclusive, state and district units can work together to close excellence gaps. Table 36 reports Colorado s tudent population trends disaggregated by race. Hispanic student population is the fastest growing demographic in Colorado, with well over double the next demo graphic group, Multi Racial . Y et , it is the most underserved population s in Colorado gifted edu cation , where American Indian (3.6%), Asian (9.7%), Black (2.7%), Hispanic (3.4%), Multi Racial (7.1%), and White (8.4%) are the percentage of public students enrolled in a gifted program by race (National Center for Educational Statistics, 2016). For pro portionality across race, there should be approximately 5% of any given population in gifted programming. This inequitable trend is not isolated in Colorado; it is a nation wide dilemma. Table 36 Colorado State Student Population Trends 2010 2017 (CDE, 2018) School Year Race 2010 2011 2016 2017 Difference American Indian 7452 6511 941 Asian 24493 28309 3816 Black 40537 41478 941 Hispanic 266098 303573 37475 Multi Racial 23565 36388 12823 White 479327 486537 7210 The literature reviewed and th is study demonstrate that underrepresentation of ethnically diverse gifted students needs to be a matter of great consideration in gifted education . Inequity of gifted referral processes will not be resolved quickly; it is woven into current systemic prac tices (Ford, 2012; Sullivan, 2011). Whatever decisions are for equitable referral and

PAGE 92

77 identification for gifted education, researchers must begin to look differently; a lternative entry points for inclusio n of minoritized groups must be considered . Without these changes, systems will continue to over identify White and Asian children and under identify Bla ck and Hispanic children . Implications and Future Research In the next sections, recommendations are addressed and introduced as follow up studies to bett er reduce the inequity of gifted underrepresentation , namely: (a) development of equity goals, (b) second screening process, and (c) talent enrichment programs and/or strategies for CLD learners. Equity goals will focus attention of gifted education to eq uity driven practice across referrals, curricula, and programming. Changing screening practices to be more inclusive will decrease the underrepresentation of minoritized groups. The development of talent enrichment programs for high ability CLD students, utilizing a combination of CLD and gifted strategies , will aid in identifying and retaining CLD gifted learners in programs. All of these equity driven strategies and further studies will aid in mitigating the disproportionality of minoritized population s in gifted education. Equity Goals One equity driven method in gifted education is to introduce equity goals (Ford, 2013). Ford proposes a formula for equitable identification practices for underrepresent ed populations, namely American Indian , Hispanic, and Black students; applying the Office of Civil Rights (OCR) Civil Rights Data Collection (CRDC), to the current district gifted identification, to determine equitability of demographic distributions. The formula is as follows: Percentage of Black/Hispan ic students in district (P) x 20% = B; Equity goal (E) = P B. Using this equity formula, current Distric t data distribution is applied in Table 37 . The equity goal encourages the

PAGE 93

78 district equity goal for gifted identification should move from the current 17% of American Indian, Hispanic, and Black populations to 22.6% , to more equitably represent the student population . Multi racial groups are equitably represented in current gifted identification and programming. White and Asian populations continue to be overrepresented. These equity goals can be accomplished through equity driven screening practices , focused on identifying our Black and Hispanic students, and inclusiveness in gifted enrichment programming. Table 3 7 District Enrollment Compared to Gif ted Identification, Disaggregated by Race Ethnic Group 2016 2017 2018 American Indian / Hispanic/ Black % District Population 31.07 31.43 31.57 % District Gifted Education 18 17 17 % Underrepresentation 42 46 46 % Equity Goal/Target for Gifted Representation (minimal discrepancy using 20% formula) Equity Goal = 22.6 Increase from 18% to 22.6% Equity Goal = 22.6 Increase from 17% to 22.6% Equity Goal = 22.6 Increase from 17% to 22.6% Multi Racial % District Population 6.22 6.63 6.91 % District Gifted Education 6 6 6 % Underrepresentation 3.5 9.5 13.2 % Equity Goal/Target for Gifted Representation (minimal discrepancy using 20% formula) Equity Goal = 4.4 Minimal equity goal of 4.4% has been met. Equity Goal = 4.4 Minimal equity goal of 4.4% has been met. Equity Goal = 4.4 Minimal equity goal of 4.4% has been met. White/Asian % District Population 62.71 61.95 61.52 % District Gifted Education 76 77 77 % Underrepresentation 0 0 0 % Equity Goal/Target for Gifted Representa tion (minimal discrepancy using 20% formula) Overrepresented Overrepresented Overrepresented (Ford, 2013; Office of Assessment and Evaluation, District, 2018)

PAGE 94

79 Second Screening as an Equity Driven Alternative It is evident from the literature that t here is not a consensus o n which testing measures to use. This is due to various studies with conflicting data (Lakin & Lohman, 2011; Naglieri & Ford, 2005) . The practice of using one test or measure will continue to omit highly able student groups, rega rdless of which measure is applied (Plucker & Peters, 2016). One test or measure is not enough (Ford, 2013b; Frasier, 1991). No one instrument should be used to exclude, nor include st udents in a program (Felder, et al., 2015; Ford, 2013 b). Research pro poses screening in multiple ways to capture highly able students from various backgrounds and with various strengths (Ford, 2013 b ; Plucker & Peters, 2016). In order to do this, the researcher recommends a double screening. T he second screening would be a focused screen through Title I schools, as Title I schools are federally designated as requiring additional support . These schools would be given a second screening test that is suggested to more equitably screen for diverse populations (Ford & Naglieri, 2003). Any student scoring in the 95 th percentile or above on either screener would be referred for additional testing. Title I school data could be used in a follow up study to determine if giving the second screener will indeed more equitably refer st udents, reducing disproportionality between racial groups. Currently in District , students screened between the 90 th and 94 th percentile s are then given achievement testing . After this achievement testing , even with a 95 th percentile or above in an acad emic content area, a student can only be placed in a talent pool designation until there is more evidence for gifted determination, per current Colorado Department of Education guidelines ( CDE , 2016 ). As the guidelines state, a student must have three dat a points for any given academic content area in the 95 th percentile to be recommended for gifted identification. With a score between the 90 th to 94 th percentile, a student may be recommend ed for talent pool

PAGE 95

80 designation. F ull identification cannot be giv en , unless a student is administered more testing beyond the one achievement test . As minoritized groups do not perform as high on achievement testing, there is controversy over whether this testing is biased against these groups, namely Black and Hispani c groups ( Plucker & Peters , 2016 ). So, based on the evidence in this current study, lowering referral to the 90 th percentile to increase equi table representation of minoritized groups, does little for Hispanic students . A lthough the Black referral rate i ncrease was significant , the underrepresentation dilemma was still significant . When comparing how many White and Asian students are also referred at the 90 th percentile, there were triple, quadruple, and more referred for gifted evaluation in these popul ations. This inequity continued when screening at the 90 th percentile. As the data revealed, even at the 90 th percentile, White and Asian students were still vastly over referred, while Black and Hispanic students were vastly under referred. The rese archer proposes more strategic use of the money spent on screening. Instead of screening for the 90 th percentile for further testing , the resea rcher proposes the following as an equ ity driven practice for gifted referral. The researcher suggests screenin g in a focused manner, using a combination of cognitive tests. CogAT (Verbal, Quantitative, and Nonverbal) would still be given as a universal screener across the district. This screener would be utilized, even with the inequity, because it draws out ver bal and quantitative thinkers. Further academic testing, to build a body of evidence, would be recommended for any student scoring at the 95 th percentile or hi gher . Students scoring 90 th to 94 th percentiles are watched and given access to opportunities t o enrich their learning through advanced courses and/or enrichment . These students are referred later if their regular academic testing demonstrates strength areas. This would be on a case by case basis. An additional nonverbal test, suggested as a more equitable measure (Ford & Naglieri, 2003), will be administered at Title I schools.

PAGE 96

81 For Title I schools, the screening practices would be slightly different. In addition to the universal screener given to all second graders across the district, Title I s tudents would be given an additional nonverbal screener, the NNAT3 , as this is suggested to more equitably screen for diverse populations ( Ford & Naglieri , 2003 ). Any student at Title I schools scoring 95 th percentile on either the CogAT7 or the NNAT3 wou ld receive further academic testing . Title I schools are the focus for additional screening, as these schools are federally identified as having high populations of students in poverty ( U.S. Department of Education, 2015) and potentially higher CLD popula tions (Mantosis, 2013) . Through a focused approach to gifted screening practices, using an equity driven second screening, systemic inequities can be reduced. More research needs to be gathered to find pathways that demonstrate equitable referral practic es. While the paperwork and identification processes catch up, much can be done to develop the talents of all students, especially high ability CLD learners. Talent Enrichment Programs for High Ability CLD Students In addition to focused funding for cogni tive and academic testing, the researcher recommends further use of and research for talent enrichment programs for high ability CLD students. The researcher recommends collaboration with school districts that have published work around equity work for hi gh ability CLD learners. High ability CLD learners are defined here as students scoring in the top quartile, namely the 75 th percentile and above . There is overwhelming research suggesting that opportunities need to be given to students of color in order to bridge the opportunity and achievement gaps (Plucker & Peters, 2016) . The researcher suggests a study where best practice CLD and gifted strategies are studied and the common overlapping strategies used to develop or determine curricula and/or strateg ies for high ability

PAGE 97

82 CLD learners. These strateg ies would be studied with high ability CLD learners to test effectiveness in closing the achievement gap. The effectiveness of strategies and/or curricula would be studied using norm referenced assessments. Through developing high ability CLD learner talents, the effects of opportunity and achievement gaps will be lessened. Students will have access to advanced coursework and receive scaffolded supports necessary for their success. In the end, while equi table referral and identification process es are adapted there will be access for current high ability CLD learners to rigorous and advanced curricula. Summary Through providing focused use of resources and funds, more CLD students can be identified as exc eptional and receive gifted services and opportunities. In this way, system s can move to increasing CLD populations in gifted programming , while striving for more equitable representation from White and Asian populations. It is clear that further researc h is necessary to decrease the inequity of systemic gifted screeners for referrals . Data in this field are needed to determine whether systemic changes are increasing the equitable representation of all racial groups in gifted programming. Then, further, there is need for rigorous advanced programming for and the retention of CLD students, through the use of CLD and gifted strategies together. Final Thoughts Inequity of gifted referral processes will not be resolved quickly; it is woven into current syste mic practices (Ford, 2012; Sullivan, 2011). This study shows that even with moving to the 90 th percentile as a referral marker, there are still statistically significant inequities in racial representation in gifted referral. Equity driven practices such as (a) equity goals, (b) second screening processes, and (c) talent enrichment programs and/or strategies for CLD learners can

PAGE 98

83 help mitigate the current issues surrounding inequity in gifted referral and thus programming . Further research is necessary to learn more regarding CLD gifted learners, for as Donna Ford (2013a) stated, " t he more we know about others, the less we make up" (p. 66). As further study is done on CLD gifted referrals and programming , the field will move closer to utilizing practices that lead to more equitable representation in gifted population s .

PAGE 99

84 REFERENCES Anderson, S., Medrich, E., & Fowler, D. (2007). Which achievement gap? Phi Delta Kappan, 88 , 547 550. Baltimore County Public Schools. (2 017). Planning the methodology: Limit ations and d elimitations. Retrieved from BCPS Independent Research Seminar: https://www.bcps.org/offices/lis/researchcourse/develop_writing_methodo logy_limitatio ns.html Beisser, S. R. (2008). Unintended consequences of the no child left behind mandates on gifted students. Forum on Public Policy: A Journal of the Oxford Round Table, 1 13. Borland, J. H. (2003). The death of giftedness: Gifted educ ation without gifted children. Rethinking Gifted Education , 10 , 105. Borland, J. H. (2009). Myth 2: The gifted constitute 3% to 5% of the population. Moreover, giftedness equals high IQ, which is a stable measure of aptitude: Spinal tap psychometrics i n gifted education. Gifted Child Quarterly , 53 (4), 236 238. Boyles, D., Carusi, T., & Attick, D. (2009). Historical and critical interpretations of social justice. In W. Ayers, T. M. Quinn, & D. Stovall (Eds.), Handbook of Social Justice in Education (pp . 30 42). New York: Routledge. Business Wire. (2013, November). Houghton Mifflin Harcourt's Cognitive Abilities Test (CogAT) evolves to meet the needs of English language learners. Retrieved from Business Wire: http://www.businesswire.com CDE: Colorado Department of Education. (2018). Colorado education facts and figures . Retrieved from https://www.cde.state.co.us/communicatio ns/coeducationfactsandfigures CDE: Office of Gifted Education (2016). Gifted identification. Colorado Department of Education, Office of Gifted Education. Denver: CDE. Retrieved from CDE Colorado Department of Education: https://www.cde.state.co.us/gt/identification District (2017). Definition and identification . Retrieved from Advanced Academic Services: web address not available. Colangelo, N., Assouline, S. G., & Gross, M. U. (2004). A na tion deceived: How schools hold back America's brightest students. The Templeton National Report on Acceleration. International Center for Gifted Education and Talent Development. Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J., Mood, A. M ., Weinfeld, F. D., & Y ork, R. L. (1966). Equality of educational opportunity. U.S. Department of Health, Education, and Welfare . Washington: National Center for Educational Statistics.

PAGE 100

85 Collins , C., & Yeskel, F. (2013). The dangerous consequence of growi ng i nequality. In M. Adams, W. J. Blumenfeld, C. Castaneda, H. Hackman, M. L. Peters, & X. Zuniga, Readings for Diversity and Social Justice (3rd ed., pp. 157 164). New York: Routledge. Colorado Department of Education (2017). CMAS Mathematics, English Language Arts, Science and Social Studies Data and Results. Retrieved from Colorado Department of Education : https://www.cde.state.co.us/assessment/cmas dataandresults Cornell, D. G., Delcourt, M. A., Goldberg, M. D., & Bland, L. C. (1995). Achievement and self concept of minority students in elementary school gifted programs. Talents and Gifts , 18 (2), 189 209. Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualita tive, quantitative, and mixed methods approaches (5th ed.). Los Angeles: Sage. Egalite, A. J., Fusarelli, L. D., & Fusarelli, B. C. (2017). Will decentralization a ffect educational i nequity? The Every Student Succeeds Act. Educational Administration Qua rterly , 53 (5), 757 781. Encyclopedia Britannica. (2016, March). National Defense Education Act (NDEA). (T. C. Hunt, Editor) Retrieved from Encyclopedia Britannica: https://www .britannica.com/topic/National Defense Education Act Erwin, J. O., & Worrell, F. C. (2012). Assessment practices and the underrepresentation of minority students in gifted and talented education. Journal of Psychoeducational Assessment , 30 (1), 74 87. E xceptional Student Services Unit. (2016, March). ECEA rules: Rules for the administration of the exceptional children's educational act. Retrieved from CDE: Colorado Department of Education: Retrieved from https://www.cde.state.co.us/sites/default/files/documents/spedlaw/download/ecearules march2013.pdf Felder, M. T., Taradash, G. D., Antoine, E., Ricci, M. C., Stemple, M., & Byamugisha, M. (2015). Increasing diversity in gifted education: Research based strategies for identification and program service . Waco, Texas: Prufrock Press Inc. Feldman, D. H. (2003). A developmental, evolutionary perspective on giftedness. In J. Borland (Ed.), Reth inking gifted education , 9 33. New York: Teachers College Press. Ford, D. Y. (1998). The underrepresentation of minority students in gifted education: Problems and promises in recruitment and retention. The Journal of Special Education , 32 (1), 4 14. Fo rd, D. Y. (2011). Multicultural gifted education. Waco, Texas: Prufrock Press Inc.

PAGE 101

86 Ford, D. Y. (2012). Culturally different students in special education: Looking backward to move forward. Exceptional Children , 78 (4), 391 405. Ford, D. Y. (2013 a ). Mult icultural issues: Gifted underrepresentation and prejudice Ñ Learning from Allport and Merton. Gifted Child Today , 36 (1), 62 67. Ford, D. Y. (2013 b ). Recruiting and retaining culturally different students in Gifted education. Waco, TX: Prufrock Press Inc. Ford, D. Y., Harris III, J. J., Tyson, C. A., & Trotman, M. F. (2001). Beyond deficit thinking: Providing access for gifted African American students. Roeper Review , 24 (2), 52 58. Ford, D. Y., Moore III, J. L., & Scott, M. T. (2011). Key theories and fra meworks for improving the recruitment and retention of African American students in gifted education. The Journal of Negro Education , 239 253. Ford, D., & Thomas, A. (1997). Underachievement among gifted minority students: Problems and promises . ERIC C learinghouse on Disabilities and Gifted Education. Frasier, M . M. (1991). Disadvantaged and culturally diverse gifted s tudents. Journal for the Education of the Gifted , 14(3), 234 245. Frasier, M. M., Hunsaker, S. L., Lee, J., Finley, V. S., Garcia, J. H., Martin, D., & Frank, E. (1995). An exploratory study of the effectiveness of the staff development m odel and the research based assessment plan in i mproving the identification of g ifted economically disadvantaged s tudents. Storrs, CT: National Resea rch Center on the Gifted and Talented. Gallagher, C. A. (2003). Color blind privilege: The social and political functions of erasing the color line in post race A merica. Race, Gender & Class, 10 (4), 22 37. Gifted Child Today. (2011, October). Javits gi fted and talented students e ducation a ct. Gifted Child Today , 8. Govil, P. (2013). Ethical considerations in educational research. International Journal of Advancement in Education and Social Sciences , 2322 0023. Grantham, T. C. (2002, Winter). Straigh t talk on the issue of underrepresentation: An interview with Dr. Mary M. Frasier. Roeper Review, 24 (2), 50 51. Grantham, T. C., & Ford, D. Y. (2003). Beyond self concept and self esteem: Racial identity and gifted African American students. The High Sc hool Journal , 87 (1), 18 29. Great Schools Partnership (2013, September). Opportuni ty g ap. Retrieved from The Glossary of Educational Reform for Journalists, Parents, and Community Members: https: //www.edglossary.org/opportunity gap/

PAGE 102

87 Greenfield, P. M. (1997). You can't take it with you: Why ability assessments don't cross cultures. American Psychologist , 52 (10), 1115. Helms, J. E. (1992). Why is there no study of cultural equivalence in standard ized cognitive ability testing?. American Psychologist , 47 (9), 1083. Houghton Mifflin Harcourt (2017). Cognitive Abilities Test . Retrieved from Houghton Mifflin Harcourt: http://www.hmhc o.com/hmh assessments/ability/cogat 6 Hos, R. (2016). Caring is not enough: Teachers' enactment of ethical care for adolescent students with limited or interrupted formal education (SLIFE) in a newcomer classroom. Education and Urban Society, 48 (5), 479 503. Islas, M. R. (2015, October). Hold on Javits works. Retrieved from On Education, Think Again: https://www.politico.com/agenda/story/2015/10/on education think again 000271 Jennings, J., & Rentner, D. S. (2006). Ten big effects of the No Child Left Behind Act on public schools. Phi Delta Kappan , 88 (2), 110 113. Johnson, A. G. (2006). Privilege, power, and difference (2nd ed.). New York, NY: McGraw Hill . Johnson, B., & Christensen, L. (2012). Educational research: Quantitative, qualitative, and mixed approaches. Los Angeles, CA: Sage. Jolly, J. L., & Makel, M. C. (2010). No Child Left Behind: The inadvertent costs for high achieving and gifted student s. Childhood Education , 87(1), 35 40. Joseph, L. M., & Ford, D. Y. (2006, Winter). Nondiscriminatory assessment: Considerations for gifted education. Gifted Child Quarterly , 50(1), 42 51. Lakin, J. M., & Lohman, D. F. (2011). The predictive accuracy of verbal, quantitative, and nonverbal reasoning tests: Consequences for talent identification and program diversity. Journal for the Education of the Gifted , 34(4), 595 623. Langley, S. D. (2015, December 11). Board Member NAGC: National Association for Gi fted Children. The State of the Nation in Gifted Education. (L. Jacobs, Interviewer) Blog Talk Radio. Education Talk Radio. Laerd Statistics (2017). Statistical tutorials and software guides . Retrieved from https://statistics.laerd.com/ Lohman, D. (n.d.). Introducing form 7 of the Cognitive Abilities Test [PDF document]. Retrieved from https://faculty.education.uiowa.ed u/docs/dlohman/CogAT7 on the road4.pdf

PAGE 103

88 Lohman, D. F. (2005). An aptitude perspective on talent: Implications for identification of academically gifted minority students. Journal for the Education of the Gifted , 28(3/4), 333 360. Lohman, D. F. (2011). Cognitively speaking: Introducing CogAT Form 7. Houghton Mifflin Harcourt . Retrieved from https://www.hmhco.com /~/media/sites/home/hmh assessments/assessments/cogat/pdf/cogat cognitively speaking v7 aug 2011.pdf?la=en Lohman, D. F. (2012). Cognitive Abilities Test, Form 7: Research and development guide . Rolling Meadows, IL: Riverside Publishing. Lohman, D. F., & Gambrell, J. L. (2012). Using nonverbal tests to help identify academically talented children. Journal of Psychoeducational Assessment , 30 (1), 25 44. Lohman, D. F., & Korb, K. A. (2006). Gifted today but not tomorrow? Longitudinal cha nges in ability and achievement during elementary school. Journal for the Education of the Gifted , 29 (4), 451 484. Lohman, D. F., Korb, K. A., & Lakin, J. M. (2008, Fall). Identifying academically gifted English language learners using nonverbal tests: A comparison of the Raven, NNAT, and CogAT. Gifted Child Quaterly , 52 (4), 275 296. Malin, J. R., Bragg, D. D., & Hackmann, D. G. (2017). College and career readiness and the Every Student Succeeds Act. Educational Administration Quarterly , 53 (5), 809 838 . Mantosis, G. (2013). Class in America 2006. In M. Adams, W. J. Blumenfeld, C. Castaneda, H. Hackman, M. L. Peters, & X. Zuniga, Readings for Diversity and Social Justice (3rd ed., pp. 150 156). New York: Routledge. Martin, E. W., Martin, R., & Terma n, D. L. (1996, Spring). The legislation and litigation history of special education. The Future of Children: Special Education for Students with Disabilities , 6 (1), 25 39. McBee, M. T. (2006). A descriptive analysis of referral sources for gifted identi fication screening by race and socioeconomic status. Journal of Secondary Gifted Education , 17 (2), 103 111. Mccain, M., & Pfeiffer, S. (2012, February). Identification of gifted students in the United States today: A look at state definitions, policies , and practices. Journal of Applied School Psychology , 28 (1), 59 88. Medina, M. (2016). Gifted identification: Chapter 3 revised. Colorado Department of Education, Office of Gifted Education. Denver: Colorado Department of Education. Meyerson, D. E. ( 2001). Tempered radicals: How people use difference to inspire change at work. Boston: Harvard Business School Publishing.

PAGE 104

89 Miller, L. S. (2004). Promoting sustained growth in the representation of African Americans, Latinos, and Native Americans among to p students in the United States at all levels of the education system. National Research Center on the Gifted and Talented NRCGT. Missett, T., & McCorm ick, K. (2014). Conceptions of g iftedness. In J. A. Plucker, & C. M. Callahan (Eds.), Critical Issues and Practices in Gifted Education: What the Research Says (pp. 143 157). Waco, TX: Prufrock Press Inc. NAGC. (2013). State of the Nation. Retrieved from National Association for Gifted Children: http://www.nagc.org/sites/default/files/Advocacy/State%20of%20the%20Nation.pdf NAGC. (2018). Jacob Javits gifted and t alented s tudents e ducation a ct. Retrieved from National Association for Gifted Children: https://www.nagc.org/resources publications/resources university professionals/jacob javits gifted talented students Naglieri, J. A. , & Ford, D. Y. (2005, Winter). Increasing minority children's participation in gifted classes using the NNAT: A response to Lohman. Gifted Child Quarterly , 49 (1), 29 36. Naglieri, J. A., & Ford, D. Y. (2015). Misconceptions about the Naglieri Nonverbal Ability Test: A commentary of concerns and disagreements. Roeper Review, 37 (4), 234 240. National Association for Gifted Children. (2008). Redefining giftedness for a new century: Shifting the paradigm. Retrieved from https://www.nagc.org/sites/default/files/Position%20Statement/Redefining%20Giftedness %20for%20a%20New%20Century.pdf National Association for Gifted Children. (2016). Gifted Education in the U.S. Retrieved from National Association for Gifted Students: https://www.nagc.org/res ources publications/resources/gifted education us National Center for Education Statistics. (2016). Percentage of public school students enrolled in gifted and talented programs, by sex, race/ethnicity, and state: 2004, 2006, and 2011 12. Digest of Ed ucational Statistics. Retrieved from https://nces.ed.gov/programs/digest/d16/tables/dt16_204.90.asp National Commission on Excellence in Education. (1983). A nation at risk: T he imperative for educational reform . United States Department of Education. NCES. (2010). The Condition of Education 2010. NCES 2010 028. National Center for Education Statistics . NEA. (2017). Students affected by achievement gaps . Retrieved from NEA: National Education Association: http://www.nea.org/home/20380.htm

PAGE 105

90 Odyssey. (2016). Minority v s. minoritized: Why the n oun just d oesn' t cut i t. Retrieved from https://www.theodysseyonline.com/minority vs minoritize Office of Advanced Academics and Gifted Services. (2011) . M ulti tiered program model gifted/advanced education continuum for student success . District, Colorado . Office of Advanced Academics and Gifted Se rvices. (2018). Identification checklists . District. Colorado . Office of Assessment and Evaluation. (2018). Disaggregated student e nrollment in District , K 12, 2015 2018. District, Colorado . Office of Assessment and Evaluation (2018). Gifted i dentification d isaggregated by e thnic g roups. District, Colorado . Office of Inclusive Excellence (2018). Retrieved from District : web address not available. Olszewski Kubilius, P., & Thomson, D. L. (2010). Gifted programming for poor or minority urban students: Issues and lessons learned. Gifted Child Today , 33 (4), 58 64. Ortiz, S. O. (2007). Best practices in nondiscriminatory assessment. In A. Thomas, & J. Grimes, Best Practices in Psychology V (Vol. 2, Chapter 40, pp. 1 18). Bethesda, MD: The Charlesworth Group. Office of Elementary and Secondary Education. (2017, M ay). Programs: Jacob K. Javits gifted and talented students education program . Retrieved from U.S. Department of Education: https://www2.ed.gov/programs/javits/index.html Perry, T., Steele, C., & Hilliard, A. G. (2003). Young, gifted, and Black: Promoting high achievement among African American students . Beacon Press. Plucker, J. A., Burroughs, N., & Song, R. (2010). Mind the ( o ther) g ap! The g rowing e xcellence g ap in K 12 e ducation. Center for Evaluation and Education Policy, Indiana University . Plucker, J., & Peters, S. (2016). Excellence gaps in education . Cambridge, MA: Harvard Education P ress. Reis, S. M., & Renzulli, J. S. (2011). Intellectual giftedness. In R. J. Sternberg & S. B. Kaufman (Eds.), The Cambridge handbook of intelligence , 235 252. Cambridge, England: Cambridge University Press. Senate. (1987). Jacob K. Javits gifted an d t alented c hildren and y outh e ducation a ct, and Office of Comprehensive School Health Education Act of 1987 . S. Hrg. 100 379 (pp. 1 50). Congressional Publication.

PAGE 106

91 Stembridge, A. (2017, August 12). Office of i nclusive e xcellence, District. Retrieved from https://twitter.com/DrYemiS Stembridge, A. (2018). My reflection matters. Retrieved from http: //www.myreflectionmatters.org/author/dr adeyemi stembridge/ Sternberg, R. J., Jarvin, L., & Grigorenko, E. L. (2011). Explorations in giftedness . Cambridge, England: Cambridge University Press. Subotnik, R. F. (1993). Genius revisited: High IQ children grown up . No rwood, NJ: Ablex Publishing Corporation. Sullivan, A. L. (2011). Disproportionality in special education identification and placement of English language learners. Exceptional Children , 77 (3), 317 334. Tatum, B. D. (2013). Defining racisim : Can we talk? In M. Adams, W. J. Blumenfeld, C. Castaneda, H. Hackman, M. L. Peters, & X. Zuniga, Readings for diversity and social justice (3rd ed., pp. 65 68). New York, NY: Routeldge. Terman, L. M. (1926). Mental and physical traits of a thousand gif ted children. Genetic studies of genius . Stanford University; GG Harrap & Company: London. U.S. Department of Education. (2015, October 5). Title I, Part A Program . Retrieved from http s://www2.ed.gov/programs/titleiparta/index.html Walker, G. (1963, July 6). Englewood and the northern dilemma. The Nation, 197 , 7 10 . Warne, R. T. (2015). Test review: Cognitive Abilities Test, Form 7 (CoAT7). Journal of Psychoeducational Assessment, 33 (2), 188 192. Whiting, G. W., & Ford, D. Y. (2009). Multicultural issues: Black students and advanced placement classes: Summary, concerns, and recommendations. Gifted Child Today , 32 (1), 23 26. Worrell, F. (2014). Ethnically diverse s tudents. In J. Plucker, & C. Callahan (Eds.) , Critical Issues and Practices in Gifted Education (pp. 237 253). Waco, TX: Prufrock Press Inc.

PAGE 107

92 APPENDIX A

PAGE 108

93 APPENDIX B