Citation
Factors that influence the quantity of charter schools in Colorado school districts

Material Information

Title:
Factors that influence the quantity of charter schools in Colorado school districts
Creator:
Kafer, Krista ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (58 pages). : ;

Subjects

Subjects / Keywords:
Charter schools -- Colorado ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
In the 2012-2013 school year, 11 percent of all Colorado public school students (88,924) attended one of the state's 187 charter schools. The number of charter schools and the percentage of district students educated in charter schools vary considerably from district to district. Research in other states shows that variables related to the level of need (percentage of low income and minority students and district academic accreditation status), school choice environment (number of private schools, adjacency to districts with charter schools, and political affiliation), and district capacity (funding, enrollment, enrollment growth, and urbanicity) are associated with the number of charter schools authorized by districts. Using two statistical models, this thesis shows many of these variables correlate with greater chartering activity. Enrollment is the strongest predictor. Districts with higher student enrollments generally have higher charter enrollments and higher numbers of charter schools. Urbanicity and adjacency to a chartering district are correlated with greater chartering activity. Per pupil funding is negatively correlated and there does not appear to be a correlation between the number of charter schools and district academic achievement or political affiliation of district voters.
System Details:
System requirements: Adobe Reader.
Thesis:
Political science
General Note:
Department of Political Science
Statement of Responsibility:
by Krista Kafer.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
903217960 ( OCLC )
ocn903217960

Downloads

This item has the following downloads:


Full Text
FACTORS THAT INFLUENCE THE QUANTITY OF CHARTER SCHOOLS IN
COLORADO SCHOOL DISTRICTS
by
KRISTA KAFER
B.A., University of Colorado Denver, 1994
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Arts
Political Science
2014


2014
KRISTA KAFER
ALL RIGHTS RESERVED
11


This thesis for the Master of Arts degree by
Krista Kafer
has been approved for the
Political Science Program
by
Michael Berry, Chair
Kathryn Cheever
Dick Carpenter
July 25, 2014
m


Kafer, Krista (M.A., Political Science)
Factors that Influence the Quantity of Charter Schools in Colorado School Districts
Thesis directed by Associate Professor Michael Berry
ABSTRACT
In the 2012-2013 school year, 11 percent of all Colorado public school students
(88,924) attended one of the states 187 charter schools. The number of charter schools
and the percentage of district students educated in charter schools vary considerably from
district to district. Research in other states shows that variables related to the level of
need (percentage of low income and minority students and district academic accreditation
status), school choice environment (number of private schools, adjacency to districts with
charter schools, and political affiliation), and district capacity (funding, enrollment,
enrollment growth, and urbanicity) are associated with the number of charter schools
authorized by districts. Using two statistical models, this thesis shows many of these
variables correlate with greater chartering activity. Enrollment is the strongest predictor.
Districts with higher student enrollments generally have higher charter enrollments and
higher numbers of charter schools. Urbanicity and adjacency to a chartering district are
correlated with greater chartering activity. Per pupil funding is negatively correlated and
there does not appear to be a correlation between the number of charter schools and
district academic achievement or political affiliation of district voters.
The form and content of this abstract are approved. I recommend its publication.
Approved: Michael Berry
IV


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION........................................1
II. REVIEW 01 THE LITERATURE..........................15
III. METHODOLOGY......................................31
VI. RESULTS AND DISCUSSION............................39
REFERENCES............................................48
v


CHAPTERI
INTRODUCTION
Charter schools are public schools that are operated independently from a school
district through a charter agreement with an authorizerusually a school district, state
governmental body, or university. Overseen by a governing board of parents and
community members, charter schools have autonomy over daily operations, budgets,
teacher contracts, training, and salary determinations, academic programs, school
calendar, pedagogy, and curriculum. Like traditional public schools, charter schools must
meet state academic standards and testing requirements, be tuition-free, and adhere to
federal civil rights laws. Charter schools do not have entrance requirements. If a charter
school fails to meet the terms of its contract, the authorizer may revoke the charter and
close the school (National Alliance for Public Charter Schools, n.d.).
The concept of charter schools predates their existence by two decades. Ray
Budde first suggested the idea at a conference in 1974 and published a paper about the
concept in 1988 (Renzulli, 2005; Budde, 1988). That same year, Albert Shanker, (1988)
former president of the American Federation of Teachers, advocated the idea in a paper
published in the Peabody Journal of Education. Three years later, Minnesota adopted the
first charter school law. Today, more than 6,000 charter schools in 42 states and the
District of Columbia educate approximately two million students (National Alliance for
Public Charter Schools, n.d.).
1


Purpose of Study
Since the adoption of the first charter school law in 1991, charter school growth
across the country has been uneven. Most states have been chartering schools for
decades, but some states, like Washington State, have just begun. Eight states do not
have charter school laws. Even within states, charter school growth varies by district.
Several researchers have examined the reasons behind variations in charter school
diffusion at the state and local level. Many have found associations between charter
school diffusion and student demographic factors like race, achievement and poverty, as
well as state/district characteristics such as size of enrollment, political climate,
adjacency to states/districts with charter schools, and other factors. No such research,
however, has been done in Colorado.
As Figure 1 shows, there is a strong correlation between student population and
the number of charter schools. There are, however, outliers. The Cherry Creek School
District, for example, has fewer charter schools than its population would predict while
Denver Public Schools has more.
Figure 1: Distribution of charter schools in Colorado school districts.
Source: Colorado Department of Education Fall 2012 data.
2


Emulating studies of other states, this research examines correlations between
charter school distribution among Colorado school districts and student demographics
(race and poverty) and district factors (accreditation level, school choice environment,
adjacency to other districts with charter schools, political climate, enrollment and
enrollment growth, urbanicity, and per pupil funding).
The study measures charter school diffusion using two dependent variables: the
number of charter schools authorized by a district and the percentage of students enrolled
in district charter schools. These represent two different ways of measuring charter
school activity.
The first dependent variable is the predominant variable used in other diffusion
studies. The second variable attempts to measure the charter school sector in a district
irrespective of district size. There are vast differences in student population among
districts. Agate 300, for example, has 10 students while Jefferson County has 85,808.
While the percentage of students educated in charter schools in a district includes some
students from other districts who transfer into the district, the bulk of students are
assumed to be district students because of proximity. The purpose of using several
variables is to see if they provide similar, and thus more valid, findings.
The study employs two statistical models. The first model examines ten
independent variables using multivariate regression analysis and the second dependent
variable, district charter school students as a percentage of all district students. Private
schools, faith-based schools, and secular schools are tested separately because they are
highly correlated. The second model uses Negative Binomial Regression to examine
3


associations between 10 independent variables and the number of charter schools
(dependent variable).
The utility of conducting quantitative analysis is that it paves the way for
qualitative research. Having singled out more easily measurable variables associated
with charter school diffusion, one can then undertake research into variables that are less
easily measured such as the qualities of superintendents and boards, the presence or
absence of policy innovators, and the strength of advocates and opponents. Researchers
have suggested that these variables have an impact on charter diffusion in other states.
The quantitative research undertaken in this thesis lays a foundation for such qualitative
research in Colorado.
Charter Schools in Colorado
History
Colorado became the third state to adopt a charter school law when Colorado
Governor Roy Romer signed the Charter Schools Act in 1993 (Benigno, 2013). In the
2012-2013 school year, 11 percent of all Colorado public school students (88,924)
attended one of the states 187 charter schools (Colorado League of Charter Schools,
n.d.).
In addition to charter schools, Colorado students may enroll in any public school
within or outside of their district, or attend one of the many district or statewide online
schools. Open enrollment was first introduced in Colorado when the legislature passed
the Public Schools of Choice Act of 1990 to enable students to attend a public school
outside of their attendance boundaries within or outside of their district (Mintrom, 2000).
Today, 9 percent of Colorado students attended public schools outside of their district
4


compared to 8 percent in 2011 and 3 percent in 2001 (Mitchell, 2011 and Colorado
Department of Education, 2013).
Colorado also has public option schools and magnet schools which are district-run
schools of choice. Such schools generally have a unique pedagogical approach to other
schools in the district. When parents choose a school other than their neighborhood
school, be it an option school, charter school, magnet school, online school, or school in
another neighborhood attendance zone, they must submit an application. Enrollment is
subject to space availability.
Private schools are also an education option. There are 277 private schools listed
in the Colorado Department of Education database (Colorado Department of Education,
n.d.). One hundred and sixty-nine are faith-based and the remainder is secular private
schools.
Table 1: School Type Definitions
Type of School Definition
Traditional Public Schools District-operated public schools that serve a designated neighborhood. Under Colorado law, students can attend, space permitting, traditional public schools other than their assigned school.
Charter Schools Independent public school. Students can attend, space permitting, any charter school.
Option and Magnet Schools District-operated public school of choice. Students can attend, space permitting, any option or magnet school.
Private Schools Independent secular and faith-based schools. In general, these schools do not receive public funding. However, Douglas County has initiated a scholarship program. The program is currently suspended pending a Colorado Supreme Court decision.
5


Student enrollment in public schools in Colorado has grown every year since
1990 (Torres, 2014). In 1993, the year Governor Romer signed the Charter Schools Act,
the state had 625,000 K-12 students. Today, there are nearly 877,000 students which
marks a 40 percent increase.
Between 2008 and 2013, 100 Colorado districts lost enrollment, five districts
maintained the same level of enrollment, and 73 districts gained enrollment. The level of
change ranges from a 79 percent drop in enrollment in the Agate 300 district to a 45
percent increase in the Mapleton 1 school district (Colorado Department of Education,
n.d.). In general rural districts declined in enrollment while Front Range communities
gained enrollment.
Student Demographics in Colorado Traditional and Charter Public Schools
As Figure 2 demonstrates, Colorado charter schools currently enroll the same
percentage of minority students (44 percent) as do traditional public schools (44 percent)
with charter schools enrolling slightly more Black and Asian students and traditional
public schools enrolling slightly more Hispanic students (Colorado League of Charter
Schools, n.d.). This was not always the case; in 2001, students of color were slightly less
likely to enroll in charter schools than in traditional public schools. Twenty-seven
percent of charter school enrollment was composed of racial minority students as
compared to 33 percent in traditional public schools (Carpenter, and Kafer, 2013).
6


Figure 2: Racial Demographics at traditional and public charter schools.
Source: Colorado League of Charter Schools 2012-2013 data.
In contrast to national trends, traditional public schools in Colorado serve a higher
percentage of low income students than charter schools (Carpenter, and Kafer 2013). The
percentage of charter school students who qualify for the federal Free and Reduced
Lunch program was 32 percent in 2012 as compared to 42 percent of traditional public
schools. The percentage of low income students in charter schools, however, has
increased significantly since 2001 when it stood at 18 percent.
The increase in the percentage of low income students served by Colorado charter
coincides with the increase urban charter schools and charter school networks seeking to
serve disadvantaged students. Charter school founders, often with the support of
philanthropists, have responded to demand by opening charter schools to serve urban low
income and minority students and by offering support such as tutoring, a longer school
day, Saturday hours, smaller class sizes, access to social services, and home visits.
Several successful charter operators such as KIPP, Denver School for Science and
Technology and STRIVE Preparatory Schools have multiple inner-city campuses across
the Denver Metro Area.
7


Achievement trends among minority and low income students in Colorado are
similar to national trends. Low income students and Black and Hispanic students have
lower proficiency levels in math and reading than more affluent students and White
students. Poor Black, White, and Hispanic students in charter schools generally
performed better than their peers in traditional public schools on reading and math state
assessments (Carpenter, and Kafer, 2013). More affluent Black and Asian students in
charter schools also achieved better outcomes than their peers in traditional public
schools in reading while more affluent Hispanic students in traditional public schools
generally performed better than their peers in charter schools. Findings were mixed for
White students. In math, more affluent Black students in charter schools performed better
than their peers in traditional public schools while findings for Hispanic, Asian, and
White students were mixed by grade level.
Charter School Diffusion
Under Colorado law, charter schools may be authorized by school districts or the
Colorado Charter School Institute (CSI). CSI may authorize charter schools in school
districts that do not have exclusive chartering authority given to them by the state board
of education. The law grants districts that enroll fewer than 3,000 students exclusive
chartering authority automatically and requires larger districts to apply for the privilege
(Conlan, 2011). The state board can revoke exclusive chartering authority if the district
fails to uphold the Charter Schools Act. In the 2012-2013 school year, nine districts
lacked exclusive chartering authority (Colorado Department of Education, n.d.). CSI also
charters schools in districts with exclusive chartering authority with the permission of the
district. In the 2012-2013, CSI held charter agreements with 28 schools (Colorado
8


Charter School Institute, n.d.). Because CSI is not a district but a state-wide authorizer
whose sole purpose is to charter schools, its schools are not included in this study. This
study focuses exclusively on districts which may or may not authorize charter schools.
Forty-four of the states 178 school districts have chartered schools. As Figure 3
shows, charter schools exist throughout the state but are more common along the
urbanized Front Range where the population density is higher. The percentage of
suburban and rural charter schools in Colorado is higher than the national average
(National Association of Public Charter Schools, n.d.). Thirty-seven percent of Colorado
charter schools are located in urban environments as compared to 52 percent nationwide.
Twenty-six percent of Colorado charter schools are located in the suburbs. Only nine
states have a higher percentage of suburban charter schools. Thirty percent of Colorados
charter school students attend a rural charter school (Stuit, and Doan, 2012). Only nine
other states have an equal or greater percentage of charter school students in rural
schools.
Figure 3: Distribution of Charter Schools in Colorado by City
Source: Colorado League of Charter Schools (n.d.). Facts and figures about Colorado
charter schools. Copied with permission.
9


The Colorado Department of Education rates each school district according to its
urban/rural setting on a scale of one to five: 1) Denver Metro; 2) Urban/Suburban; 3)
Outlying City; 4) Outlying Town; and 5) Rural (Colorado Department of Education,
n.d.). One hundred and five districts are either Denver Metro (19) or part of another
urban/suburban area (86). The remainder is part of an outlying city (51), outlying town
(18), or rural area (4). Figure 4 shows the relationship between urbanicity and the number
of charter schools. Urban and suburban districts are more likely to have charter schools.
There some interesting outliers, however; three urban districts, Adams County 14,
Englewood 1 and Sheridan do not have charter schools while five rural districts,
Elizabeth C-l, Keenesburg RE-3J, Strasburg 31J, and West End RE-2 have charter
schools. On the other end of scale, Cherry Creek has only one charter school and Adams
County 14, Sheridan and Englewood have no charter schools even though they all have a
Denver Metro setting designation.
40 35 30 o 0 -5 25 V) 1 20 (0 U 15 0 £ 10 .Q 1 5 Z n






t f
< t t t
0 1 2 3 4 5 District setting
Figure 4, Distribution of charter schools in Colorado school districts by district setting.
Source: Colorado Department of Education Fall 2012 data.
10


While the urban-rural difference in chartering rates is clear, the pattern of
distribution of charter schools within urban and suburban school districts is not as easily
apparent. Table 2 shows the top 10 districts by student population along with the number
of charter schools. The table shows that the size of the district in terms of student
enrollment is not a completely reliable predictor of charter school presence. For
example, Colorado Springs 11 ranks ninth in student population but has the fourth highest
number of charter schools. Denver Public Schools, the second largest district by
enrollment, has more than twice as many charter schools as the largest district, Jeffco
Public Schools.
Table 2: Top 10 Colorado school districts by student enrollment and number of charter
schools
School District Total Enrollment Number of Charter Schools
1. Jeffco Public Schools 85,508 14
2. Denver Public Schools 83,377 37
3. Douffllas Countf School District 64,657 11 1
4. Cherry Creek Public Schools 53,368 1
5. Adams 12 Five Star Schools 43,268 6 I
6. Aurora Public Schools 39,835 6
7. Boulder Valley Public Schools 30,041 5 1
8. St Vrain Valley School District 29,382 6
9. CO Svrinfl School District 11 28,993 7 1
10. Poudre School District 27,909 4
Source: Colorado Department of Education Fall 2012 data.
When viewed as a scatterplot in Figure 5, it appears that student enrollment alone
does not predict the number of charter schools.
11


90.000 80.000 £ 70,000 £ 60,000 2 50,000 Z 40,000 -g 30,000 3 P( 55 20,000 rt 10,000 n
Jefferson Denver

Douglas
Cherrv Creek
a,imn t Adams 12
flU 1 ur u ^ jjJSoulder
ludre^ctTLk0 Springs 11


0 10 20 30 40 Number of charter schools
Figure 5, Distribution of charter schools in top 10 largest Colorado school districts.
Source: Colorado Department of Education Fall 2012 data.
Ranking districts of different sizes (according to the number of students) by the
number of charter schools provides an interesting comparison but it does not take into
account the impact of size. One would expect larger districts to have more charter schools
because they have more students thus more demand and more per-pupil funds. A smaller
district may have a large proportion of students in charter schools but fewer charter
schools overall simply by virtue of its size. By viewing district chartering activity as a
ratio of students per charter school, one can compare charter activity among districts of
different student enrollment sizes.
Table 2, provides just such a comparison. The fourth largest school district,
Cherry Creek Public Schools, has 53,368 students per charter school, while Colorado
Springs 11, the ninth largest district by population, has 4142 students per charter school.
Jeffco Public Schools has 35,000 more students than the Boulder Valley School District
but a similar sized charter schools sector.
12


Table 3: Top 10 CO school districts by ratio of students per charter school
School District District Students per Charter School Percent of District Students Enrolled in Charter Schools
1. Denver Public Schools 2253/1 14%
2. CO Springs School District 11 4142/1 8%
3. St Vrain Valley School District 4897/1 11%
4. Douglas County School District 5878/1 15%
5. Boulder Valley Public Schools 6008/1 8%
6. Jeffco Public Schools 6108/1 8%
7. Aurora Public Schools 6639/1 10%
8. Poudre School District 6977/1 6%
9. Adams 12 Five Star Schools 7211/1 20%
10. Cherry Creek Public Schools 53,368/1 Less than 1%
Source: Colorado Department of Education Fall 2012 data
Another way to look at the size of the size of a school districts charter school
sector is to examine the percentage of students that are enrolled in charter schools in a
district. Table 3 also shows the percentage of students educated in charter schools in
each district. While some of the students educated in the districts charter schools are not
from the district, the same is true for the districts traditional public schools. Under
Colorado law, students may transfer to other district charter and traditional public
schools, space permitting.
In the top ten largest school districts, as few as 1 percent and as many as 20
percent of students educated in a district are enrolled in a charter school. Adams 12, the
fifth largest school by total district population, enrolls the highest percentage of students
in charter schools. Though not in the top 10 districts by enrollment, Brighton 27, Falcon
49, and Greeley school districts have a higher proportion students enrolled in charter
schools21 percent, 17 percent, and 17 percent respectivelythan nine of the 10 largest
13


school districts. Several rural districts have a high proportion of students enrolled charter
schools: Park County (25 percent), West End (15 percent), Elizabeth (17 percent) and
Clear Creek (13 percent).
While there is a positive correlation between district size and charter school
presence, there are other factors that likely exert an influence on the prevalence of charter
schools. As the following analysis will show, student demographics (race and poverty)
and district factors (accreditation level, school choice environment, adjacency to other
districts with charter schools, political climate, enrollment and enrollment growth,
urbanicity, and per pupil funding) impact the degree of charter school diffusion.
14


CHAPTER II
REVIEW OF THE LITERATURE
Most of the researchers found positive associations between charter diffusion and
non-white students, low district achievement, adjacency to districts with charter schools,
and district student density. Some researchers found correlations between charter school
diffusion and enrollment of low income students, percentage of district private schools,
and political climate. Others, however, did not. Table 4 provides a summary of the
current state of knowledge in this field. The literature review that follows provides a
discussion of most of these variables studies have associated with charter school
diffusion.
15


Table 4: Summary Table of Literature Review
Vari- Percentage Percentage District Percentage Adjacency Political District Other
able of Minority of Low Academic of Private Factors Capacity Variables
Students Income Students Quality Schools
Renzulli and Renzulli Renzulli Renzulli Rincke Renzulli Renzullis Renzulli
Roscigno (2005) to (2002) +to and (2007) +to (2002)+ (2005) + (2005) +
(2005) +to charter application Roscigno district between between special
number of application submissions. (2005)+to charter Democratic urbanicity education
charters in submissions. a point, adoption. registration and charter students and
state then and charter school charter
negative. application submissions. application
submissions. submissions
Renzulli Rincke Rincke Renzulli Witte, Renzulli and Witte, Rincke
(2005) +to (2007) - (2007) + to (2005) + Schlomer, Roscigno Schlomer, (2007) +
charter with district district secular and Shober (2005) - and Shober with magnet
application charter charter schools (2007) +to between (2007) + schools and
submissions. adoption adoption. only. district Republican between charter
charter governor district school
school and charter enrollment adoption.
adoption. submissions. and charter schools.
Rincke Witte, Minstrom Wong, Zhang and Shober, Renzulli Renzulli and
(2007) - Schlomer, (1997) + K.K. and Yang Manna, and (2005) + Roscigno
with district and Shober between Langevin (2008) + to Witte (2006) between (2005) +
charter (2007) +to decrease in (2007) +to charter + correlation district size between
adoption. district test scores charter law schools between and charter open
Hispanic adoption. and charter adoption. openings state charter application enrollment
only. law schools and submissions. law and
adoption. Republican charters in
governor or leMlature. state.
Witte. Stoddard Stoddard Zhang and Wong and Witte, Minstrom
Schlomer, and and Yangs Langevine Schlomer (1997)-
and Shober Corcoran Corcoran (2008) - (2007) + and Shober between
(2007) + to (2006) + (2006) + correlation between (2007) + union
charter between states with with Republican between strength and
school higher low charter governor district state
openings dropout achievement school and chartering adoption of
rates and and higher creation. adopting a and federal charter law.
charter charter charter law. funds.
school enrollments in districts. enrollments.
Wong, K.K. Wong and Zhang and Wong, and Minstrom
and Shen (2002) Yang (2008) Langevin (1997)+
Langevin + between + (2007)+ between
(2007) + to state dropout Democratic between presence of
state rates and registration lower state policy
adoption of charter law and district funding and entrepreneurs
charter law adoption. chartering to law and policy
a point, then negative. adoption. adoption.
Wong, K.K. Wong and Zhang and Wong, and Minstrom
and Shen (2002) Yang (2008) Langevin (1997) +
Langevin + between + (2007) + between
(2007) + to state dropout Democratic between presence of
state rates and registration lower state policy
adoption of charter law and district funding and entrepreneurs
charter law adoption. chartering to law and policy
a point, then negative. adoption. adoption.
16


Variable Percentage Percentage District Percentage Political
of of Low Academic of Private Adjacency Factors
Minority Income Quality Schools
Students Students
District Other
Capacity Variables
Wong, Wong and Zhang and Wong, and Minstrom
K.K. and Shen Yang Langevin (1997) +
Langevin (2002) + (2008) + (2007) + between
(2007) + to between Democratic between presence of
state state registration lower state policy
adoption of dropout and district funding and entrepreneurs
charter law rates and chartering law and policy
charter law to a point, adoption. adoption.
adoption. then negative.
Zhang and Zhang and Renzulli Renzulli and
Yang Yang (2002) + Roscigno
(2008) +to (2008)-to between (2005) -
charter charter lower between union
schools school funding and strength and
openings. openings. higher state adoption
Black charter of charter law.
students only. submissions
Stoddard Zhang and Zhang and
and Yang Yangs (2008)
Corcoran (2008) to + between
(2006) +to charter appointed
charter schools superintendent
schools openings. and chartering
openings. Black in districts.
students onlV.
Renzulli (2005)+ between number of district administrators and charter submissions.
Overall Mostly Mixed Positive Mixed Positive Mixed Positive N/A
positive
Researchers have examined variables that correlate with charter school diffusion
across the country in terms of passage of charter school enabling laws and charter school
openings (or charter school application submissions). Although this thesis focuses on
variables that influence charter diffusion in districts, findings of both types of existing
diffusion research are reviewed here for two reasons: There are few studies that analyze
17


factors that influence district chartering rates so it is useful to broaden the analysis to
include other diffusion analysis. Secondly, both types of research examine similar
variables. The main factors that influence charter school adoption at the state and district
level are related to student demographics, district school quality, political environment,
district size, proximity to other districts or states with charter schools, union strength, and
revenue. The subheadings below provide a discussion of extant research as it applies to
each of these characteristics.
Disadvantaged Students
There is good reason to believe that minorities are attracted to school-choice
options precisely because they have been so disadvantaged in the public education
system observed Renzulli and Roscigno in their 2005 study (350). Students of color and
low income students have traditionally performed at lower academic levels than White
students and more affluent students (Barton, 2004). For these traditionally underserved
populations, charter schools have become a popular alternative education environment.
Nationwide charter schools enroll, on average, a greater percentage of Black and Latino
students (27 percent and 26 percent, respectively) than traditional public schools (15
percent and 22 respectively) (National Alliance of Public Charter Schools, n.d.). Charter
schools also enroll a higher percentage of low income students (53 percent) than
traditional public schools (47 percent) (Lake, 2012).
Although there is considerable diversity within the charter school sector, several
studies suggest that charter schools, in general, produce positive academic impacts for
disadvantaged student populations. A 2011 national randomized study of the effect of
18


attending a charter school on academic progress by Mathematica, a policy research
organization, found positive impacts for more disadvantaged schools and students and
negative impacts for the more advantaged (Clark, M.A. et al., 1). A more recent study
by the Center for Research on Education Outcomes at Stanford University showed that
attending a charter school had positive impacts for Hispanic students who are English
language learners and Black students in poverty (Raymond et al. 2013). Hispanic
English-language learners gained the equivalent of 50 additional days of learning in
reading and 43 additional days in math over their peers in traditional public schools.
Poor, Black students gained 29 additional days in reading and 36 additional days of
learning in math over their peers in traditional public schools. More affluent Hispanics
experienced similar achievement to their traditional public school peers while White and
Asian students lost ground.
Several researchers have tested the hypothesis of whether a higher presence of
minority, low income, or learning challenged students is associated with a greater
likelihood of a state passing a charter school law or of a district chartering schools.
Minority Students
Renzulli and Roscignos (2005) found that as the percentage of nonwhite students
increased so did the number of charter schools within a state. In similar state-level
adoption research, Wong, and Langevin, (2007) demonstrated that the percentage of
minority students has positive impact on state adoption of charter school legislation.
At the district level, Renzullis 2005 study found the percentage of nonwhite and
special education students to have a positive correlation with the number of charter
19


school application submissions. Witte, Schlomer, and Shober (2007) also hypothesized
that districts with more nonwhite students would open charter schools. Their analysis
demonstrated a positive correlation with the number of charter schools in districts in
Wisconsin. Zhang and Yang (2008) found a greater percentage of Black students to be
positively associated with more charter schools in Florida districts. Stoddard and
Corcorans 2006 research also showed districts with high or increasing percentage of
Black students, a high or increasing percentage of college graduates and growing income
inequality had larger charter school enrollments than did more homogeneous districts.
Rincke (2007), however, found greater enrollment of Hispanic students to be negatively
associated with charter adoption in California districts. Rincke acknowledged that this
finding conflicted with the claim asserted by charter schools advocates that charter
schools benefit disadvantaged children. He asserts that the more favorable the social
conditions under which local public school producers operate, the more likely is the
establishment of additional charter schools (538). This may have been true during the
late 1990s and early 2000s from which Rinckes data are drawn. As discussed earlier, the
proportion of disadvantaged students served by Colorado charter schools in the early
years was lower than it is now.
Low Income Students
A number of existing studies have found that increases in low income student
population exert a suppressing effect on charter school establishment (Rincke, 2007;
Zhang and Yang 2008; and Renzulli, 2005). Zhang and Yang (2008) also examined the
impact of the percentage of learning disabled students and found no impact. Witte,
Schlomer, and Shober (2007), however, found districts with charter schools to have
20


higher average low income student enrollments. It is not clear why the findings are
contradictory. All use data from the late 1990s and early 2000s. Witte, Schlomer, and
Shobers analysis is on Wisconsin schools, Rinckes California, Renzullis North
Carolina, and Zhang and Yangs Florida. Perhaps state differences explain the variance in
findings. Overall, researchers agree that the percentage of non-white students is
associated with higher charter activity.
District Academic Quality
The presence of students in poverty and students of color in a district is not the
only potential predictor of chartering activity. Students in low-performing districts
regardless of ethnicity or poverty are likely to desire higher quality schooling options.
Testing this hypothesis between district school quality and chartering activity, Renzulli
(2002) found that a greater proportion of low performing schools in a district increased
the mean number of charter school application submissions. Similarly, Rinckes (2007)
research into charter school diffusion in California found that districts with low
achievement were more likely to establish charter schools. Zhang and Yang (2008),
however, found that the percentage of failing schools in a district had a negative
correlation with charter school openings in Florida. Zhang and Yang had hypothesized
that failing schools would lead to more charter schools but concluded that improving
educational performance may be a major concern for potential charter founders but not be
the guiding principle for school boards and local politics (583). The authors
acknowledge that their findings are different from Renzulli but offer no explanation as to
the reason.
21


In research considering state charter laws, Minstrom (1997) found that decreasing
test score averages increased the likelihood of the passage of a charter law. Analysis by
Stoddard and Corcoran (2006) found states with low student achievement and districts
with higher dropout rates had higher charter school enrollments. Wong and Shen (2002)
found that a states adoption of a charter law was inversely related to its graduation rate,
that is, states with lower rates were more likely to adopt charter laws. These studies
suggest that charter schools may provide one avenue to improve student learning when
test scores indicate decreases in student performance.
School Choice Environment
The school choice environment refers to the demand for and availability of
education options in the district and adjacent districts, and the political climate. School
choice, that is, the ability of parents to choose from among public or private options with
the support of public funds is not a new concept in Colorado or the nation. In the
colonial era and early republic, student were educated through a variety of independent
schools financed by local communities, churches, and charitable organizations (Jeynes,
2003; 2007). Although tax supported public schools became the norm in the late 19th
century and early 20th Century (Carpenter and Kafer 2012), the desire for alternatives
resumed in the mid and late 20th century. In 1955 Minnesota adopted the first tax credit
for private school tuition. Today there are 23 tuition tax credit programs exist in 15 states
(Friedman Foundation for Educational Choice n.d.). In 1990, the first modem voucher
program was enacted (Vermont and Maines century and a half old rural voucher
programs aside). Today there are 22 programs in 12 states and the District of Columbia.
22


Presence of Private Schools
One way to measure the demand for school choice is the number of private
schools in a district. According to research by Schaeffer (2012), approximately 8 percent
of charter elementary students and 11 percent of middle and high school students come
from the private sector. In other words, they left their private school to attend a public
charter school. In urban centers, private school students constitute substantial portions of
charter elementary (32 percent), middle (23 percent) and high school school (15 percent)
students come from private schools. Buddins (2012) research and that of Chakrabarti
and Roy (2011) also suggest that charter schools pull a significant number of students
from private schools. Taken together, these results indicate that private schools may both
bolster school choice legitimacy and demand and serve as an alternative source for
prospective students.
A study by Renzulli and Roscigno (2005) showed that the number of private
schools increased the number of charter schools operating in a state. They suggest that
the presence of competition, in the form of private schools, increases the number of
charter schools that operate in the state (358). They also point out that correlation waned
in states with the highest level of private schools, indicating that there may be a saturation
point for education alternatives.
Another of Renzullis studies (2005) showed that the number of secular private
schools, but not faith-based schools, had a positive impact on the number of charter
school application submissions in school districts. She suggests that education
alternatives increase the legitimacy of education choice. Wong, K.K. and Langevin
23


(2007) found the number of private schools to positively correlate with the likelihood that
a state would adopt charter school legislation. Zhang and Yangs 2008 research, however,
found the number of private schools to be negatively correlated with charter school
creation. They suggest that charter schools are substitutes of private to some degree
(585).
Presence of Public School Options
The presence of other public school options appears to increase the likelihood of
charter openings. A study examining showed the presence of magnet schools is a
positive predictor for chartering activity (Rinckes 2007). Magnet schools are district-run
schools of choice which are generally established for voluntary racial integration
purposes. According to Public School Review, there are 24 magnet schools in Colorado
(n.d.). Renzulli and Roscigno (2005) found that the passage of a statewide open
enrollment law increased the likelihood of chartering.
Even the presence of other charter schools appears to predict that more will open.
Renzulli and Roscigno (2005) suggest that there is some degree of path dependence in
this area, [OJpen charter schools and the increased number of states with charter school
laws may increase the legitimacy for new charter legislation and the creation of more
charter schools (348). Renzulli (2005), however, found that the number of existing
charter schools in a district had a negative impact on submissions. This suggests that as
the number of charter schools increases, competition for a finite number of students also
increases. The number of charter schools in the state, on the other hand, had a positive
impact, suggesting that existing charter schools in other districts increases the legitimacy
24


of charter schools without reducing the capacity within the district for more charter
schools. As with private schools, the presence of nearby charter schools appears to raise
interest and demand for charter schools.
Adjacency to Districts with Charter Schools
As previously mentioned, students of today are more likely to attend schools
outside of their residential district than they had been in the past. Because districts
lacking charter schools may see enrollments decline as students choose to attend charter
schools in nearby districts, there exists an incentive to offer their own charter schools. As
parents and district boards become aware of nearby districts charter schools, it seems
likely that board members would experience pressure to authorize charter schools within
the district.
Witte, Scholomer, and Shobers (2007) research in Wisconsin shows a positive
correlation between proximity to a district with charter schools and the number of charter
schools. Similar diffusion effects have also been found in in California (Rincke 2007)
and Florids (Zang and Yang 2008) Further research on inter-district school choice and
charter school diffusion conclude that policy makers are significantly influenced by their
peers actions in nearby districts (Rincke, 2006, 2007).
Political Climate
According to DeBray-Pelot, Lubienski, and Scott (2007), charter schools
represent the marriage of market-oriented neoliberals working from a series of state-
level think tanks and progressive reformers committed to creating options with a public
25


system (p. 212). Although Republicans tend to be more open to school choice; charter
schools enjoy support from both sides of the aisle (Kirst 2007).
The research on the impact of political climate on charter diffusion is mixed.
Renzulli (2002) hypothesized that a higher presence of Democrats would correlate with
more chartering activity. She found that an increase in registered Democrats increased the
number of submissions, but that other factors were more determinative. A follow-up
study found that state level political factors had no impact on whether a state adopted a
charter school law but having a Republican governor decreased the expected number of
charter school foundings (Renzulli and Roscigno 2005, 358).
Shober, Manna, and Witte (2006) found that having large Republican
representation in the state legislature was correlated with more charter schools in a state
and a Republican governor even more so. Wong and Langevine (2007) found that states
with a Republican governor were more likely to adopt a charter school law.
Zhang and Yang (2008) found that the percentage of Democratic voters was
correlated with an increase in the number of charters to a point. When the percentage of
Democrats exceeded 79.5 percent, the effect was reversed. This would suggest that
charter schools enjoy bipartisan support except in heavily Democratic districts.
Union opposition to charter schools is another variable identified by researchers
that impacts both charter law adoption and charter activity. Minstrom (1997) found union
strength reduced the likelihood of charter law approval as did Renzulli and Roscigno
(2005). However, they also found that the National Education Associations presence had
26


a marginally positive impact on the founding of charter schools once a law had passed.
Stoddard and Corcoran (2006) had the same findings.
Research by the Thomas B. Fordham Institute ranked the strength of Colorados
teacher union 35th among the states and the District of Columbia (Winkler, Scull, and
Zeehandelaar 2012). According to the report, union strength is not very strong in
Colorado. Because state level union strength is weak and district level union information
would be difficult to obtain, this variable will not be analyzed in the thesis.
School District Capacity
School district capacity refers to several district level factors including population
density, presence of policy innovators, the superintendent, enrollment and enrollment
growth, and funding.
Population Density
Renzullis (2005) research showed that urban districts had more charter school
submissions than suburban or rural districts. Total population and population density both
impact the capacity of a district to open a charter school. The presence of advocacy
groups will likely be stronger in urban and suburban areas. Coalitions that generally
support the opening of charter schools include parents dissatisfied by local schools,
business and community organizations, state charter school associations, national
advocacy groups with local affiliates, real estate developers, faith-based organizations,
higher education institutions, and foundations and philanthropists (Kirst 2007).
Presence of Innovators
27


Minstroms (1997) research found that the presence of policy entrepreneurs
significantly increases the rate at which the public approves of school choice as a policy
innovation. These groups work with policy makers to effect change (Minstrom and
Vergari, 1998). He notes that [PJolicy entrepreneurs operating at the state level will
most often develop their ideas for policy innovation through their conversations and
interactions with members of interstate and external policy networks'" (1997, p.130,
italics in the original).
The Superintendent
In their case studies, Teske et al. (2000) found mixed support for charter school
initiatives from school superintendents. Some exhibited antagonistic relationships with
charter schools, while others were much more positive. Even in districts where district
schools faced considerable competition from charter schools, some superintendents had
positive views of charter schools. Teske et al. concluded that [T]he attitude of the
district superintendent and, through the superintendent, the attitudes of other high level
administrators seem to be more a function of their individual beliefs rather than the
market share of the charter or district schools (p. 10). In all 19 case studies conducted by
Witte, Scholomer and Shober (2007), there were entrepreneurial administrators, school
board members, parents, and teachers who wanted to open charter schools.
Zhang and Yangs (2008) research showed that the presence of an appointed
superintendent versus an elected superintendent was positively with higher rates of
chartering. They suggest that this is because appointed superintendents are less
distracted by electoral politics in the community and more motivated by a desire to
28


enhance their reputation and labor-market value by adopting innovations. Alternatively, it
may indicate that appointed superintendents treat charter schools as a dumping ground for
their at-risk students so that they can boost test schools in current schools (585). All
district superintendents are appointed in Colorado.
Enrollment and Enrollment Growth
High enrollment growth could lead school districts to open charter schools as a
means of increasing district capacity to absorb new students. Districts in suburban areas
experiencing high growth in student enrollment become receptive to charter schools to
address overcrowding (Pushpam, 2002). A study by RPI International for U.S.
Department of Education examined the impact of charter schools in 49 districts in five
states -Arizona, California, Colorado, Massachusetts, and Michigan (Ericson, J. et al.,
2001). The researchers found, In 35 percent of the total districts, administrators reported
that charter schools had relieved overcrowding or the pressure to construct new facilities
caused by an increasing student population (p.14).
The research on enrollment growth and charter activity is lean. Zhang and Yang,
however, found no association between enrollment growth and chartering (2008). Higher
enrollment, in general though, has been shown to correlate with higher chartering
activity. Witte, Schlomer, and Shober (2007) found that districts with higher enrollments
were more likely to charter schools. They suggest that larger districts have larger
administrative infrastructure to support the chartering process.
Renzulli (2005) found that the larger the size of the district and the higher the
number of district administrators the higher the charter school submissions. She
29


hypothesizes that high levels of bureaucracy increase inefficiencies and red tape and will
therefore increase the number of submissions.
Funding
Witte, Schlomer and Shober (2007) found that districts with more federal
revenues were more likely to start chartering schools and attributed it to the fact that
these districts have more low income students. In their analysis of factors related to
charter law adoption, Wong and Langevin (2007) found lower classroom spending
associated with charter law adoption. Renzulli (2002) also found that districts with lower
state and local funding had higher charter school submissions.
30


CHAPTER III
METHODOLOGY
This study will test for relationships between the size of the charter school sector
and ten variables associated with student demographics and district factors. Although the
number charter schools open in the 2013-2014 is available, academic data are available
only for the 2012-2013 school year. Therefore all data used in the study come from the
2012-2013 school year.
Hypotheses
Level of Need
One of the primary reasons for founding charter schools according to advocates is
that they can serve disadvantaged children who are not well served in the traditional
system. The Colorado Charter School Acts of 1993 states:
(2) The general assembly further finds and declares that this part 1 is enacted for the
following purposes:
a) To improve pupil learning by creating schools with high,
rigorous standards for pupil performance; (b) To increase learning
opportunities for all pupils, with special emphasis on expanded
learning experiences for pupils who are identified as academically
low achieving; (Colorado Charter Schools Act (1993).
31


For the purposes of this study, level of need is defined as the percentage of
disadvantaged students (minority and low income students) and student achievement as
measured by district accreditation status.
H]: School districts with a higher percentage of nonwhite students will have a
greater quantity of charter schools.1 2 3
H2: School districts with a higher percentage of low income students will have a
greater quantity of charter schools."
H3: School districts with a lower accreditation plan will have a greater quantity of
charter schools.
School Choice Environment
School choice environment refers to the level of school choice within the district.
All districts are impacted by Colorados open enrollment law. Some districts have private
schools. As the researchers noted, private schools act as competition for public schools
and as a pool of potential student recruits. Parents who enroll their children in private
1 These data are found at the CDE website at www.cde.state.co.us/cdereval/pupilcurrentdistrict.htm Data
set for minority students from CDE that included all nonwhite students. Percentages have been
transformed into decimals for data processing.
2
These data are found at the CDE website at www.cde.state.co.us/cdereval/pupilcurrentdistrict.htm
This CDE data includes students who are eligible for the Federal Free and Reduced Lunch Program. In
2012, free lunch eligibility was set was 130 percent of the federal poverty level (approximately $29,965 for
a family of four) and at 185 percent of the federal poverty level (approximately $42,643 for a family of
four) for a reduced price lunch. Percentages have been transformed into decimals for data processing.
3
The state of Colorado assigns districts an accreditation category based on the overall District Performance
Framework score that includes academic achievement, academic growth, academic growth gaps, and
postsecondary and workforce readiness. Accreditation categories are as follows: Accredited with
Distinction, Accredited, Accredited with Improvement Plan, Accredited with Priority Improvement Plan,
and Accredited with Turnaround Plan. To simplify, districts with accreditation status of Accredited with
Distinction, Accredited, Accredited with Improvement Plan were coded as 0 and those with Priority
Improvement Plan, and Accredited with Turnaround Plan were coded as a 1.
32


schools are accustomed to choosing schools. Some of the research shows that high
number of private schools is associated with a greater number of charter schools.
District adjacency to other districts with charter schools is also included. If a
sufficient number of parents leave the district to attend a charter school in another
district, district boards will feel pressure to charter schools of their own.
Political climate is also included in this category as a null hypothesis. Colorado
history shows bipartisan support for charter schools so it is expected to have no impact on
charter diffusion. The weakness in using countywide voter registration statistics is that
they do not capture political diversity within each county. For example, political
registration in the Manitou Springs school district and the Colorado Springs 11 school
district located near Peterson Air Force Base are likely different. Both districts are
assigned 22 percent Democratic registration because that is the county level. Similarly,
there is considerable diversity in Jefferson and Arapahoe Counties, as well, that is not
reflected in the countywide percentages. Finally, the rising number of unaffiliated voters,
dilutes the usefulness of this variable. Independents, which make up about a third of
voters in Jefferson and Arapahoe Counties vote for Democrats, Republicans, and other
parties. Other researchers may consider using more complicated variable to better
capture political culture but none has yet embarked down that road.
H4: School districts with a higher number of private (independent and faith-based)
schools will have a greater quantity of charter schools.4
4 Data for independent and faith-based schools are found at the CDE website at
www.cde.state.co.us/cdereval/pupilcurrentnonpublic.htm. The data set includes the total number of private
schools. Because research variable on the whether the number of total private schools, number of secular
33


H5: School districts adjacent to districts with charter schools are more likely to
have charter schools.5
He: The Percentage of Democratic registered voters will have no impact on the
quantity of charter schools in school districts.6 7
District Capacity
District capacity refers to the capacity to open additional schools. Districts with
higher student enrollments can authorize more schools. Such district will have both the
demand and the capacity to operate additional schools. Districts experiencing enrollment
increases over the past few years will have additional capacity. Districts in urban and
suburban areas should have more capacity to open additional schools than rural schools.
Finally, districts with lower per-pupil funding should be more likely to charter schools as
the research findings predict.
H7: School districts with higher levels of enrollment growth will have a greater
n
quantity of charter schools.
H8: School districts with higher levels of enrollment will have a greater quantity of
o
charter schools.
private schools, or number of secular private correlated to chartering activity the study also tests this
hypothesis for secular and faith-based schools.
5 These data are found at the CDE website at
http://www.cde.state.co.us/cdeedserv/coloradoschooldistrictsmap. I identified charter districts on the map
and noted which non-charter districts were next to charter districts. Districts adjacent to charter districts are
coded as a 1 and districts that do not touch another district with charter schools are coded as 0.
6 These data are found at Colorado Secretary of States office at
www.sos.state.co.us/pubs/elections/VoterRegNumbers/2013/June/VotersByPartyStatus.pdf. Each school
district fits within the boundaries of a county. Most counties have more than one district usually between
two and six. The average is 2.7. Outliers El Paso County and Weld County have 15 and 12 twelve school
districts respectively. Denver, Douglas, and Jefferson, three of largest school districts, are county-wide.
7 These data are found at Membership Trends District Totals 2006-2013 at CDE website at
http://www.cde.state.co.us/cdereval/pupilcurrentdistrict.
34


H9: School districts with a Denver Metro, Urban/Suburban, and Outlying City
designation will have a greater quantity of charter schools than Outlying Town,
and Rural school districts.8 9
Hio: School districts with lower levels of total per pupil funding will have a higher
quantity of charter schools.10
Dependent Variables
The study measures the quantity of charter schools in districts using two
dependent variables to be analyzed separately. The use of the two variables is useful for
validation of the results.
The first dependent variable is the number of charter schools in a district. This is
the most common dependent variable used in the literature.
The reason for using a second variable, that of student population, is that some
districts may appear to be enthusiastic chartering authorities by the absolute number of
charter schools, but may in fact, serve a proportionally small number of kids. Researchers
Teske, Schneider and Cassese noted in their 2005 study of school district authorizers that
many school boards view charter schools as competitors for their own schools and
attempt to limit charter school creation and growth. School boards have held a monopoly
8 These data are found at Membership Trends District Totals 2006-2013 at CDE website at
http://www.cde.state.co.us/cdereval/pupilcurrentdistrict
9 These data are at CDE website at
http://www.cde.state.co.us/sites/default/files/documents/cdereval/download/pdf/districtslistedbysetting.pdf
CDE designates each school district as one of the following: Denver Metro; Urban/Suburban; Outlying
City; Outlying Town; and Rural. To simplify, Denver Metro, Urban/Suburban, and Outlying City districts
were coded with 1 and Outlying Town; and Rural districts coded as a 0. This includes state and local
funding.
10 Data represent Total Formula Funding and are found at CDE website at
http://www.cde.state.co.us/cdefinance/dbvdfvl3. This includes state and local funding.
35


for a long time and do not wish to give it up. Boards, however, experience pressure to
respond to local political forces to open schools. The researchers hypothesize that school
boards are more likely to favor niche schools that serve special populations and do not
compete with the school districts general education schools. A district may charter a
high number of charter schools but because they are primarily niche schools, the district
actually serves a relatively low percentage of district students.
The study analyzes open charter schools instead of charter schools application
submissions. Researchers examining charter school diffusion use both. Of the five
studies examining charter school diffusion, two studies tracked charter school application
submissions (cite these two here) and three used charter school openings as dependent
variables (cite these three here) (Renzulli, 2005, Renzulli and Roscigno, 2005, Renzulli,
2002, Teske, Schneider and Cassese, 2005, Witte, Scholomer, and Shober, 2007).
Renzulli (2005) used application submissions because examination of charter school
openings would exclude the number of rejected applications and underestimate the
initiation patterns for charter schools and ignore important information about how
organizational environments foster or subdue efforts at innovation (Renzulli 2005, p. 4).
She found that between the years 1991 and 1998, 1,147 applications were submitted and
418 schools received a contract and opened. Indeed, using openings alone could
underestimate the interest in establishing charter schools.
In Colorado, however, one may not need to gather submission data to capture the
breadth of the charter school initiation efforts, given the ease with which a rejected
applicant may appeal to the State Board of Education. Over the past seven years (2006-
2012), there have been 32 appeals to the State Board. In 12 cases, the State Board upheld
36


the school boards decision. In 10 cases, the school board was ordered to reconsider. In
four cases on second appeal, the State Board ordered the school board to allow the school
to be opened. Parties settled the issue and dropped the appeal in four cases, and in two
cases, the appeal was rejected because of failure by the applicant to follow the appeals
process. In all, 83 charter schools opened during those years (Colorado Department of
Education 2013).
Moreover, Colorado school districts may not be openly hostile to charter schools,
otherwise they risk losing their Exclusive Chartering Authority granted by the State
Board. In the 10 school districts without exclusive authority, the states alternative
authorizer, the CSI, may open schools without district approval.
Given the rate of charter school openings and successful appeals, this study will
use the number of existing schools, rather than submissions as its dependent variable.
Quantitative Methods
First, a multivariate regression model is used for the second dependent variable,
percentage of students educated in district charter schools. A second analysis uses
negative binomial regression model using a simplified dependent variable namely the
number of charter schools. This modeling approach is suitable because of the distribution
of this dependent variable indicating the raw number of charter schools. This approach
has similarly been employed by other scholars Zhang and Yang (2008) and Renzulli
(2002 and 2005).
Model 1: Multivariate Analysis was used with dependent variable 2 (percentage of
students in districts charter schools) and 10 independent variables. Statistics for private
37


schools, secular private schools, and faith-based private schools were analyzed separately
because of high covariance levels.
Model 2: Negative Binomial Regression was used with the first dependent variable
(number of charter schools) and 10 independent variables. Minority and low income
students were analyzed together and separately.
38


CHAPTER TV
RESULTS AND DISCUSSION
Descriptive Statistics
Table 5 provides descriptive statistics for the dependent and independent variables
used in this analysis.
Table 5: Descriptive Statistics Dependent Variables
Mean Std. Dev. Min. Max.
Number of 0.87 3.28 0 37
Charter Schools in District % of Charter School Students in District 0.02 0.05 0 0.25
% Minority Students 0.33 0.22 0 0.95
% Poor Students 0.47 0.18 0.07 0.90
High District Accreditation Status 0.88 0.33 0 1
Low District Accreditation Status 0.12 0.33 0 1
Secular Private Schools 0.61 2.20 0 15
Faith-based Private Schools 0.94 3.05 0 28
All Private Schools 1.56 4.95 0 41
Adjacency to Districts with Charter Schools 0.58 0.50 0 1
Percentage of Democrats 0.28 0.12 0.1 0.72
Change in Enrollment -0.03 0.16 -0.79 0.45
Enrollment 4780.14 12539.58 10 85508
More Urban/Suburban 0.24 0.43 0 1
More Rural 0.76 0.43 0 1
Funding 9732.59 2545.6 7207.46 16539.42
39


Correlations
As Table 6 shows, several of the independent variables are highly correlated
within the sample used for this study. Specifically, the demographic factors of economic
affluence and racial minority population are somewhat correlated with a correlation
coefficient of 0.63. Variables indicating the number of private, faith-based, and secular
private schools are highly correlated. Enrollment and the number of private schools are
also highly correlated. This is not surprising since large, populous districts are more
likely to have more private schools than smaller, less populous areas.
Table 6: Correlation Matrix of Independent Variables
Mn Pr High Low Sec FB Pri Adj Dem Enr Enr Urb Rur Fund
Std Std Acc Acc Schl Schl Schl Chg Sub
Min 1.00
Std
Pr .63 1.00
Std
High -.45 -.31 1.00
Acc
Low .45 .31 -1.00 1.00
Acc
Sec .10 -.15 -.05 .05 1.00
Schl
FB .21 -.01 -.18 .18 .78 1.00
Schl
Pri .17 -.07 -.13 .13 .92 .96 1.00
Schl
Adj -.04 -.28 -.01 .00 .23 .24 .25 1.00
Dem .47 .30 -.23 .22 .17 .14 .16 -.01 1.00
Enr .15 -.15 .10 -.10 .18 .18 .19 .24 .02 1.00
Chg
Enr 0.20 -.13 -.13 .13 .77 .89 .89 .29 .13 .02 1.00
Urb .32 -.03 -.19 .18 .45 .50 .50 .30 .10 .13 .57 1.00
Sub
Rur -.32 0.03 .19 -.19 -.45 -.50 -.50 -.30 -.10 .10 -.57 - 1.00
1.00
Fund -.30 0.13 .13 -.13 -.23 -.25 -.25 -.37 -.09 -.10 -.30 -.47 .46 1.00
40


Model 1: Percentage of District Charter Students and 10 Independent Variables
The first analysis presented in this section is a multivariate regression of
Dependent Variable 2 (students educated in district charter schools as a percentage of all
district students) and 10 independent variables. Private schools, faith-based private
schools and secular private schools were analyzed separately as are poor and minority
students. In Model 1, district enrollment, urbanicity, and district adjacency to at least one
other district with charter schools were correlated with a higher percentage of district
charter school students. When all private schools and secular private schools are analyzed
in this model, there is a negative correlation between the number of schools and
percentage of district charter school students. When just secular private schools are
analyzed, the percentage of poor students in the district is found to be negatively
associated with the size of the charter school sector in the district. Similarly, when the
percentage of poor and minority students are analyzed separately, the percentage of poor
students is negatively correlated at a statistically significant level of. 10.
41


Table 6: Regression Results for Model 1, Dependent Variable 2 (percentage of students
in district charter schools) for all Colorado school districts in 2012-13
Independent Variable
Minority Students .00 .00 .00
(.02) (.02) (.02)
% of Poor Students -.03 . Q4** -.03
(.02) (.03) (.03)
Low Accreditation .00 -.00 .00
(.01) (.01) (.01)
# of Private Schools -.00*
(.00)
# of Secular Private -.01*
Schools (.00)
# of Faith-based -.02
Private Schools (.03)
Adjacent to Charter .01* 01* .01*
Districts (.01) (.01) (.01)
% of Democratic -.01 -.00 -.02
Registration (.03) (.03) (-03)
Enrollment Change .02 .02 .02
2008-2013 (-02) (-02) (.02)
District Total 2.12* 1.73* 1.76*
Enrollment (5.55) (4.16) (5.66)
Greater Urbanicity .03* .03* .03*
(.01) (.01) (.01)
District Funding -1.42 -1.36 -1.44
(1.48) (1.48) (1.50)
Constant .03 .03 .04
(-02) (-02) (.02)
* Statistically ** Statistically
significant at .05. significant at .10.
42


Model 2: Number of Charter Schools and 10 Independent Variables
In Model 2, Negative Binomial Regression was used with the first dependent
variable (number of charter schools) and 10 independent variables. As Table 7 shows,
only the results for total number of private schools are shown. Analysis of secular private
schools did not produce different findings and data that included faith-based private
schools did not converge in Stata. In this model, the number of private schools and the
degree of enrollment change are correlated with greater numbers of charter schools at a
.10 statistically significant level. Adjacency to charter school districts, total enrollment,
and greater urbanicity are positively correlated and funding is negatively correlated. No
correlation was found when poverty and minority rates were analyzed separately.
Table 7: Negative Binomial Regression
Independent Variable
Minority Students 1.16
% of Poor Students (1.29) -.22
(1.26)
Low Accreditation .07
(.44)
# of Private Schools 04**
(.02)
Adjacent to Charter Districts 2.00*
(.74)
% of Dem Registration -.79 (1.48)
Enrollment Change 1.75** (.99)
District Total Enrollment .00* (8.42)
Greater Urbanicity .91* (.36)
District Funding -.00* (.00)
Constant 2.30 f2.325
* Statistically significant at .05. ** Statistically significant at .10.
43


Discussion
As Table 7 shows, several of the hypotheses were confirmed.
Table 7: Hypotheses and Results
Hypotheses Model 1 Model 2
H]: School districts with a higher percentage of nonwhite students will have a greater quantity of charter schools. No correlation No correlation
H2: School districts with a higher percentage of low income students will have a greater quantity of charter schools. No correlation (in most cases) No correlation
H3: School districts with a lower accreditation plan will have a greater quantity of charter schools. No correlation No correlation
H4: School districts with a higher number of private schools will Negative correlation Positive
have a greater quantity of charter schools. in most cases correlation
H5: School districts adjacent to districts with charter schools are more likely to have charter schools. Positive correlation Positive correlation
H6: The Percentage of Democratic registered voters will have no impact on the quantity of charter schools in school districts. No correlation found No correlation found
H7: School districts with higher levels of enrollment growth will have a greater quantity of charter schools. No correlation found Positive correlation
H8: School districts with higher levels of enrollment will have a greater quantity of charter schools. Positive correlation Positive correlation
H9: School districts with a Denver Metro, Urban/Suburban, and Outlying City designation will have a greater quantity of charter schools than Outlying Town, and Rural school districts. Positive correlation Positive correlation
H10: School districts with lower levels of total per pupil funding will have a higher quantity of charter schools. Negative correlation Negative correlation
44


Level of Need
Hypotheses in this group predicted that the percentage of minority and low
income students and lower academic achievement would positively correlate with
chartering activity. The majority research showed a link between minority students and
lower academic achievement with chartering and the research is mixed on the impact of
low income students. In this study, the two models showed no statistically significant
findings on level of need except in one case when only data for secular private schools
was included. In this case, a negative correlation was found. It is possible that the racial
and economic demographics in Colorado differ than in states studied by other
researchers. Many of the Colorados poorest districts are not urban districts but rural
districts in the states San Luis Valley and other rural areas. These districts do not have
charter schools. It is also possible that district accreditation status, a holistic measure of
school quality and outcomes, produces different results than subject proficiency rates.
School Choice Environment
The hypotheses regarding school choice environment predicted that the number of
private schools would be associated with more charter schools. The models, however,
produced contradictory results. Model 1 found a negative association and Model 2, a
positive association. The existing research is also mixed for this variable. As was
expected, the percentage of Democratic registration is not associated with charter
activity. Charter schools have significant bipartisan support in Colorado which may
explain why some states produce different findings on the impact of partisanship on
charter school growth. A positive association between adjacency to districts with charter
45


schools concurs with existing research. Most Front Range districts, where charter activity
is the greatest, abut one another, so this finding is not surprising.
District Capacity
The hypotheses in this group concern enrollment, enrollment growth, funding,
and urbanicity. Both models showed a positive association between enrollment and
urbanicity and greater charter activity. This finding concurs with existing research. Only
Model 2 found an association between enrollment growth and more charter schools. Both
models found an association between lower funding and great charter activity which
aligns with existing research. Colorados highest funded districts are its rural districts, so
the finding is not surprising.
There is also a possibility of reverse causation in this analysis as it pertains to
enrollment and dependent variable 1 (Model 1), the number of charter schools. Since
charter schools are schools of choice, they draw students from outside of the district.
High performing charter schools attract students from outside of the district and thus
increase district enrollment and enrollment growth. Thus the dependent variable may
cause increases in the variables regarding enrollment and enrollment growth.
Conclusion
Using two statistical models and a variety of data, this thesis shows enrollment,
urbanicity and adjacency to a chartering district are correlated with greater chartering
activity. Per pupil funding is negatively correlated and there does not appear to be a
correlation between the number of charter schools and district academic achievement or
political affiliation of district voters. The presence of private schools may have a negative
46


or positive impact on chartering activity. A more sophisticated statistical model may be
able to solve the differing results in this study.
The findings in this study provide some insight into why some districts have more
charter schools than others. Though urban, adjacent to charter districts, and similarly
funded to other Denver Metro areas, the Englewood and Sheridan districts have no
charter schools. A potential reason could be that their student populations are small; they
are more similar to rural districts in terms of size. The paucity of charter schools in
isolated, rural, low population areas accurately predicts fewer charter schools on the
eastern plains and mountain regions.
The results, however, do not illuminate reasons for why Cherry Creek has so few
charter schools compared to other districts of its size and funding level. Boulder Valley
has a similar funding level and far fewer students but far more charter schools. In the
same way, Denver has a similar student population size and higher per-pupil funding than
Jefferson County, yet has twice as many charter schools.
As with many studies, the results prompt new questions. Are there additional
quantifiable variables that can account for such differences or do the differences lie in
characteristics not measurable by numbers? The answer may lie in the qualities of
superintendents and boards, the presence or absence of policy innovators, parent demand,
or the strength of charter advocates and opponents. This thesis paves the way for other
researchers to ask new questions.
47


REFERENCES
Benigno, P. and Morin, K. (2013) On the road of innovation: Colorados charter school
law turns 20. (Independence Institute) Retrieved from the Independence Institute
website at http://www.i2i.org/files/file/IP-4-2013.pdf.
Barton, P.E. Why does the gap persist? Association for Supervision and Curriculum
Development Educational Leadership 62(3), 8-13.
Bierlein Palmer, L. and Gau, R. (2005) Charter school authorizing: policy implications
from a national study. Phi Beta Kappan 86(5), 353-357.
Budde, R. (1988). Education by charter: Restructuring school districts. San Francisco:
WestEd.
Buddin, R. (2012) The impact of charter schools on public and private school
enrollments. (Cato Institute Policy Analysis No. 707). Retrieved from the Cato
Institute website: http://www.cato.org/sites/cato.org/files/pubs/pdf/PA707.pdf.
Carpenter, D. and Kafer, K. (2013) The State of Charter Schools, Colorado Department
of Education, 1-82. Retrieved from the Colorado Department of Education
website at http://www.cde.state.co.us/cdechart/download/STATEREPORT.pdf.
Carpenter, D.M., and Kafer, K (2012). A History of Private School Choice. Peabody
Journal of Education: Issues of Leadership, Policy, and Organizations, (87) 3.
336-350 doi: 10.1080/0161956X.2012.679587
Chakrabarti, R. and Roy, J. (2011). Do charter schools crowd out private school
enrollment? Evidence from Michigan. Federal Reserve Bank of New York Staff
Report no. 472, 1-54. Retrieved at
http://www.newvorkfed.org/research/staff reports/sr472.pdf.
Clark, M.A. et al. (2011). Do Charter Schools Improve Student Achievement? Evidence
from a National Randomized Study (Mathematica Policy Research). Retrieved
from the Mathematica website http://www.mathematica-
mpr.com/publications/PDF s / education/charterschool s_WP .pdf
Colorado League of Charter Schools (n.d.) Facts and figures about Colorado charter
schools. Retrieved at the Colorado League of Charter Schools website at
http://www.coloradoleague.org/colorado-charter-schools/20th-anniversary-facts-
and-figures.php
Colorado Charter Schools Act, Colorado Revised Statutes § 22-30.5-101 et. seq (1993).
Colorado Charter School Institute (n.d.) CSI at a glance. Retrieved at Colorado Charter
School Institute website at
48


http://www.csi.state.co.us/UserFiles/Servers/Server 234507l/Image/CSI%20at%
20a%20gl ance-1314.pdf
Colorado Department of Education (n.d.) Colorado School Districts Listed by Setting.
Retrieved at the Colorado Department of Education website at
http://www.cde.state.co.us/sites/default/files/documents/cdereval/download/pdf/di
strictslistedbvsetting.pdf
Colorado Department of Education Membership (n.d.). Trends for District Totals (2006-
2013) Retrieved at the Colorado Department of Education
http://www.cde.state.co.us/cdereval/fall2013membershiptrendsfordistricttotals200
6-2013xls
Colorado Department of Education (n.d.) Summary of School Districts Regarding
Exclusive Chartering Authority Status 2012-13 School Year. Retrieved at the
Colorado Department of Educations website at
http://www.cde.state.co.us/sites/default/files/ECA 2012%20Final O.pdf
Colorado Department of Education (2013) CDE 2013 District Accreditation Ratings
2010-2013. Retrieved at the Colorado Department of Education website at
http://www.cde. state.co.us/accountabilitv/performanceframeworkresults#twentvt
welve
Colorado Department of Education. (2013) Students Attending Public Schools Not in
Parent's District of Residence. Retrieved at the Colorado Department of Education
website at
http://www.cde. state.co.us/cdereval/pupilcurrentdistrict#sthash.usazlqCv.dpuf
Colorado Department of Education. (2013) Pupil Counts by Year 1993 through 2013.
Retrieved at the Colorado Department of Education website at
http://www.cde.state.co.us/datapipeline/pupilctbwrl993through2013xls
Colorado Department of Education, (n.d.) Summary of School Districts Regarding
Exclusive Chartering Authority Status 2012-13 School Year. Retrieved at the
Colorado Department of Education website at
http://www.cde.state.co.us/sites/default/fdes/ECA 2012%20Final O.pdf
Colorado Department of Education (n.d.). Nonpublic Schools Fall 2011. Retrieved at the
Colorado Department of Education website at
http://www.cde.state.co.us/sites/default/fdes/documents/choice/download/2011-
12nonpublicdirectory.pdf
49


Conlan, S. et al. (2011) Building Charter School Quality in Colorado. Building Charter
School Quality. Retrieved at the Building Charter School Quality website at
http://charterschoolquality.org/media/1178/BCSQ_BuildingQualityColorado.pdf
DeBray-Pelot, E.H., Lubienski, C.A., and Scott, J.T. (2007). The institutional landscape
of interest group politics and school choice. Peabody Journal of Education, 82 (2-
3), 204-230. doi: 10.1080/01619560701312947
Department of Agriculture Child Nutrition Programs Income Eligibility Guidelines, 57
C.F.R. 17005 (2012). Retrieved at http://www.gpo.gov/fdsys/pkg/FR-2012-03-
23/pdf/2012-7036.pdf
Ericson, J. et al. (2001). Challenge and Opportunity: The Impact of Charter Schools on
School Districts. (RPI International for the U.S. Department of Education Office
of Educational Research and Improvement National Study of Charter Schools).
Retrieved at the US Department of Education website at
http://www2.ed.gov/rschstat/eval/choice/district impact.doc.
Friedman Foundation for Educational Choice, (n.d.) School Choice Programs. Retrieved
at the Friedman Foundation for Educational Choice website at
http://www.edchoice.org/School-Choice/School-Choice-Programs
Jeynes, W. H. (2003). Religion, education, and academic success. Greenwhich, CT:
Information Age.
Jeynes, W. H. (2007). American educational history: School, society, and the common
good. Thousand Oaks, CA: Sage.
Carpenter, D.M., and Kafer, K (2012). A History of Private School Choice. Peabody
Journal of Education: Issues of Leadership, Policy, and Organizations, (87) 3.
336-350 doi: 10.1080/0161956X.2012.679587
Kirst, M. W. (2007). Politics of charter schools: Competing national advocacy coalitions
meet local politics. Peabody Journal of Education, 82 (2-3), 184-203.
doi: 10.1080/01619560701312939
Lake, R. Ed. (2012) A Balanced Look at American Charter Schools in 2012. Center on
Reinventing Public Education University of Washington Bothell. Retrieved for
the Center on Reinventing Public Educations website at
http://www.crpe.org/sites/default/files/pub_hfrl2_mayl3.pdf
50


Minstrom, M. and Vergari, S. (1998). Policy networks and innovation diffusion: The case
of state education reforms. The Journal of Politics, 60 (1), 126-148.
doi: 10.2307/2648004.
Minstrom, M. (1997). Policy Entrepreneurs and the diffusion of innovation. American
Journal of Political Science, 41 (3), 738-770.
Mintrom, M. (2000). Policy Entrepreneurs and School Choice. Washington D.C.:
Georgetown University Press.
Mitchell, N. (2011, January 18). More students choicing out of district. Chalkbeat
Colorado. Retrieved from http://co.chalkbeat.org/2011/01/18/more-students-
choicing-out-of-di strict/
National Alliance for Public Charter Schools, (n.d.). Charter schools 101: The most
frequently asked questions. Retrieved at the National Alliance for Public Charter
Schools at http://www.publiccharters.org/About-Charter-Schools/Frequentlv-
Asked-Questions.aspx.
National Alliance of Public Charter Schools (n.d.) Details from the Dashboard: Charter
School Race/Ethnicity Demographics. Retrieved from National Alliance of Public
Charter Schools at http://www.publiccharters.org/wp-
content/uploads/2014/01/NAPCS-2010-2011 -Race Ethnicitv-Details-from-the-
Dashboard 20120516T152831.pdf
National Association of Public Charter Schools (n.d.) Details from the Dashboard:Charter
Schools by Geographic Region Retrieved from the National Association of Public
Charter Schools website at http://www.publiccharters.org/wp-
content/uploads/2014/01/Geographic-Location-Details-from-the-Dashboard-
Report_20120224T143 95 5. pdf
Public School Review (n.d.). Colorado Magnet Public Schools.
www.publicschoolreview.com/state_magnets/stateid/CO
Pushpam, J. (2002) The Approval Barrier to Suburban Charter Schools. (Thomas B.
Fordham Foundation) Retrieved from the Thomas B. Fordham Foundation
website at
http://www.edexcellence.net/sites/default/files/publication/pdfs/approvalbarrier_l
O.pdf
Raymond, M. et al. (2013) National Charter School Study. Center for Research on
Education Outcomes at Stanford University. Retrieved from the CREDO website
at http://credo.stanford.edu/documents/NCSS%202013%20Final%20Draft.pdf
51


Renzulli, L. A. (2002). Entrepreneurial ambitions in the public sector: A random effects
model of the emergence of charter schools in North Carolina. Education Policy
Analysis Archives, 70(19), 1-25. Retrieved from
http://epaa.asu.edu/epaa/vl0nl9/.
Renzulli, L. A. (2005). Organizational environments and the emergence of charter
schools in the United States. Sociology of Education, 78 (1), 1-26. doi:
10.1177/003804070507800101.
Renzulli, L. A. and Roscigno, V. J. (2005). Charter school policy, implementation, and
diffusion across the United States. Sociology of Education, 78 (4), 344-366.
doi: 10.1177/003804070507800404.
Rincke, J. (2006). Policy innovation in local jurisdiction: testing for neighborhood
influence in school choice policies. Public Choice 126 (1), 189-200.
doi: 10.1007/s 11127-006-9022-3.
Rincke, J. (2007). Policy diffusion in space and time: the case of charter schools in
California school districts. Regional Science and Urban Economics (37), 527-541.
doi: 10.1016./j.regsciurbeco.2007.02.001.
Shanker, A. (1988). Restructuring our schools. Peabody Journal of Education 65 (3), 88-
100. doi: 10.1080/01619568809538615
Schaeffer, A.B. (2012). The Charter School Paradox (Cato Institute Policy Analysis No.
707). Retrieved from the Cato Institute website:
http://www.cato.org/sites/cato.org/files/pubs/pdf/charter-school-paradox.pdf
Shober, A.F., Manna, P., and Witte, J.F. (2006) Flexibility Meets Accountability: State
Charter School Laws and Their Influence on the Formation of Charter Schools in
the United States Policy Studies Journal, (34)4, 563-587, doi: 10.1111/j. 1541-
0072.2006.00191.x
Teske, P., Schneider, M., and Cassese, E. (2005). Local school boards as authorizers of
charter schools. In W. G. Howell (Ed.), Besieged: School boards and the future of
education politics. (1229-1279). Washington, D.C.: Brookings Institution Press.
Stoddard, C. and Corcoran, S (2006). The Political Economy of School Choice: Support
for Charter Schools across States and School Districts. New York: National
Center for the Study of Privatization in Education.
Stuit, D. and Doan, S. (2012). Beyond city limits: expanding public charter schools in
rural America. National Alliance for Public Charter Schools Issue Brief, 1-16.
Retrieved from at National Alliance for Public Charter Schools at
52


http://www.publiccharters.org/wp-content/uploads/2014/01/NAPCS_Beyond-
City-Limits_Issue-Brief_2_27_2012_20120223 T101259.pdf
Teske, P., Schneider, M., Buckley, J., and Clark, S. (2000). Does charter school
competition improve traditional public schools? Center for Civic Innovation at the
Manhattan Institute Civic Report, 10 (June 2000), 1-31. Retrieved from
http://www.manhattan-institute.org/html/cr 10.htm.
Winkler, A.M., Scull, J. and Zeehandelaar, D. (2012). How Strong are U.S. Teacher
Unions? A state-by-state comparison. (Thomas B. Fordham Institute). Retrieved
from the Thomas B. Fordham Institute website at
http://www.edexcellence.net/sites/default/files/publication/pdfs/20121029-Union-
Strength-Full-Report_7_0.pdf
Witte, J.F., Schlomer, P.A., and Shober, A.F. (2007) Going charter? A study of school
district competition in Wisconsin. Peabody Journal of Education, 82 (2-3), 410-
439. doi: 10.1080/01619560701313051.
Wong, K.K. and Langevin, W.E. (2007). Policy expansion of school choice in the
American states. Peabody Journal of Education, 82 (2-3), 440-472.
doi: 10.1080/01619560701313085
Wong, K.K. and Shen, F.X. (2002). Politics of state-led reform in education: Market
Competition and electoral dynamics. Educational Policy 16 (16)1.
doi: 10.1177/0895904802016001009
Zhang, Y. and Yang, K. (2008). What drives charter school diffusion at the local level:
educational needs or political and institutional forces? The Policy Studies Journal,
36(4), 571-591. doi: 10.1111/j.l541-0072.2007.00284.x.
Torres, Z. (2014, January 15). Student enrollment up in Colorado; DPS now largest
school district. The Denver Post, Retrieved from
http ://www. denverpost. com/news/ci_24913752/student-enrollment-up-col orado-
dps-now-largest-school#ixzz30Epvqm
53


Full Text

PAGE 1

FACTORS THAT INFLUENCE THE QUANTITY OF CHARTER SCHO OLS IN COLORADO SCHOOL DISTRICTS by KRISTA KAFER B.A., University of Colorado Denver, 1994 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Arts Political Science 2014

PAGE 2

ii 2014 KRISTA KAFER ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Arts degree by Krista Kafer has been approved for the Political Science Program by Michael Berry, Chair Kathryn Cheever Dick Carpenter July 25, 201 4

PAGE 4

iv Kafer, Krista (M.A., Political Science) Factors that Influence the Quantity of Charter Scho ols in Colorado School Districts Thesis directed by Associate Professor Michael Berr y ABSTRACT In the 2012–2013 school year, 11 percent of all Col orado public school students (88,924) attended one of the state’s 187 charter sc hools. The number of charter schools and the percentage of district students educated in charter schools vary considerably from district to district. Research in other states show s that variables related to the level of need (percentage of low income and minority student s and district academic accreditation status), school choice environment (number of priva te schools, adjacency to districts with charter schools, and political affiliation), and di strict capacity (funding, enrollment, enrollment growth, and urbanicity) are associated w ith the number of charter schools authorized by districts. Using two statistical mode ls, this thesis shows many of these variables correlate with greater chartering activit y. Enrollment is the strongest predictor. Districts with higher student enrollments generally have higher charter enrollments and higher numbers of charter schools. Urbanicity and a djacency to a chartering district are correlated with greater chartering activity. Per pu pil funding is negatively correlated and there does not appear to be a correlation between t he number of charter schools and district academic achievement or political affiliat ion of district voters. The form and content of this abstract are approved. I recommend its publication. Approved: Michael Berry

PAGE 5

v TABLE OF CONTENTS CHAPTER I. INTRODUCTION.................................... ................................................... .....................1 II. REVIEW OF THE LITERATURE....................... ................................................... .....15 III. METHODOLOGY................................... ................................................... ................31 VI. RESULTS AND DISCUSSION......................... ................................................... .....39 REFERENCES......................................... ................................................... ......................48

PAGE 6

1 CHAPTER I INTRODUCTION Charter schools are public schools that are operate d independently from a school district through a charter agreement with an author izer—usually a school district, state governmental body, or university. Overseen by a gov erning board of parents and community members, charter schools have autonomy ov er daily operations, budgets, teacher contracts, training, and salary determinati ons, academic programs, school calendar, pedagogy, and curriculum. Like traditiona l public schools, charter schools must meet state academic standards and testing requireme nts, be tuition-free, and adhere to federal civil rights laws. Charter schools do not h ave entrance requirements. If a charter school fails to meet the terms of its contract, the authorizer may revoke the charter and close the school (National Alliance for Public Char ter Schools, n.d.). The concept of charter schools predates their exist ence by two decades. Ray Budde first suggested the idea at a conference in 1 974 and published a paper about the concept in 1988 (Renzulli, 2005; Budde, 1988). That same year, Albert Shanker, (1988) former president of the American Federation of Teac hers, advocated the idea in a paper published in the Peabody Journal of Education Three years later, Minnesota adopted the first charter school law. Today, more than 6,000 ch arter schools in 42 states and the District of Columbia educate approximately two mill ion students (National Alliance for Public Charter Schools, n.d.).

PAGE 7

2 Purpose of Study Since the adoption of the first charter school law in 1991, charter school growth across the country has been uneven. Most states ha ve been chartering schools for decades, but some states, like Washington State, ha ve just begun. Eight states do not have charter school laws. Even within states, chart er school growth varies by district. Several researchers have examined the reasons behin d variations in charter school diffusion at the state and local level. Many have found associations between charter school diffusion and student demographic factors li ke race, achievement and poverty, as well as state/district characteristics such as size of enrollment, political climate, adjacency to states/districts with charter schools, and other factors. No such research, however, has been done in Colorado. As Figure 1 shows, there is a strong correlation be tween student population and the number of charter schools. There are, however, outliers. The Cherry Creek School District, for example, has fewer charter schools th an its population would predict while Denver Public Schools has more. Figure 1: Distribution of charter schools in Colora do school districts. Source: Colorado Department of Education Fall 2012 data. n r r r r r nrrrr

PAGE 8

3 Emulating studies of other states, this research ex amines correlations between charter school distribution among Colorado school d istricts and student demographics (race and poverty) and district factors (accreditat ion level, school choice environment, adjacency to other districts with charter schools, political climate, enrollment and enrollment growth, urbanicity, and per pupil fundin g). The study measures charter school diffusion using t wo dependent variables: the number of charter schools authorized by a district and the percentage of students enrolled in district charter schools. These represent two di fferent ways of measuring charter school activity. The first dependent variable is the predominant var iable used in other diffusion studies. The second variable attempts to measure th e charter school sector in a district irrespective of district size. There are vast diffe rences in student population among districts. Agate 300, for example, has 10 students while Jefferson County has 85,808. While the percentage of students educated in charte r schools in a district includes some students from other districts who transfer into the district, the bulk of students are assumed to be district students because of proximit y. The purpose of using several variables is to see if they provide similar, and th us more valid, findings. The study employs two statistical models. The first model examines ten independent variables using multivariate regression analysis and the second dependent variable, district charter school students as a per centage of all district students. Private schools, faith-based schools, and secular schools a re tested separately because they are highly correlated. The second model uses Negative B inomial Regression to examine

PAGE 9

4 associations between 10 independent variables and t he number of charter schools (dependent variable). The utility of conducting quantitative analysis is that it paves the way for qualitative research. Having singled out more easi ly measurable variables associated with charter school diffusion, one can then underta ke research into variables that are less easily measured such as the qualities of superinten dents and boards, the presence or absence of policy innovators, and the strength of a dvocates and opponents. Researchers have suggested that these variables have an impact on charter diffusion in other states. The quantitative research undertaken in this thesis lays a foundation for such qualitative research in Colorado. Charter Schools in Colorado History Colorado became the third state to adopt a charter school law when Colorado Governor Roy Romer signed the Charter Schools Act i n 1993 (Benigno, 2013). In the 2012–2013 school year, 11 percent of all Colorado p ublic school students (88,924) attended one of the state’s 187 charter schools (Co lorado League of Charter Schools, n.d.). In addition to charter schools, Colorado students m ay enroll in any public school within or outside of their district, or attend one of the many district or statewide online schools. Open enrollment was first introduced in Co lorado when the legislature passed the Public Schools of Choice Act of 1990 to enable students to attend a public school outside of their attendance boundaries within or ou tside of their district (Mintrom, 2000). Today, 9 percent of Colorado students attended publ ic schools outside of their district

PAGE 10

5 compared to 8 percent in 2011 and 3 percent in 2001 (Mitchell, 2011 and Colorado Department of Education, 2013). Colorado also has public option schools and magnet schools which are district-run schools of choice. Such schools generally have a un ique pedagogical approach to other schools in the district. When parents choose a scho ol other than their neighborhood school, be it an option school, charter school, mag net school, online school, or school in another neighborhood attendance zone, they must sub mit an application. Enrollment is subject to space availability. Private schools are also an education option. There are 277 private schools listed in the Colorado Department of Education database (C olorado Department of Education, n.d.). One hundred and sixty-nine are faith-based a nd the remainder is secular private schools. Table 1: School Type Definitions Type of School Definition Traditional Public Schools District-operated public schools that serve a designated neighborhood. Under Colorado law, students can attend, space permitting, traditional public schools other than their assigned school. Charter Schools Independent public school. Students can attend, space permitting, any charter school. Option and Magnet Schools District-operated public school of choice. Students can attend, space permitting, any option or magnet school. Private Schools Independent secular and faith-based schools. In general, these schools do not receive public funding. However, Douglas County has initiated a scholarship program. The program is currently suspended pending a Colorado Supreme Court decision.

PAGE 11

6 Student enrollment in public schools in Colorado ha s grown every year since 1990 (Torres, 2014). In 1993, the year Governor Rom er signed the Charter Schools Act, the state had 625,000 K-12 students. Today, there a re nearly 877,000 students which marks a 40 percent increase. Between 2008 and 2013, 100 Colorado districts lost enrollment, five districts maintained the same level of enrollment, and 73 dis tricts gained enrollment. The level of change ranges from a 79 percent drop in enrollment in the Agate 300 district to a 45 percent increase in the Mapleton 1 school district (Colorado Department of Education, n.d.). In general rural districts declined in enrol lment while Front Range communities gained enrollment. Student Demographics in Colorado Traditional and Ch arter Public Schools As Figure 2 demonstrates, Colorado charter schools currently enroll the same percentage of minority students (44 percent) as do traditional public schools (44 percent) with charter schools enrolling slightly more Black and Asian students and traditional public schools enrolling slightly more Hispanic stu dents (Colorado League of Charter Schools, n.d.). This was not always the case; in 2 001, students of color were slightly less likely to enroll in charter schools than in traditi onal public schools. Twenty-seven percent of charter school enrollment was composed o f racial minority students as compared to 33 percent in traditional public school s (Carpenter, and Kafer, 2013).

PAGE 12

7 Figure 2: Racial Demographics at traditional and pu blic charter schools. Source: Colorado League of Charter Schools 2012–201 3 data. In contrast to national trends, traditional public schools in Colorado serve a higher percentage of low income students than charter scho ols (Carpenter, and Kafer 2013). The percentage of charter school students who qualify f or the federal Free and Reduced Lunch program was 32 percent in 2012 as compared to 42 percent of traditional public schools. The percentage of low income students in c harter schools, however, has increased significantly since 2001 when it stood at 18 percent. The increase in the percentage of low income studen ts served by Colorado charter coincides with the increase urban charter schools a nd charter school networks seeking to serve disadvantaged students. Charter school founde rs, often with the support of philanthropists, have responded to demand by openin g charter schools to serve urban low income and minority students and by offering suppor t such as tutoring, a longer school day, Saturday hours, smaller class sizes, access to social services, and home visits. Several successful charter operators such as KIPP, Denver School for Science and Technology and STRIVE Preparatory Schools have mult iple inner-city campuses across the Denver Metro Area. r n r n r r n n nr

PAGE 13

8 Achievement trends among minority and low income st udents in Colorado are similar to national trends. Low income students and Black and Hispanic students have lower proficiency levels in math and reading than m ore affluent students and White students. Poor Black, White, and Hispanic students in charter schools generally performed better than their peers in traditional pu blic schools on reading and math state assessments (Carpenter, and Kafer, 2013). More aff luent Black and Asian students in charter schools also achieved better outcomes than their peers in traditional public schools in reading while more affluent Hispanic stu dents in traditional public schools generally performed better than their peers in char ter schools. Findings were mixed for White students. In math, more affluent Black studen ts in charter schools performed better than their peers in traditional public schools whil e findings for Hispanic, Asian, and White students were mixed by grade level. Charter School Diffusion Under Colorado law, charter schools may be authoriz ed by school districts or the Colorado Charter School Institute (CSI). CSI may au thorize charter schools in school districts that do not have “exclusive chartering au thority” given to them by the state board of education. The law grants districts that enroll fewer than 3,000 students exclusive chartering authority automatically and requires lar ger districts to apply for the privilege (Conlan, 2011). The state board can revoke exclusiv e chartering authority if the district fails to uphold the Charter Schools Act. In the 201 2–2013 school year, nine districts lacked exclusive chartering authority (Colorado Dep artment of Education, n.d.). CSI also charters schools in districts with exclusive charte ring authority with the permission of the district. In the 2012–2013, CSI held charter agreem ents with 28 schools (Colorado

PAGE 14

9 Charter School Institute, n.d.). Because CSI is not a district but a state-wide authorizer whose sole purpose is to charter schools, its schoo ls are not included in this study. This study focuses exclusively on districts which may or may not authorize charter schools. Forty-four of the stateÂ’s 178 school districts have chartered schools. As Figure 3 shows, charter schools exist throughout the state b ut are more common along the urbanized Front Range where the population density is higher. The percentage of suburban and rural charter schools in Colorado is h igher than the national average (National Association of Public Charter Schools, n. d.). Thirty-seven percent of Colorado charter schools are located in urban environments a s compared to 52 percent nationwide. Twenty-six percent of Colorado charter schools are located in the suburbs. Only nine states have a higher percentage of suburban charter schools. Thirty percent of ColoradoÂ’s charter school students attend a rural charter scho ol (Stuit, and Doan, 2012). Only nine other states have an equal or greater percentage of charter school students in rural schools. Figure 3: Distribution of Charter Schools in Colora do by City Source: Colorado League of Charter Schools (n.d.). Facts and figures about Colorado charter schools. Copied with permission.

PAGE 15

10 The Colorado Department of Education rates each sch ool district according to its urban/rural setting on a scale of one to five: 1) D enver Metro; 2) Urban/Suburban; 3) Outlying City; 4) Outlying Town; and 5) Rural (Colo rado Department of Education, n.d.). One hundred and five districts are either De nver Metro (19) or part of another urban/suburban area (86). The remainder is part of an outlying city (51), outlying town (18), or rural area (4). Figure 4 shows the relatio nship between urbanicity and the number of charter schools. Urban and suburban districts ar e more likely to have charter schools. There some interesting outliers, however; three urb an districts, Adams County 14, Englewood 1 and Sheridan do not have charter school s while five rural districts, Elizabeth C-1, Keenesburg RE-3J, Strasburg 31J, and West End RE-2 have charter schools. On the other end of scale, Cherry Creek h as only one charter school and Adams County 14, Sheridan and Englewood have no charter s chools even though they all have a Denver Metro setting designation. Figure 4, Distribution of charter schools in Colora do school districts by district setting. Source: Colorado Department of Education Fall 2012 data. r r r r rrrrrr

PAGE 16

11 While the urban-rural difference in chartering rate s is clear, the pattern of distribution of charter schools within urban and su burban school districts is not as easily apparent. Table 2 shows the top 10 districts by stu dent population along with the number of charter schools. The table shows that the size of the district in terms of student enrollment is not a completely reliable predictor o f charter school presence. For example, Colorado Springs 11 ranks ninth in student population but has the fourth highest number of charter schools. Denver Public Schools, t he second largest district by enrollment, has more than twice as many charter sch ools as the largest district, Jeffco Public Schools. Table 2: Top 10 Colorado school districts by student enrollm ent and number of charter schools School District Total Enrollment Number of Charter Schools 1. Jeffco Public Schools 85,508 14 2. Denver Public Schools 83,377 37 3. Douglas County School District 64,657 11 4. Cherry Creek Public Schools 53,368 1 5. Adams 12 Five Star Schools 43,268 6 6. Aurora Public Schools 39,835 6 7. Boulder Valley Public Schools 30,041 5 8. St Vrain Valley School District 29,382 6 9. CO Springs School District 11 28,993 7 10. Poudre School District 27,909 4 Source: Colorado Department of Education Fall 2012 data. When viewed as a scatterplot in Figure 5, it appear s that student enrollment alone does not predict the number of charter schools.

PAGE 17

12 Figure 5, Distribution of charter schools in top 10 largest Colorado school districts. Source: Colorado Department of Education Fall 2012 data. Ranking districts of different sizes (according to the number of students) by the number of charter schools provides an interesting c omparison but it does not take into account the impact of size. One would expect larger districts to have more charter schools because they have more students thus more demand an d more per-pupil funds. A smaller district may have a large proportion of students in charter schools but fewer charter schools overall simply by virtue of its size. By vi ewing district chartering activity as a ratio of students per charter school, one can compa re charter activity among districts of different student enrollment sizes. Table 2, provides just such a comparison. The four th largest school district, Cherry Creek Public Schools, has 53,368 students pe r charter school, while Colorado Springs 11, the ninth largest district by populatio n, has 4142 students per charter school. Jeffco Public Schools has 35,000 more students than the Boulder Valley School District but a similar sized charter schools sector. "##$n n $%& '( %$ $%' )*n !)n& +$%' r rrrr

PAGE 18

13 Table 3: Top 10 CO school districts by ratio of students per charter school School District District Students per Charter School Percent of District Students Enrolled in Charter Schools 1. Denver Public Schools 2253/1 14% 2. CO Springs School District 11 4142/1 8% 3. St Vrain Valley School District 4897/1 11% 4. Douglas County School District 5878/1 15% 5. Boulder Valley Public Schools 6008/1 8% 6. Jeffco Public Schools 6108/1 8% 7. Aurora Public Schools 6639/1 10% 8. Poudre School District 6977/1 6% 9. Adams 12 Five Star Schools 7211/1 20% 10. Cherry Creek Public Schools 53,368/1 Less than 1% Source: Colorado Department of Education Fall 2012 data Another way to look at the size of the size of a sc hool district’s charter school sector is to examine the percentage of students tha t are enrolled in charter schools in a district. Table 3 also shows the percentage of stu dents educated in charter schools in each district. While some of the students educated in the districts’ charter schools are not from the district, the same is true for the distric t’s traditional public schools. Under Colorado law, students may transfer to other distri ct charter and traditional public schools, space permitting. In the top ten largest school districts, as few as 1 percent and as many as 20 percent of students educated in a district are enro lled in a charter school. Adams 12, the fifth largest school by total district population, enrolls the highest percentage of students in charter schools. Though not in the top 10 distri cts by enrollment, Brighton 27, Falcon 49, and Greeley school districts have a higher prop ortion students enrolled in charter schools—21 percent, 17 percent, and 17 percent resp ectively—than nine of the 10 largest

PAGE 19

14 school districts. Several rural districts have a hi gh proportion of students enrolled charter schools: Park County (25 percent), West End (15 per cent), Elizabeth (17 percent) and Clear Creek (13 percent). While there is a positive correlation between distr ict size and charter school presence, there are other factors that likely exert an influence on the prevalence of charter schools. As the following analysis will show, stud ent demographics (race and poverty) and district factors (accreditation level, school c hoice environment, adjacency to other districts with charter schools, political climate, enrollment and enrollment growth, urbanicity, and per pupil funding) impact the degre e of charter school diffusion.

PAGE 20

15 CHAPTER II REVIEW OF THE LITERATURE Most of the researchers found positive associations between charter diffusion and non-white students, low district achievement, adjac ency to districts with charter schools, and district student density. Some researchers foun d correlations between charter school diffusion and enrollment of low income students, pe rcentage of district private schools, and political climate. Others, however, did not. Ta ble 4 provides a summary of the current state of knowledge in this field. The liter ature review that follows provides a discussion of most of these variables studies have associated with charter school diffusion.

PAGE 21

16 Table 4: Summary Table of Literature Review Vari-able Percentage of Minority Students Percentage of Low Income Students District Academic Quality Percentage of Private Schools Adjacency Political Factors District Capacity Other Variables Renzulli and Roscigno (2005) + to number of charters in state Renzulli (2005) to charter application submissions. Renzulli (2002) + to application submissions. Renzulli and Roscigno (2005) + to a point, then negative. Rincke (2007) + to district charter adoption. Renzulli (2002) + between Democratic registration and charter application submissions. RenzulliÂ’s (2005) + between urbanicity and charter school submissions. Renzulli (2005) + special education students and charter application submissions Renzulli (2005) + to charter application submissions. Rincke (2007) with district charter adoption Rincke (2007) + to district charter adoption. Renzulli (2005) + secular schools only. Witte, Schlomer, and Shober (2007) + to district charter school adoption. Renzulli and Roscigno (2005) between Republican governor and charter submissions. Witte, Schlomer, and Shober (2007) + between district enrollment and charter schools. Rincke (2007) + with magnet schools and charter school adoption. Rincke (2007) with district charter adoption. Hispanic only. Witte, Schlomer, and Shober (2007) + to district adoption. Minstrom (1997) + between decrease in test scores and charter law adoption. Wong, K.K. and Langevin (2007) + to charter law adoption. Zhang and Yang (2008) + to charter schools openings Shober, Manna, and Witte (2006) + correlation between state charter schools and Republican governor or legislature. Renzulli (2005) + between district size and charter application submissions. Renzulli and Roscigno (2005) + between open enrollment law and charters in state. Witte, Schlomer, and Shober (2007) + to charter school openings Stoddard and Corcoran (2006) + between higher dropout rates and charter school enrollments in districts. Stoddard and Corcoran (2006) + states with low achievement and higher charter enrollments. Zhang and YangÂ’s (2008) correlation with charter school creation. Wong and Langevine (2007) + between Republican governor and adopting a charter law. Witte, Schlomer and Shober (2007) + between district chartering and federal funds. Minstrom (1997) between union strength and state adoption of charter law. Wong, K.K. and Langevin (2007) + to state adoption of charter law Wong and Shen (2002) + between state dropout rates and charter law adoption. Zhang and Yang (2008) + Democratic registration and district chartering to a point, then negative. Wong, and Langevin (2007) + between lower state funding and law adoption. Minstrom (1997) + between presence of policy entrepreneurs and policy adoption. Wong, K.K. and Langevin (2007) + to state adoption of charter law Wong and Shen (2002) + between state dropout rates and charter law adoption. Zhang and Yang (2008) + Democratic registration and district chartering to a point, then negative. Wong, and Langevin (2007) + between lower state funding and law adoption. Minstrom (1997) + between presence of policy entrepreneurs and policy adoption.

PAGE 22

17 Researchers have examined variables that correlate with charter school diffusion across the country in terms of passage of charter s chool enabling laws and charter school openings (or charter school application submissions ). Although this thesis focuses on variables that influence charter diffusion in distr icts, findings of both types of existing diffusion research are reviewed here for two reason s: There are few studies that analyze Variable Percentage of Minority Students Percentage of Low Income Students District Academic Quality Percentage of Private Schools Adjacency Political Factors District Capacity Other Variables Wong, K.K. and Langevin (2007) + to state adoption of charter law Wong and Shen (2002) + between state dropout rates and charter law adoption. Zhang and Yang (2008) + Democratic registration and district chartering to a point, then negative. Wong, and Langevin (2007) + between lower state funding and law adoption. Minstrom (1997) + between presence of policy entrepreneurs and policy adoption. Zhang and Yang (2008) + to charter schools openings. Black students only. Zhang and Yang (2008) – to charter school openings. Renzulli (2002) + between lower funding and higher charter submissions Renzulli and Roscigno (2005) between union strength and state adoption of charter law. Stoddard and Corcoran (2006) + to charter schools openings. Black students only. Zhang and Yang (2008) to charter schools openings. Zhang and Yang’s (2008) + between appointed superintendent and chartering in districts. Renzulli (2005) + between number of district administrators and charter submissions. Overall Mostly positive Mixed Positive Mixed Positive Mixed Positive N/A

PAGE 23

18 factors that influence district chartering rates so it is useful to broaden the analysis to include other diffusion analysis. Secondly, both ty pes of research examine similar variables. The main factors that influence charter school adoption at the state and district level are related to student demographics, district school quality, political environment, district size, proximity to other districts or stat es with charter schools, union strength, and revenue. The subheadings below provide a discussion of extant research as it applies to each of these characteristics. Disadvantaged Students “There is good reason to believe that minorities are attracted to school-choice options precisely because they have been so disadva ntaged in the public education system” observed Renzulli and Roscigno in their 200 5 study (350). Students of color and low income students have traditionally performed at lower academic levels than White students and more affluent students (Barton, 2004). For these traditionally underserved populations, charter schools have become a popular alternative education environment. Nationwide charter schools enroll, on average, a gr eater percentage of Black and Latino students (27 percent and 26 percent, respectively) than traditional public schools (15 percent and 22 respectively) (National Alliance of Public Charter Schools, n.d.). Charter schools also enroll a higher percentage of low inco me students (53 percent) than traditional public schools (47 percent) (Lake, 2012 ). Although there is considerable diversity within the charter school sector, several studies suggest that charter schools, in general, p roduce positive academic impacts for disadvantaged student populations. A 2011 national randomized study of the effect of

PAGE 24

19 attending a charter school on academic progress by Mathematica, a policy research organization, found “positive impacts for more disa dvantaged schools and students and negative impacts for the more advantaged” (Clark, M .A. et al., 1). A more recent study by the Center for Research on Education Outcomes at Stanford University showed that attending a charter school had positive impacts for Hispanic students who are English language learners and Black students in poverty (Ra ymond et al. 2013). Hispanic English-language learners gained the equivalent of 50 additional days of learning in reading and 43 additional days in math over their p eers in traditional public schools. Poor, Black students gained 29 additional days in r eading and 36 additional days of learning in math over their peers in traditional pu blic schools. More affluent Hispanics experienced similar achievement to their traditiona l public school peers while White and Asian students lost ground. Several researchers have tested the hypothesis of w hether a higher presence of minority, low income, or learning challenged studen ts is associated with a greater likelihood of a state passing a charter school law or of a district chartering schools. Minority Students Renzulli and Roscigno’s (2005) found that as the pe rcentage of nonwhite students increased so did the number of charter schools with in a state. In similar state-level adoption research, Wong, and Langevin, (2007) demon strated that the percentage of minority students has positive impact on state adop tion of charter school legislation. At the district level, Renzulli’s 2005 study found the percentage of nonwhite and special education students to have a positive corre lation with the number of charter

PAGE 25

20 school application submissions. Witte, Schlomer, an d Shober (2007) also hypothesized that districts with more nonwhite students would op en charter schools. Their analysis demonstrated a positive correlation with the number of charter schools in districts in Wisconsin. Zhang and Yang (2008) found a greater p ercentage of Black students to be positively associated with more charter schools in Florida districts. Stoddard and Corcoran’s 2006 research also showed districts with high or increasing percentage of Black students, a high or increasing percentage of college graduates and growing income inequality had larger charter school enrollments th an did more homogeneous districts. Rincke (2007), however, found greater enrollment of Hispanic students to be negatively associated with charter adoption in California dist ricts. Rincke acknowledged that this finding conflicted with the claim asserted by chart er schools advocates that charter schools benefit disadvantaged children. He asserts that “the more favorable the social conditions under which local public school producer s operate, the more likely is the establishment of additional charter schools” (538). This may have been true during the late 1990s and early 2000s from which Rincke’s data are drawn. As discussed earlier, the proportion of disadvantaged students served by Colo rado charter schools in the early years was lower than it is now. Low Income Students A number of existing studies have found that increa ses in low income student population exert a suppressing effect on charter sc hool establishment (Rincke, 2007; Zhang and Yang 2008; and Renzulli, 2005). Zhang and Yang (2008) also examined the impact of the percentage of learning disabled stude nts and found no impact. Witte, Schlomer, and Shober (2007), however, found distric ts with charter schools to have

PAGE 26

21 higher average low income student enrollments. It i s not clear why the findings are contradictory. All use data from the late 1990s and early 2000s. Witte, Schlomer, and Shober’s analysis is on Wisconsin schools, Rincke’s California, Renzulli’s North Carolina, and Zhang and Yang’s Florida. Perhaps sta te differences explain the variance in findings. Overall, researchers agree that the perce ntage of non-white students is associated with higher charter activity. District Academic Quality The presence of students in poverty and students of color in a district is not the only potential predictor of chartering activity. St udents in low-performing districts regardless of ethnicity or poverty are likely to de sire higher quality schooling options. Testing this hypothesis between district school qua lity and chartering activity, Renzulli (2002) found that a greater proportion of low perfo rming schools in a district increased the mean number of charter school application submi ssions. Similarly, Rincke’s (2007) research into charter school diffusion in Californi a found that districts with low achievement were more likely to establish charter s chools. Zhang and Yang (2008), however, found that the percentage of failing schoo ls in a district had a negative correlation with charter school openings in Florida Zhang and Yang had hypothesized that failing schools would lead to more charter sch ools but concluded that “improving educational performance may be a major concern for potential charter founders but not be the guiding principle for school boards and local p olitics” (583). The authors acknowledge that their findings are different from Renzulli but offer no explanation as to the reason.

PAGE 27

22 In research considering state charter laws, Minstro m (1997) found that decreasing test score averages increased the likelihood of the passage of a charter law. Analysis by Stoddard and Corcoran (2006) found states with low student achievement and districts with higher dropout rates had higher charter school enrollments. Wong and Shen (2002) found that a stateÂ’s adoption of a charter law was inversely related to its graduation rate, that is, states with lower rates were more likely t o adopt charter laws. These studies suggest that charter schools may provide one avenue to improve student learning when test scores indicate decreases in student performan ce. School Choice Environment The school choice environment refers to the demand for and availability of education options in the district and adjacent dist ricts, and the political climate. School choice, that is, the ability of parents to choose f rom among public or private options with the support of public funds is not a new concept in Colorado or the nation. In the colonial era and early republic, student were educa ted through a variety of independent schools financed by local communities, churches, an d charitable organizations (Jeynes, 2003; 2007). Although tax supported public schools became the norm in the late 19th century and early 20th Century (Carpenter and Kafer 2012), the desire for alternatives resumed in the mid and late 20th century. In 1955 Minnesota adopted the first tax c redit for private school tuition. Today there are 23 tuit ion tax credit programs exist in 15 states (Friedman Foundation for Educational Choice n.d.). In 1990, the first modern voucher program was enacted (Vermont and MaineÂ’s century an d a half old rural voucher programs aside). Today there are 22 programs in 12 states and the District of Columbia.

PAGE 28

23 Presence of Private Schools One way to measure the demand for school choice is the number of private schools in a district. According to research by Sch aeffer (2012), approximately 8 percent of charter elementary students and 11 percent of mi ddle and high school students come from the private sector. In other words, they left their private school to attend a public charter school. In urban centers, private school st udents constitute substantial portions of charter elementary (32 percent), middle (23 percent ) and high school school (15 percent) students come from private schools. Buddin’s (2012 ) research and that of Chakrabarti and Roy (2011) also suggest that charter schools pu ll a significant number of students from private schools. Taken together, these results indicate that private schools may both bolster school choice legitimacy and demand and ser ve as an alternative source for prospective students. A study by Renzulli and Roscigno (2005) showed that the number of private schools increased the number of charter schools ope rating in a state. They suggest that “the presence of competition, in the form of privat e schools, increases the number of charter schools that operate in the state” (358). T hey also point out that correlation waned in states with the highest level of private schools indicating that there may be a saturation point for education alternatives. Another of Renzulli’s studies (2005) showed that th e number of secular private schools, but not faith-based schools, had a positiv e impact on the number of charter school application submissions in school districts. She suggests that education alternatives increase the legitimacy of education c hoice. Wong, K.K. and Langevin

PAGE 29

24 (2007) found the number of private schools to posit ively correlate with the likelihood that a state would adopt charter school legislation. Zha ng and Yang’s 2008 research, however, found the number of private schools to be negativel y correlated with charter school creation. They suggest that “charter schools are su bstitutes of private to some degree” (585). Presence of Public School Options The presence of other public school options appear s to increase the likelihood of charter openings. A study examining showed the pre sence of magnet schools is a positive predictor for chartering activity (Rincke’ s 2007). Magnet schools are district-run schools of choice which are generally established f or voluntary racial integration purposes. According to Public School Review, there are 24 magnet schools in Colorado (n.d.). Renzulli and Roscigno (2005) found that the passage of a statewide open enrollment law increased the likelihood of charteri ng. Even the presence of other charter schools appears to predict that more will open. Renzulli and Roscigno (2005) suggest that there is some degree of path dependence in this area, “[O]pen charter schools and the increase d number of states with charter school laws may increase the legitimacy for new charter le gislation and the creation of more charter schools” (348). Renzulli (2005), however, found that the number of existing charter schools in a district had a negative impact on submissions. This suggests that as the number of charter schools increases, competitio n for a finite number of students also increases. The number of charter schools in the sta te, on the other hand, had a positive impact, suggesting that existing charter schools in other districts increases the legitimacy

PAGE 30

25 of charter schools without reducing the capacity wi thin the district for more charter schools. As with private schools, the presence of n earby charter schools appears to raise interest and demand for charter schools. Adjacency to Districts with Charter Schools As previously mentioned, students of today are more likely to attend schools outside of their residential district than they had been in the past. Because districts lacking charter schools may see enrollments decline as students choose to attend charter schools in nearby districts, there exists an incent ive to offer their own charter schools. As parents and district boards become aware of nearby district’s charter schools, it seems likely that board members would experience pressure to authorize charter schools within the district. Witte, Scholomer, and Shober’s (2007) research in W isconsin shows a positive correlation between proximity to a district with ch arter schools and the number of charter schools. Similar diffusion effects have also been found in in California (Rincke 2007) and Florids (Zang and Yang 2008) Further research on inter-district school choice and charter school diffusion conclude that policy maker s are significantly influenced by their peers’ actions in nearby districts (Rincke, 2006, 2 007). Political Climate According to DeBray-Pelot, Lubienski, and Scott (20 07), charter schools represent the “marriage of market-oriented neoliber als working from a series of statelevel think tanks and progressive reformers committ ed to creating options with a public

PAGE 31

26 system” (p. 212). Although Republicans tend to be m ore open to school choice; charter schools enjoy support from both sides of the aisle (Kirst 2007). The research on the impact of political climate on charter diffusion is mixed. Renzulli (2002) hypothesized that a higher presence of Democrats would correlate with more chartering activity. She found that an increas e in registered Democrats increased the number of submissions, but that other factors were more determinative. A follow-up study found that state level political factors had no impact on whether a state adopted a charter school law but having a Republican governor decreased the “expected number of charter school foundings” (Renzulli and Roscigno 20 05, 358). Shober, Manna, and Witte (2006) found that having large Republican representation in the state legislature was correla ted with more charter schools in a state and a Republican governor even more so. Wong and La ngevine (2007) found that states with a Republican governor were more likely to adop t a charter school law. Zhang and Yang (2008) found that the percentage of Democratic voters was correlated with an increase in the number of charte rs to a point. When the percentage of Democrats exceeded 79.5 percent, the effect was rev ersed. This would suggest that charter schools enjoy bipartisan support except in heavily Democratic districts. Union opposition to charter schools is another vari able identified by researchers that impacts both charter law adoption and charter activity. Minstrom (1997) found union strength reduced the likelihood of charter law appr oval as did Renzulli and Roscigno (2005). However, they also found that the National Education Association’s presence had

PAGE 32

27 a marginally positive impact on the founding of cha rter schools once a law had passed. Stoddard and Corcoran (2006) had the same findings. Research by the Thomas B. Fordham Institute ranked the strength of ColoradoÂ’s teacher union 35th among the states and the District of Columbia (Win kler, Scull, and Zeehandelaar 2012). According to the report, union strength is not very strong in Colorado. Because state level union strength is wea k and district level union information would be difficult to obtain, this variable will no t be analyzed in the thesis. School District Capacity School district capacity refers to several district level factors including population density, presence of policy innovators, the superin tendent, enrollment and enrollment growth, and funding. Population Density RenzulliÂ’s (2005) research showed that urban distri cts had more charter school submissions than suburban or rural districts. Total population and population density both impact the capacity of a district to open a charter school. The presence of advocacy groups will likely be stronger in urban and suburba n areas. Coalitions that generally support the opening of charter schools include pare nts dissatisfied by local schools, business and community organizations, state charter school associations, national advocacy groups with local affiliates, real estate developers, faith-based organizations, higher education institutions, and foundations and philanthropists (Kirst 2007). Presence of Innovators

PAGE 33

28 Minstrom’s (1997) research found that the presence of policy entrepreneurs significantly increases the rate at which the publi c approves of school choice as a policy innovation. These groups work with policy makers t o effect change (Minstrom and Vergari, 1998). He notes that “[P]olicy entreprene urs operating at the state level will most often develop their ideas for policy innovatio n through their conversations and interactions with members of interstate and external policy networks ” (1997, p.130, italics in the original). The Superintendent In their case studies, Teske et al. (2000) found mi xed support for charter school initiatives from school superintendents. Some exhib ited antagonistic relationships with charter schools, while others were much more positi ve. Even in districts where district schools faced considerable competition from charter schools, some superintendents had positive views of charter schools. Teske et al. con cluded that “[T]he attitude of the district superintendent and, through the superinten dent, the attitudes of other high level administrators seem to be more a function of their individual beliefs” rather than the market share of the charter or district schools (p. 10). In all 19 case studies conducted by Witte, Scholomer and Shober (2007), there were entr epreneurial administrators, school board members, parents, and teachers who wanted to open charter schools. Zhang and Yang’s (2008) research showed that the pr esence of an appointed superintendent versus an elected superintendent was positively with higher rates of chartering. They suggest that this is because “appo inted superintendents are less distracted by electoral politics in the community a nd more motivated by a desire to

PAGE 34

29 enhance their reputation and labor-market value by adopting innovations. Alternatively, it may indicate that appointed superintendents treat c harter schools as a dumping ground for their at-risk students so that they can boost test schools in current schools” (585). All district superintendents are appointed in Colorado. Enrollment and Enrollment Growth High enrollment growth could lead school district s to open charter schools as a means of increasing district capacity to absorb new students. Districts in suburban areas experiencing high growth in student enrollment beco me receptive to charter schools to address overcrowding (Pushpam, 2002). A study by R PI International for U.S. Department of Education examined the impact of cha rter schools in 49 districts in five states –Arizona, California, Colorado, Massachusett s, and Michigan (Ericson, J. et al., 2001). The researchers found, “In 35 percent of the total districts, administrators reported that charter schools had relieved overcrowding or t he pressure to construct new facilities caused by an increasing student population” (p.14). The research on enrollment growth and charter activ ity is lean. Zhang and Yang, however, found no association between enrollment gr owth and chartering (2008). Higher enrollment, in general though, has been shown to co rrelate with higher chartering activity. Witte, Schlomer, and Shober (2007) found that districts with higher enrollments were more likely to charter schools. They suggest t hat larger districts have larger administrative infrastructure to support the charte ring process. Renzulli (2005) found that the larger the size of t he district and the higher the number of district administrators the higher the ch arter school submissions. She

PAGE 35

30 hypothesizes that high levels of bureaucracy increa se inefficiencies and red tape and will therefore increase the number of submissions. Funding Witte, Schlomer and Shober (2007) found that distri cts with more federal revenues were more likely to start chartering schoo ls and attributed it to the fact that these districts have more low income students. In t heir analysis of factors related to charter law adoption, Wong and Langevin (2007) foun d lower classroom spending associated with charter law adoption. Renzulli (200 2) also found that districts with lower state and local funding had higher charter school s ubmissions.

PAGE 36

31 CHAPTER III METHODOLOGY This study will test for relationships between the size of the charter school sector and ten variables associated with student demograph ics and district factors. Although the number charter schools open in the 2013–2014 is ava ilable, academic data are available only for the 2012–2013 school year. Therefore all d ata used in the study come from the 2012–2013 school year. Hypotheses Level of Need One of the primary reasons for founding charter sch ools according to advocates is that they can serve disadvantaged children who are not well served in the traditional system. The Colorado Charter School Acts of 1993 st ates: (2) The general assembly further finds and declares that this part 1 is enacted for the following purposes: a) To improve pupil learning by creating schools wi th high, rigorous standards for pupil performance; (b) To in crease learning opportunities for all pupils, with special emphasis on expanded learning experiences for pupils who are identified as academically low achieving; (Colorado Charter Schools Act (1993)

PAGE 37

32 For the purposes of this study, level of need is de fined as the percentage of disadvantaged students (minority and low income stu dents) and student achievement as measured by district accreditation status. H1: School districts with a higher percentage of nonw hite students will have a greater quantity of charter schools.1 H2: School districts with a higher percentage of low income students will have a greater quantity of charter schools.2 H3: School districts with a lower accreditation plan will have a greater quantity of charter schools.3 School Choice Environment School choice environment refers to the level of sc hool choice within the district. All districts are impacted by Colorado’s open enrol lment law. Some districts have private schools. As the researchers noted, private schools act as competition for public schools and as a pool of potential student recruits. Parent s who enroll their children in private These data are found at the CDE website at www.cde. state.co.us/cdereval/pupilcurrentdistrict.htm Data set for minority students from CDE that included al l nonwhite students. Percentages have been transformed into decimals for data processing. 2 These data are found at the CDE website at www.cde.state.co.us/cdereval/pupilcurrentdistrict.h tm This CDE data includes students who are eligible fo r the Federal Free and Reduced Lunch Program. In 2012, free lunch eligibility was set was 130 percen t of the federal poverty level (approximately $29,9 65 for a family of four) and at 185 percent of the federal poverty level (approximately $42,643 for a family of four) for a reduced price lunch. Percentages have been transformed into decimals for data processing. 3 The state of Colorado assigns districts an accredit ation category based on the overall District Perfor mance Framework score that includes academic achievement, academic growth, academic growth gaps, and postsecondary and workforce readiness. Accreditatio n categories are as follows: Accredited with Distinction, Accredited, Accredited with Improvemen t Plan, Accredited with Priority Improvement Plan, and Accredited with Turnaround Plan. To simplify, d istricts with accreditation status of Accredited w ith Distinction, Accredited, Accredited with Improvemen t Plan were coded as “0” and those with Priority Improvement Plan, and Accredited with Turnaround Pl an were coded as a “1.”

PAGE 38

33 schools are accustomed to choosing schools. Some of the research shows that high number of private schools is associated with a grea ter number of charter schools. District adjacency to other districts with charter schools is also included. If a sufficient number of parents leave the district to attend a charter school in another district, district boards will feel pressure to cha rter schools of their own. Political climate is also included in this category as a null hypothesis. Colorado history shows bipartisan support for charter school s so it is expected to have no impact on charter diffusion. The weakness in using countywide voter registration statistics is that they do not capture political diversity within each county. For example, political registration in the Manitou Springs school district and the Colorado Springs 11 school district located near Peterson Air Force Base are l ikely different. Both districts are assigned 22 percent Democratic registration because that is the county level. Similarly, there is considerable diversity in Jefferson and Ar apahoe Counties, as well, that is not reflected in the countywide percentages. Finally, t he rising number of unaffiliated voters, dilutes the usefulness of this variable. Independen ts, which make up about a third of voters in Jefferson and Arapahoe Counties vote for Democrats, Republicans, and other parties. Other researchers may consider using more complicated variable to better capture political culture but none has yet embarked down that road. H4: School districts with a higher number of private (independent and faith-based) schools will have a greater quantity of charter sch ools.4 4 Data for independent and faith-based schools are f ound at the CDE website at www.cde.state.co.us/cdereval/pupilcurrentnonpublic. htm. The data set includes the total number of priv ate schools. Because research variable on the whether t he number of total private schools, number of secul ar

PAGE 39

34 H5: School districts adjacent to districts with chart er schools are more likely to have charter schools.5 H6: The Percentage of Democratic registered voters wi ll have no impact on the quantity of charter schools in school districts.6 District Capacity District capacity refers to the capacity to open ad ditional schools. Districts with higher student enrollments can authorize more schoo ls. Such district will have both the demand and the capacity to operate additional schoo ls. Districts experiencing enrollment increases over the past few years will have additio nal capacity. Districts in urban and suburban areas should have more capacity to open ad ditional schools than rural schools. Finally, districts with lower per-pupil funding sho uld be more likely to charter schools as the research findings predict. H7: School districts with higher levels of enrollment growth will have a greater quantity of charter schools.7 H8: School districts with higher levels of enrollment will have a greater quantity of charter schools.8 private schools, or number of secular private corr elated to chartering activity the study also tests this hypothesis for secular and faith-based schools. 5 These data are found at the CDE website at http://www.cde.state.co.us/cdeedserv/coloradoschool districtsmap. I identified charter districts on the map and noted which non-charter districts were next to charter districts. Districts adjacent to charter di stricts are coded as a 1 and districts that do not touch anothe r district with charter schools are coded as 0. 6 These data are found at Colorado Secretary of Stat e’s office at www.sos.state.co.us/pubs/elections/VoterRegNumbers/ 2013/June/VotersByPartyStatus.pdf. Each school district fits within the boundaries of a county. Mo st counties have more than one district usually bet ween two and six. The average is 2.7. Outliers El Paso C ounty and Weld County have 15 and 12 twelve school districts respectively. Denver, Douglas, and Jeffer son, three of largest school districts, are countywide. 7 These data are found at Membership Trends District Totals 2006–2013 at CDE website at http://www.cde.state.co.us/cdereval/pupilcurrentdis trict.

PAGE 40

35 H9: School districts with a Denver Metro, Urban/Subur ban, and Outlying City designation will have a greater quantity of charter schools than Outlying Town, and Rural school districts.9 H10: School districts with lower levels of total per pup il funding will have a higher quantity of charter schools.10 Dependent Variables The study measures the quantity of charter schools in districts using two dependent variables to be analyzed separately. The use of the two variables is useful for validation of the results. The first dependent variable is the number of chart er schools in a district. This is the most common dependent variable used in the lite rature. The reason for using a second variable, that of stu dent population, is that some districts may appear to be enthusiastic chartering authorities by the absolute number of charter schools, but may in fact, serve a proportio nally small number of kids. Researchers Teske, Schneider and Cassese noted in their 2005 st udy of school district authorizers that many school boards view charter schools as competit ors for their own schools and attempt to limit charter school creation and growth School boards have held a monopoly 8 These data are found at Membership Trends District Totals 2006–2013 at CDE website at http://www.cde.state.co.us/cdereval/pupilcurrentdis trict 9 These data are at CDE website at http://www.cde.state.co.us/sites/default/files/docu ments/cdereval/download/pdf/districtslistedbysettin g.pdf CDE designates each school district as one of the f ollowing: Denver Metro; Urban/Suburban; Outlying City; Outlying Town; and Rural. To simplify, Denver Metro, Urban/Suburban, and Outlying City districts were coded with “1” and Outlying Town; and Rural di stricts coded as a “0.” This includes state and loc al funding. 10 Data represent Total Formula Funding and are found at CDE website at http://www.cde.state.co.us/cdefinance/dbydfy13. This includes state and local funding.

PAGE 41

36 for a long time and do not wish to give it up. Boar ds, however, experience pressure to respond to local political forces to open schools. The researchers hypothesize that school boards are more likely to favor niche schools that serve special populations and do not compete with the school district’s general educatio n schools. A district may charter a high number of charter schools but because they are primarily niche schools, the district actually serves a relatively low percentage of dist rict students. The study analyzes open charter schools instead of charter schools application submissions. Researchers examining charter school d iffusion use both. Of the five studies examining charter school diffusion, two stu dies tracked charter school application submissions (cite these two here) and three used ch arter school openings as dependent variables (cite these three here) (Renzulli, 2005, Renzulli and Roscigno, 2005, Renzulli, 2002, Teske, Schneider and Cassese, 2005, Witte, Sc holomer, and Shober, 2007). Renzulli (2005) used application submissions becaus e examination of charter school openings would exclude the number of rejected appl ications “and underestimate the initiation patterns for charter schools and ignore important information about how organizational environments foster or subdue effort s at innovation” (Renzulli 2005, p. 4). She found that between the years 1991 and 1998, 1,1 47 applications were submitted and 418 schools received a contract and opened. Indeed, using openings alone could underestimate the interest in establishing charter schools. In Colorado, however, one may not need to gather su bmission data to capture the breadth of the charter school initiation efforts, g iven the ease with which a rejected applicant may appeal to the State Board of Educatio n. Over the past seven years (2006– 2012), there have been 32 appeals to the State Boar d. In 12 cases, the State Board upheld

PAGE 42

37 the school boardÂ’s decision. In 10 cases, the schoo l board was ordered to reconsider. In four cases on second appeal, the State Board ordere d the school board to allow the school to be opened. Parties settled the issue and dropped the appeal in four cases, and in two cases, the appeal was rejected because of failure b y the applicant to follow the appeals process. In all, 83 charter schools opened during those years (Colorado Department of Education 2013). Moreover, Colorado school districts may not be open ly hostile to charter schools, otherwise they risk losing their Exclusive Charteri ng Authority granted by the State Board. In the 10 school districts without exclusive authority, the stateÂ’s alternative authorizer, the CSI, may open schools without distr ict approval. Given the rate of charter school openings and succe ssful appeals, this study will use the number of existing schools, rather than sub missions as its dependent variable. Quantitative Methods First, a multivariate regression model is used for the second dependent variable, percentage of students educated in district charter schools. A second analysis uses negative binomial regression model using a simplifi ed dependent variable namely the number of charter schools. This modeling approach i s suitable because of the distribution of this dependent variable indicating the raw numbe r of charter schools. This approach has similarly been employed by other scholars Zhang and Yang (2008) and Renzulli (2002 and 2005). Model 1: Multivariate Analysis was used with depend ent variable 2 (percentage of students in districtsÂ’ charter schools) and 10 inde pendent variables. Statistics for private

PAGE 43

38 schools, secular private schools, and faith-based p rivate schools were analyzed separately because of high covariance levels. Model 2: Negative Binomial Regression was used with the first dependent variable (number of charter schools) and 10 independent vari ables. Minority and low income students were analyzed together and separately.

PAGE 44

39 CHAPTER IV RESULTS AND DISCUSSION Descriptive Statistics Table 5 provides descriptive statistics for the dep endent and independent variables used in this analysis. Table 5: Descriptive Statistics Dependent Variables Mean Std. Dev. Min. Max. Number of Charter Schools in District 0.87 3.28 0 37 % of Charter School Students in District 0.02 0.05 0 0.25 % Minority Students 0.33 0.22 0 0.95 % Poor Students 0.47 0.18 0.07 0.90 High District Accreditation Status 0.88 0.33 0 1 Low District Accreditation Status 0.12 0.33 0 1 Secular Private Schools 0.61 2.20 0 15 Faith-based Private Schools 0.94 3.05 0 28 All Private Schools 1.56 4.95 0 41 Adjacency to Districts with Charter Schools 0.58 0.50 0 1 Percentage of Democrats 0.28 0.12 0.1 0.72 Change in Enrollment -0.03 0.16 -0.79 0.45 Enrollment 4780.14 12539.58 10 85508 More Urban/Suburban 0.24 0.43 0 1 More Rural 0.76 0.43 0 1 Funding 9732.59 2545.6 7207.46 16539.42

PAGE 45

40 Correlations As Table 6 shows, several of the independent variab les are highly correlated within the sample used for this study. Specifically the demographic factors of economic affluence and racial minority population are somewh at correlated with a correlation coefficient of 0.63. Variables indicating the numb er of private, faith-based, and secular private schools are highly correlated. Enrollment a nd the number of private schools are also highly correlated. This is not surprising sinc e large, populous districts are more likely to have more private schools than smaller, l ess populous areas. Table 6: Correlation Matrix of Independent Variable s Mn Std Pr Std High Acc Low Acc Sec Schl FB Schl Pri Schl Adj Dem Enr Chg Enr Urb Sub Rur Fund Min Std 1.00 Pr Std .63 1.00 High Acc -.45 -.31 1.00 Low Acc .45 .31 -1.00 1.00 Sec Schl .10 -.15 -.05 .05 1.00 FB Schl .21 -.01 -.18 .18 .78 1.00 Pri Schl .17 -.07 -.13 .13 .92 .96 1.00 Adj -.04 -.28 -.01 .00 .23 .24 .25 1.00 Dem .47 .30 -.23 .22 .17 .14 .16 -.01 1.00 Enr Chg .15 -.15 .10 -.10 .18 .18 .19 .24 .02 1.00 Enr 0.20 -.13 -.13 .13 .77 .89 .89 .29 .13 .02 1.00 Urb Sub .32 -.03 -.19 .18 .45 .50 .50 .30 .10 .13 .57 1.00 Rur -.32 0.03 .19 -.19 -.45 -.50 -.50 -.30 -.10 .10 -.57 1.00 1.00 Fund -.30 0.13 .13 -.13 -.23 -.25 -.25 -.37 -.09 -.10 -. 30 -.47 .46 1.00

PAGE 46

41 Model 1: Percentage of District Charter Students an d 10 Independent Variables The first analysis presented in this section is a m ultivariate regression of Dependent Variable 2 (students educated in district charter schools as a percentage of all district students) and 10 independent variables. Pr ivate schools, faith-based private schools and secular private schools were analyzed s eparately as are poor and minority students. In Model 1, district enrollment, urbanici ty, and district adjacency to at least one other district with charter schools were correlated with a higher percentage of district charter school students. When all private schools a nd secular private schools are analyzed in this model, there is a negative correlation betw een the number of schools and percentage of district charter school students. Whe n just secular private schools are analyzed, the percentage of poor students in the di strict is found to be negatively associated with the size of the charter school sect or in the district. Similarly, when the percentage of poor and minority students are analyz ed separately, the percentage of poor students is negatively correlated at a statisticall y significant level of .10.

PAGE 47

42 Table 6: Regression Results for Model 1, Dependent Variable 2 (percentage of students in district charter schools) for all Colorado schoo l districts in 2012-13 Independent Variable Minority Students .00 (.02) .00 (.02) .00 (.02) % of Poor Students -.03 (.02) -.04** (.03) -.03 (.03) Low Accreditation .00 (.01) -.00 (.01) .00 (.01) # of Private Schools -.00* (.00) ------------# of Secular Private Schools ------.01* (.00) .-----# of Faith-based Private Schools -------------.02 (.03) Adjacent to Charter Districts .01* (.01) 01* (.01) .01* (.01) % of Democratic Registration -.01 (.03) -.00 (.03) -.02 (.03) Enrollment Change 2008-2013 .02 (.02) .02 (.02) .02 (.02) District Total Enrollment 2.12* (5.55) 1.73* (4.16) 1.76* (5.66) Greater Urbanicity .03* (.01) .03* (.01) .03* (.01) District Funding -1.42 (1.48) -1.36 (1.48) -1.44 (1.50) Constant .03 (.02) .03 (.02) .04 (.02) Statistically significant at .05. ** Statistically significant at .10.

PAGE 48

43 Model 2: Number of Charter Schools and 10 Independe nt Variables In Model 2, Negative Binomial Regression was used w ith the first dependent variable (number of charter schools) and 10 indepen dent variables. As Table 7 shows, only the results for total number of private school s are shown. Analysis of secular private schools did not produce different findings and data that included faith-based private schools did not converge in Stata. In this model, t he number of private schools and the degree of enrollment change are correlated with gre ater numbers of charter schools at a .10 statistically significant level. Adjacency to c harter school districts, total enrollment, and greater urbanicity are positively correlated an d funding is negatively correlated. No correlation was found when poverty and minority rat es were analyzed separately. Table 7: Negative Binomial Regression Independent Variable Minority Students 1.16 (1.29) % of Poor Students -.22 (1.26) Low Accreditation .07 (.44) # of Private Schools .04** (.02) Adjacent to Charter Districts 2.00* (.74) % of Dem Registration -.79 (1.48) Enrollment Change 1.75** (.99) District Total Enrollment .00* (8.42) Greater Urbanicity .91* (.36) District Funding -.00* (.00) Constant 2.30 (2.32) Statistically significant at .05. ** Statistically significant at .10.

PAGE 49

44 Discussion As Table 7 shows, several of the hypotheses were co nfirmed. Table 7: Hypotheses and Results Hypotheses Model 1 Model 2 H1: School districts with a higher percentage of nonw hite students will have a greater quantity of charter schools. No correlation No correlation H2: School districts with a higher percentage of low income students will have a greater quantity of charter sc hools. No correlation (in most cases) No correlation H3: School districts with a lower accreditation plan will have a greater quantity of charter schools. No correlation No correlation H4: School districts with a higher number of private schools will have a greater quantity of charter schools. Negative correlation in most cases Positive correlation H5: School districts adjacent to districts with chart er schools are more likely to have charter schools. Positive correlation Positive correlation H6: The Percentage of Democratic registered voters wi ll have no impact on the quantity of charter schools in school districts. No correlation found No correlation found H7: School districts with higher levels of enrollment growth will have a greater quantity of charter schools. No correlation found Positive correlation H8: School districts with higher levels of enrollment will have a greater quantity of charter schools. Positive correlation Positive correlation H9: School districts with a Denver Metro, Urban/Subur ban, and Outlying City designation will have a greater quant ity of charter schools than Outlying Town, and Rural school distri cts. Positive correlation Positive correlation H10: School districts with lower levels of total per pup il funding will have a higher quantity of charter schools. Negative correlation Negative correlation

PAGE 50

45 Level of Need Hypotheses in this group predicted that the percent age of minority and low income students and lower academic achievement woul d positively correlate with chartering activity. The majority research showed a link between minority students and lower academic achievement with chartering and the research is mixed on the impact of low income students. In this study, the two models showed no statistically significant findings on level of need except in one case when o nly data for secular private schools was included. In this case, a negative correlation was found. It is possible that the racial and economic demographics in Colorado differ than i n states studied by other researchers. Many of the ColoradoÂ’s poorest distric ts are not urban districts but rural districts in the stateÂ’s San Luis Valley and other rural areas. These districts do not have charter schools. It is also possible that district accreditation status, a holistic measure of school quality and outcomes, produces different res ults than subject proficiency rates. School Choice Environment The hypotheses regarding school choice environment predicted that the number of private schools would be associated with more chart er schools. The models, however, produced contradictory results. Model 1 found a neg ative association and Model 2, a positive association. The existing research is also mixed for this variable. As was expected, the percentage of Democratic registration is not associated with charter activity. Charter schools have significant bipartis an support in Colorado which may explain why some states produce different findings on the impact of partisanship on charter school growth. A positive association betwe en adjacency to districts with charter

PAGE 51

46 schools concurs with existing research. Most Front Range districts, where charter activity is the greatest, abut one another, so this finding is not surprising. District Capacity The hypotheses in this group concern enrollment, en rollment growth, funding, and urbanicity. Both models showed a positive assoc iation between enrollment and urbanicity and greater charter activity. This findi ng concurs with existing research. Only Model 2 found an association between enrollment gro wth and more charter schools. Both models found an association between lower funding a nd great charter activity which aligns with existing research. ColoradoÂ’s highest f unded districts are its rural districts, so the finding is not surprising. There is also a possibility of reverse causation in this analysis as it pertains to enrollment and dependent variable 1 (Model 1), the number of charter schools. Since charter schools are schools of choice, they draw st udents from outside of the district. High performing charter schools attract students fr om outside of the district and thus increase district enrollment and enrollment growth. Thus the dependent variable may cause increases in the variables regarding enrollme nt and enrollment growth. Conclusion Using two statistical models and a variety of data, this thesis shows enrollment, urbanicity and adjacency to a chartering district a re correlated with greater chartering activity. Per pupil funding is negatively correlate d and there does not appear to be a correlation between the number of charter schools a nd district academic achievement or political affiliation of district voters. The prese nce of private schools may have a negative

PAGE 52

47 or positive impact on chartering activity. A more s ophisticated statistical model may be able to solve the differing results in this study. The findings in this study provide some insight int o why some districts have more charter schools than others. Though urban, adjacent to charter districts, and similarly funded to other Denver Metro areas, the Englewood a nd Sheridan districts have no charter schools. A potential reason could be that t heir student populations are small; they are more similar to rural districts in terms of siz e. The paucity of charter schools in isolated, rural, low population areas accurately pr edicts fewer charter schools on the eastern plains and mountain regions. The results, however, do not illuminate reasons for why Cherry Creek has so few charter schools compared to other districts of its size and funding level. Boulder Valley has a similar funding level and far fewer students but far more charter schools. In the same way, Denver has a similar student population s ize and higher per-pupil funding than Jefferson County, yet has twice as many charter sch ools. As with many studies, the results prompt new quest ions. Are there additional quantifiable variables that can account for such di fferences or do the differences lie in characteristics not measurable by numbers? The ans wer may lie in the qualities of superintendents and boards, the presence or absence of policy innovators, parent demand, or the strength of charter advocates and opponents. This thesis paves the way for other researchers to ask new questions.

PAGE 53

48 REFERENCES Benigno, P. and Morin, K. (2013) On the road of innovation: Colorado’s charter scho ol law turns 20 (Independence Institute) Retrieved from the Indep endence Institute website at http://www.i2i.org/files/file/IP-4-2013.pdf Barton, P.E. Why does the gap persist? Association for Supervision and Curriculum Development Educational Leadership 62 (3), 8-13. Bierlein Palmer, L. and Gau, R. (2005) Charter scho ol authorizing: policy implications from a national study. Phi Beta Kappan 86 (5), 353-357. Budde, R. (1988). Education by charter: Restructuri ng school districts. San Francisco: WestEd. Buddin, R. (2012) The impact of charter schools on public and private school enrollments. (Cato Institute Policy Analysis No. 707). Retrieve d from the Cato Institute website: http://www.cato.org/sites/cato.org/files/pubs/pdf/P A707.pdf Carpenter, D. and Kafer, K. (2013) The State of Charter Schools Colorado Department of Education, 1-82. Retrieved from the Colorado Dep artment of Education website at http://www.cde.state.co.us/cdechart/download/STATER EPORT.pdf. Carpenter, D.M., and Kafer, K (2012). A History of Private School Choice. Peabody Journal of Education: Issues of Leadership, Policy, and Organizations, (87) 3. 336-350 doi:10.1080/0161956X.2012.679587 Chakrabarti, R. and Roy, J. (2011). Do charter scho ols crowd out private school enrollment? Evidence from Michigan. Federal Reserve Bank of New York Staff Report no. 472, 1-54. Retrieved at http://www.newyorkfed.org/research/staff_reports/sr 472.pdf Clark, M.A. et al. (2011). Do Charter Schools Impro ve Student Achievement? Evidence from a National Randomized Study (Mathematica Poli cy Research). Retrieved from the Mathematica website http://www.mathematica mpr.com/publications/PDFs/education/charterschools_ WP.pdf Colorado League of Charter Schools (n.d.) Facts and figures about Colorado charter schools. Retrieved at the Colorado League of Charte r Schools website at http://www.coloradoleague.org/colorado-charter-scho ols/20th-anniversary-factsand-figures.php Colorado Charter Schools Act, Colorado Revised Stat utes § 22-30.5-101 et. seq (1993). Colorado Charter School Institute (n.d.) CSI at a glance. Retrieved at Colorado Charter School Institute website at

PAGE 54

49 http://www.csi.state.co.us/UserFiles/Servers/Server _2345071/Image/CSI%20at% 20a%20glance-1314.pdf Colorado Department of Education (n.d.) Colorado Sc hool Districts Listed by Setting. Retrieved at the Colorado Department of Education w ebsite at http://www.cde.state.co.us/sites/default/files/docu ments/cdereval/download/pdf/di strictslistedbysetting.pdf Colorado Department of Education Membership (n.d.). Trends for District Totals (20062013) Retrieved at the Colorado Department of Educa tion http://www.cde.state.co.us/cdereval/fall2013members hiptrendsfordistricttotals200 6-2013xls Colorado Department of Education (n.d.) Summary of School Districts Regarding Exclusive Chartering Authority Status 2012-13 Schoo l Year. Retrieved at the Colorado Department of EducationÂ’s website at http://www.cde.state.co.us/sites/default/files/ECA_ 2012%20Final_0.pdf Colorado Department of Education (2013) CDE 2013 Di strict Accreditation Ratings 2010-2013. Retrieved at the Colorado Department of Education website at http://www.cde.state.co.us/accountability/performan ceframeworkresults#twentyt welve Colorado Department of Education. (2013) Students A ttending Public Schools Not in Parent's District of Residence. Retrieved at the Co lorado Department of Education website at http://www.cde.state.co.us/cdereval/pupilcurrentdis trict#sthash.usaz1qCy.dpuf Colorado Department of Education. (2013) Pupil Coun ts by Year 1993 through 2013. Retrieved at the Colorado Department of Education w ebsite at http://www.cde.state.co.us/datapipeline/pupilctbyyr 1993through2013xls Colorado Department of Education. (n.d.) Summary of School Districts Regarding Exclusive Chartering Authority Status 2012-13 School Year. Retrieved at the Colorado Department of Education website at http://www.cde.state.co.us/sites/default/files/ECA_ 2012%20Final_0.pdf Colorado Department of Education (n.d.). Nonpublic Schools Fall 2011. Retrieved at the Colorado Department of Education website at http://www.cde.state.co.us/sites/default/files/docu ments/choice/download/201112nonpublicdirectory.pdf

PAGE 55

50 Conlan, S. et al. (2011) Building Charter School Qu ality in Colorado Building Charter School Quality. Retrieved at the Building Charter S chool Quality website at http://charterschoolquality.org/media/1178/BCSQ_Bui ldingQualityColorado.pdf DeBray-Pelot, E.H., Lubienski, C.A., and Scott, J.T (2007). The institutional landscape of interest group politics and school choice. Peabody Journal of Education, 82 (23), 204-230. doi:10.1080/01619560701312947 Department of Agriculture Child Nutrition Programs Income Eligibility Guidelines, 57 C.F.R. 17005 (2012). Retrieved at http://www.gpo.go v/fdsys/pkg/FR-2012-0323/pdf/2012-7036.pdf Ericson, J. et al. (2001). Challenge and Opportunit y: The Impact of Charter Schools on School Districts. (RPI International for the U.S. D epartment of Education Office of Educational Research and Improvement National St udy of Charter Schools). Retrieved at the US Department of Education website at http://www2.ed.gov/rschstat/eval/choice/district_im pact.doc Friedman Foundation for Educational Choice. (n.d.) School Choice Programs. Retrieved at the Friedman Foundation for Educational Choice w ebsite at http://www.edchoice.org/School-Choice/School-Choice -Programs Jeynes, W. H. (2003). Religion, education, and acad emic success. Greenwhich, CT: Information Age. Jeynes, W. H. (2007). American educational history: School, society, and the common good. Thousand Oaks, CA: Sage. Carpenter, D.M., and Kafer, K (2012). A History of Private School Choice. Peabody Journal of Education: Issues of Leadership, Policy, and Organizations, (87) 3. 336-350 doi:10.1080/0161956X.2012.679587 Kirst, M. W. (2007). Politics of charter schools: C ompeting national advocacy coalitions meet local politics. Peabody Journal of Education, 82 (2-3), 184-203. doi:10.1080/01619560701312939 Lake, R. Ed. (2012) A Balanced Look at American Cha rter Schools in 2012. Center on Reinventing Public Education University of Washingt on Bothell. Retrieved for the Center on Reinventing Public EducationÂ’s websit e at http://www.crpe.org/sites/default/files/pub_hfr12_m ay13.pdf

PAGE 56

51 Minstrom, M. and Vergari, S. (1998). Policy network s and innovation diffusion: The case of state education reforms. The Journal of Politics, 60 (1), 126-148. doi:10.2307/2648004. Minstrom, M. (1997). Policy Entrepreneurs and the d iffusion of innovation. American Journal of Political Science, 41 (3), 738-770. Mintrom, M. (2000). Policy Entrepreneurs and School Choice. Washington D.C.: Georgetown University Press. Mitchell, N. (2011, January 18). More students choi cing out of district. Chalkbeat Colorado. Retrieved from http://co.chalkbeat.org/20 11/01/18/more-studentschoicing-out-of-district/ National Alliance for Public Charter Schools. (n.d. ). Charter schools 101: The most frequently asked questions. Retrieved at the Nation al Alliance for Public Charter Schools at http://www.publiccharters.org/About-Charter-Schools /FrequentlyAsked-Questions.aspx National Alliance of Public Charter Schools (n.d.) Details from the Dashboard: Charter School Race/Ethnicity Demographics Retrieved from National Alliance of Public Charter Schools at http://www.publiccharters.org/wpcontent/uploads/2014/01/NAPCS-2010-2011-Race_Ethnic ity-Details-from-theDashboard_20120516T152831.pdf National Association of Public Charter Schools (n.d .) Details from the Dashboard:Charter Schools by Geographic Region Retrieved from the Nat ional Association of Public Charter SchoolsÂ’ website at http://www.publiccharte rs.org/wpcontent/uploads/2014/01/Geographic-Location-Details -from-the-DashboardReport_20120224T143955.pdf Public School Review (n.d.). Colorado Magnet Public Schools. www.publicschoolreview.com/state_magnets/stateid/CO Pushpam, J. (2002) The Approval Barrier to Suburban Charter Schools. (Thomas B. Fordham Foundation) Retrieved from the Thomas B. Fo rdham Foundation website at http://www.edexcellence.net/sites/default/files/pub lication/pdfs/approvalbarrier_1 0.pdf Raymond, M. et al. (2013) National Charter School Study. Center for Research on Education Outcomes at Stanford University. Retrieve d from the CREDO website at http://credo.stanford.edu/documents/NCSS%202013%20F inal%20Draft.pdf

PAGE 57

52 Renzulli, L. A. (2002). Entrepreneurial ambitions i n the public sector: A random effects model of the emergence of charter schools in North Carolina. Education Policy Analysis Archives, 10 (19), 1-25. Retrieved from http://epaa.asu.edu/epaa/v10n19/ Renzulli, L. A. (2005). Organizational environments and the emergence of charter schools in the United States. Sociology of Education 78 (1), 1-26. doi: 10.1177/003804070507800101. Renzulli, L.A. and Roscigno, V.J. (2005). Charter s chool policy, implementation, and diffusion across the United States. Sociology of Education 78 (4), 344-366. doi:10.1177/003804070507800404. Rincke, J. (2006). Policy innovation in local juris diction: testing for neighborhood influence in school choice policies. Public Choice 126 (1), 189-200. doi:10.1007/s11127-006-9022-3. Rincke, J. (2007). Policy diffusion in space and ti me: the case of charter schools in California school districts. Regional Science and Urban Economics (37), 527-541. doi: 10.1016./j.regsciurbeco.2007.02.001. Shanker, A. (1988). Restructuring our schools. Peabody Journal of Education 65 (3), 88100. doi:10.1080/01619568809538615 Schaeffer, A.B. (2012). The Charter School Paradox (Cato Institute Policy Analysis No. 707). Retrieved from the Cato Institute website: http://www.cato.org/sites/cato.org/files/pubs/pdf/c harter-school-paradox.pdf Shober, A.F., Manna, P., and Witte, J.F. (2006) Fle xibility Meets Accountability: State Charter School Laws and Their Influence on the Form ation of Charter Schools in the United States Policy Studies Journal, (34) 4, 5 63–587, doi: 10.1111/j.15410072.2006.00191.x Teske, P., Schneider, M., and Cassese, E. (2005). L ocal school boards as authorizers of charter schools. In W. G. Howell (Ed.), Besieged: School boards and the future of education politics (1229-1279). Washington, D.C.: Brookings Institut ion Press. Stoddard, C. and Corcoran, S (2006). The Political Economy of School Choice: Support for Charter Schools across States and School Distri cts. New York: National Center for the Study of Privatization in Education. Stuit, D. and Doan, S. (2012). Beyond city limits: expanding public charter schools in rural America. National Alliance for Public Charter Schools Issue Brief, 1-16. Retrieved from at National Alliance for Public Char ter Schools at

PAGE 58

53 http://www.publiccharters.org/wp-content/uploads/20 14/01/NAPCS_BeyondCity-Limits_Issue-Brief_2_27_2012_20120223T101259.p df Teske, P., Schneider, M., Buckley, J., and Clark, S (2000). Does charter school competition improve traditional public schools? Center for Civic Innovation at the Manhattan Institute Civic Report, 10 (June 2000), 1-31. Retrieved from http://www.manhattan-institute.org/html/cr_10.htm Winkler, A.M., Scull, J. and Zeehandelaar, D. (2012 ). How Strong are U.S. Teacher Unions? A state-by-state comparison. (Thomas B. Fordham Institute). Retrieved from the Thomas B. Fordham Institute website at http://www.edexcellence.net/sites/default/files/pub lication/pdfs/20121029-UnionStrength-Full-Report_7_0.pdf Witte, J.F., Schlomer, P.A., and Shober, A.F. (2007 ) Going charter? A study of school district competition in Wisconsin. Peabody Journal of Education, 82 (2-3), 410439. doi:10.1080/01619560701313051. Wong, K.K. and Langevin, W.E. (2007). Policy expans ion of school choice in the American states. Peabody Journal of Education, 82 (2-3), 440-472. doi:10.1080/01619560701313085 Wong, K.K. and Shen, F.X. (2002). Politics of state -led reform in education: Market Competition and electoral dynamics. Educational Policy 16 (16)1. doi:10.1177/0895904802016001009 Zhang, Y. and Yang, K. (2008). What drives charter school diffusion at the local level: educational needs or political and institutional fo rces? The Policy Studies Journal 36 (4), 571-591. doi: 10.1111/j.1541-0072.2007.00284. x. Torres, Z. (2014, January 15). Student enrollment u p in Colorado; DPS now largest school district. The Denver Post Retrieved from http://www.denverpost.com/news/ci_24913752/studentenrollment-up-coloradodps-now-largest-school#ixzz30Epvqm