Citation
Child care and women's earnings

Material Information

Title:
Child care and women's earnings
Creator:
Harris, Pavla
Publication Date:
Language:
English
Physical Description:
175 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Child care services ( lcsh )
Wages -- Women ( lcsh )
Women -- Economic conditions ( lcsh )
Child care services ( fast )
Wages -- Women ( fast )
Women -- Economic conditions ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 168-175).
General Note:
Department of Sociology
Statement of Responsibility:
by Pavla Harris.

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:
51820719 ( OCLC )
ocm51820719
Classification:
LD1190.L66 2002m .H37 ( lcc )

Full Text
CHILD CARE AND WOMEN'S EARNINGS
by
Pavla Harris
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
Sociology
2002
|AL
L,__
1


This thesis for the Master of Arts
degree by
Pavla Harris
has been approved
by

W //.
Date
Virginia S. Fink


Harris, Pavla (M.A., Sociology)
Child Care and Women's Earnings
Thesis Directed by Associate Professor Richard H. Anderson
ABSTRACT
The goal of this thesis was to examine the relationship between child care and
women's earnings and to find out if there is an effect of child care on earnings
of working women with young children. The Integrated Theory of Gender
Stratification posits that child care plays an important role in reproductive
labor and, consequently, influences productive labor conditions by influencing
the compatibility of productive and reproductive labor. The author proposes
that different types of child care arrangements influence women's earnings
differently, because they influence compatibility of productive and
reproductive labor in different ways. Since women's earnings in contemporary
American society are the key to women's access to economic resources and
thus to economic power (and in extension, according to structural theorists, to
political power), child care becomes a very important factor in gender equality.
The author used the NLS Y79 data set to determine if different types of child
in


care arrangements influenced income (salary, wages and tips) of 1136 working
women with children aged five or younger. The results of multiple regression
statistical analysis support the working hypothesis that there is an association
between child care arrangements and women's earnings. Using a relative to
care for a child in the child's own home is associated with lower income, using
a commercial child care center is associated with higher income. Overall,
about four percent of the variation in income in the sample was explained by
the variation in the type of child care arrangement. Further research would be
necessary to determine the precise mechanism of the process, but the author
speculates that varied reliability of different child care arrangements might be
important in the compatibility of productive and reproductive labor, thus
influencing the women's income.
This abstract accurately represents the content of the candidate's thesis. I
recommend its publication.
IV


DEDICATION
I dedicate this thesis to Mike.


ACKNOWLEDGMENT
My thanks to Dr. Richard Anderson for his unwavering optimism and faith, to
Dr. Candan Duran-Aydintug for leading by example and believing in me, to
Dr. Virginia Fink for support in more ways than one, and last, but not least, to
Steve McClaskie in Ohio, who showed me that even large data sets can be
manageable.


CONTENTS
Figures...................................................x
Tables...................................................xi
CHAPTER
1. INTRODUCTION......................................... 1
Choosing the Topic: Women's Earnings and Child Care.1
Statement of the Problem to Be Studied..............4
2. THEORETICAL PERSPECTIVES...............................8
3. REVIEW OF THE LITERATURE..............................18
Women's Employment and Child Care..................18
Women's Earnings...................................19
Child Care Choice..................................21
Child Care Costs and Quality.......................24
Vll


4. METHODOLOGY................................................31
Study Design............................................31
Research Problem, Questions and Hypotheses..............33
Choice of Dataset.......................................34
Description of the Dataset..............................36
Dependent Variable: Income........................40
Independent Variable: Child Care Arrangements.:...41
5. RESULTS....................................................45
Descriptive Statistics..................................45
Statistical Analysis....................................48
ANOVA.............................................48
Cross Tabulation with ETA.........................49
Multiple Regression...............................51
Bivariate Correlations............................51
Stepwise Regression...............................52
Logistic Regression...............................56
Summary...........................................60
Vlll


6. DISCUSSION AND IMPLICATIONS
61
Discussion and Conclusions....................61
Implications of the Study and Recommendations.67
APPENDIX............................................70
A. LIST OF EXTRACTED VARIABLES....................70
B. CODEBOOK ENTRIES FOR EXTRACTED VARIABLES...73
C. DESCRIPTION OF THE CHILD CARE ARRANGEMENT
VARIABLES.....................................110
D. FREQUENCY TABLES..............................112
E. ANOVA.........................................123
F. CROSS TABULATION WITH ETA.....................127
G. CORRELATIONS..................................132
H. STEPWISE REGRESSION MODELS....................134
I. LOGISTIC REGRESSION MODELS....................156
BIBLIOGRAPHY........................................168
IX


FIGURES
Figure
2.1 Gender Organization of Production
15
x


TABLES
Table
5.1 Child Care Arrangement for Child 1 by Child Care Arrangement
for Child 2 (Reported in Percentages).........................47
5.2 Multiple Regression Models with "Spouse's Income" Variable
Included......................................................54
5.3 Multiple Regression Models with "Spouse's Income" Variable
Excluded......................................................54
5.4 Significant Predictors for Specific Types of Child Care
Arrangements (Spouse's Income Included).......................58
5.5 Significant Predictors for Specific Types of Child Care
Arrangements (Spouse's Income Excluded).......................59
xi


CHAPTER 1
INTRODUCTION
Choosing the Topic: Womens Earnings and Childcare
Since I started my sociological studies, I talked informally with many women about
gender, inequality, and the American society. One issue came up time and again
while discussing women's wages: child care. Dropping off my children at a day
care center, researching articles in the library, or walking the dog, when women
found out I was interested in researching gender issues, they always mentioned the
double workload they faced by being employed and taking care of their family,
and, invariably, how difficult it was to solve day care issues. Very often I heard that
they could not accept certain jobs or had to work limited hours or quit because of
child care problems. Matters of child care seemed to be one of the most important
aspects of women's labor force experience. Even women with executive-level jobs
and nannies in place were concerned when, for example, the nanny had a boyfriend
and was thinking about getting married.
1


These conversations brought me to thinking about how important child care is in
connection with access to resources in the form of paid labor, and, consequently,
gender inequality for women.
Most women in the United States are in the labor force, and that includes mothers
and caregivers of very young children. Bureau of Labor Statistics reported (1987)
that in 1986, half of all married mothers with infants under age one were in the
workforce and 60% of mothers whose youngest child was 3-5 years old were
employed, and the numbers are still rising (Brayfield 1992, Cohen and Bianchi
1999). In 1995, approximately 44% of children under five years old and 75% of
children five to fourteen years old regularly spent time in more than one child care
arrangement per week (Smith 2000).
About thirteen million children under age six and 31 million children between the
ages six and seventeen have both parents or their only parent in the work force
(Bureau of Labor Statistics 1998 and Child Care Action calculations 2002).
Between 1970 and 1996 the percentage of working married women with preschool
aged children rose from 30.3% to 62.7% (Barrow 1999).
2


United States Commission on Civil Rights (1981) established that lack of child care
or inadequate child care prevents women from taking paid jobs, keeps them in part
time jobs or jobs they are overqualified for, restricts them from participation in
federally supported education programs, or taking job promotions. Thus the
unavailability of adequate child care restricts equal opportunity for women in many
ways. The Civil Rights Commission's report is over twenty years old, but, in my
opinion, the issues it addresses are still of valid concern, as evidenced, for example,
in the cross-national study of public policies (Gomick, Meyers, and Ross 1998).
Gomick, Meyers, and Ross (1998) compared public policies and employment of
mothers in 14 industrialized countries. The data came from the Luxembourg
Income Study (LIS). All countries that included needed variables (relevant public
policies and mothers' employment) were selected. The independent variable was
national policy performance two composite indexes of policy indicators. The
dependent variable was each country's "child penalty": a regression-adjusted
estimate of the decrease in mothers' employment probability given the presence of
young children at home. The countries were ranked separately twice: first assuming
child three and under for a Base mother, then a child over the age of three. The
findings were consistent with the central hypothesis that policies influence maternal
3


employment behavior. There was a strong association between cross-national
variation in employment supporting policies and the continuity of maternal
employment. (Labor supply theory assumes that at least short-term labor supply
decisions are sensitive to policy.)
In the United States, Australia, and the United Kingdom, the government policies
were the least developed and very large reductions in employment were observed,
especially among mothers with children under age three. Belgium, Denmark,
Finland, France, and Sweden provided the most generous benefits (Gomick,
Meyers, and Ross 1998).
Statement of the Problem to Be Studied
While trying to learn more about child care situation in the current American
society I discovered that, as Corsaro (1997) points out, the debates about child care
seldom distinguish between different types of child care arrangements, such as care
by relatives v. non-relatives, center or nanny care, in the child's home or other
setting etc.
4


Some authors indicate that unavailability or affordability of child care and/ or
problems with existing child care prevent many women from holding jobs or force
them to accept part-time jobs or jobs for which they are overqualified. It is then
logical that these constraints would influence the amount of money women are able
to make. Why would women be unable to get a promotion or a better job as a result
of child care characteristics? In my opinion, the reasons are related not only to the
costs of child care as many authors suggest, but also to the type of care. For
example, it might be possible that care with flexible hours of operation or nanny
care will enable mothers to work better paid shifts or longer hours such as those
required by executive jobs. My goal, therefore, was to investigate if there was an
association between the type of child care women use and the amount of money
working women with children make.
I propose that different types of child care will influence women's earnings, that
there will be a difference in women's earnings in relation to the type of child care
they use. I assume that the relationship is bi-directional, i.e., that women with
higher earnings will be able to afford more expensive child care (nannies, child care
centers), but also that the child care utilized will influence the ability to make
higher or lower income.
5


Here also arises a question if women find employment first and then look for child
care or if they arrange for child care first in order to look for work or if their
decisions are virtually simultaneous. Studies on the topic are inconclusive. For
example, Cattan (1991) states that the lack of adequate child care kept over a
million young mothers from working or seeking a job in 1986. This finding seems
to imply that women are looking for an acceptable child care situation before
finding employment or, at least, at the same time. Michalopoulos, Robbins and
Garfinkel (1992) present a structural model in which the decision to purchase
market child care is made simultaneously with the employment decision of the
mother. In my opinion, all scenarios are plausible and perhaps they might happen at
different times in women's lives. The greatest dilemma for women occurs when
they are in their prime childbearing years, because at the same time they are likely
to be developing their career or making important career decisions. For women
who are trying to reenter the work force after giving birth or staying at home with
young children, the child care decision might come before or at the same time as
the employment decision. After all, someone has to watch the children while the
woman is looking for work. Later in life, when a woman's job is secure, other
factors come into play and different behavior might be observed.
6


I want to examine the matters of child care as related to women's earnings, because
greater understanding of these issues can lead to better understanding of the system
of gender inequality in our society. Perhaps the deeper knowledge will then
contribute to solutions that will ultimately decrease the level of inequality between
women and men.
7


CHAPTER 2
THEORETICAL PERSPECTIVES
After an exhaustive research of literature, mainly in the areas of sociology and
economics I determined that social scientists generally employ two different
approaches in explaining women's wages and wage inequality between men and
women: individualist and structural (Peterson, Dimitrova, and Lambe 1993).
The individualist approach focuses on the equality of opportunity for women and
how women use or do not use the opportunity. Many economists favor this
theoretical framework as expressed in the human capital theory (Becker 1991,
1993). Human capital, according to Gary Becker, is characterized as education,
training, experience, and other investments (for example, medical care) that enable
people to achieve higher earnings than without these investments (Becker 1993). In
my understanding, "human capital" would mean simply the aggregate of all these
factors, but Becker specifically uses the word investments.
When investigating why mothers earn less than other women do, human capital has
been suggested as one of the reasons. However, even after controlling for human
8


capital, there is a wage gap present, as Budig and England suggest (Budig and
England 2001).
Budig and England (2001) strove to explain why mothers earn less than other
women. The authors analyzed pooled data of the 1982-1993 waves of the NLSY
study. They limited their sample to women employed part-time or full-time during
at least two of the years from 1982 to 1993 and used fixed-effects regression
modeling for their analysis. Budig and England found an approximately seven
percent per child wage penalty among young American women.
Only about one-third of the penalty was explained by years of past job experience
and seniority, including whether past work was part-time. "Mother-friendly" job
characteristics explained little of the penalty, except of the tendency of more
mothers than non-mothers to work part-time. The resulting five percent per child
difference was left unexplained. Budig and England suggest it could be a
consequence of (as yet not fully understood or documented) the effects of
motherhood on productivity and/or of discrimination by employers against
mothers. I would like to introduce the idea that part of the difference might be
attributed to child care factors.
9


The structural approach emphasizes structures that are present in society and create
existing situation. The works of Joan Huber or Janet Saltzman Chafetz belong in
this category and I will analyze their theoretical thoughts in greater detail.
Janet Saltzman Chafetz had perhaps the greatest influence on my theoretical
thinking. Chafetz's approach is mainly structural. I consider her best theoretical
work to be Gender Equity (1990). The purpose of Gender Equity is to discover how
gender systems change or could be changed and to introduce a theory of stability
and change of gender systems. To achieve this goal Chafetz first investigates the
stability of gender systems and what influences it.
Chafetz (1990) starts Gender Equity with a brief overview. To summarize her
chapter in my own words:
1. There is a system of gender stratification.
2. Gender division of labor is responsible for reproduction of gender
inequality in society.
3. Women's opportunities are dependent on their access to resources
(or resources generating work).
4. The access to resources is controlled by male elites.
10


5. The access to resources changes by forces that are not under
women's control.
6. Things have changed but many changes are "more apparent than
real."
In Part I., Gender Systems Stability Chafetz posits that there exist change targets -
variables that have a certain characteristic: if they change, other variables are likely
to change also. The same variables determine if the system will stay the same -
sustain the status quo.
In Chafetz's view, the critical change target of gender stratification is the gender
division of labor. Gender division of labor consists of three elements (Chafetz
1990): household and familial labor (taking care of children would belong into this
category), extradomestic labor, and representation among incumbents of elite roles.
Since household and familial labor (in other words, reproductive labor) is one of e
the factors of the gender division of labor, and gender division of labor is, in turn, a
critical change target of the system of gender inequality, it becomes obvious that
child care does play a role in the system of gender inequality. Consequently,
changes in child care should have an influence on the gender division of labor and
gender inequality.
11


Chafetz (1990) also seeks to reconcile two different theoretical approaches to
gender inequality: those that put an emphasis on coercive aspects of gender systems
and those that stress the voluntaristic ones. In this aspect, her approach is similar to
Kathleen Gerson's (1985).
Gerson provides very good theoretical overview of theories of women's behavior.
She divides the theories into the structural coercion approach, the voluntarist
approach, and the developmental approach. Her interpretation of the structural
coercion approach encompasses structural and conflict theories (Marxist -
feminist). Gerson explains female subordination by the economic system of
capitalism, but claims that this line of thinking underestimates women's active role
in their own lives. The voluntarist approach deals with various childhood
socialization theories, but I agree with her critique that these theories do not
provide for explaining variations among women or accounting for the dynamics of
(individual and/or social) change. The developmental approach is based on
constructing life history analyses, cohort analyses and similar.
I agree with Chafetz that women's opportunities are dependent on their access to
resources. This notion is also very well explained in Joan Huber's work (Huber
12


1983, 1988,1990.) Huber's theory of gender stratification the Intersection of
Family, Economy, and Gender, identifies three propositions (or principles, as she
sometimes calls them): microlevel, macrolevel, and general stratification
proposition. The three propositions explain how ecological conditions and tool use
interacted with childbearing and suckling to shape gender stratification in different
societies. Huber clearly suggests that resources available to people (groups,
genders) determine the power they have in a society. The macrolevel perspective is
the one most relevant to my topic. It can be implied that without adequate child
care women's access to work-resources, and therefore distribution of goods and
power, is limited, which increases men's power in society.
Collins, Chafetz, Blumberg, Coltrane, and Turner collaborated on an Integrated
Theory of Gender Stratification, published in 1993 in Sociological Perspectives. In
different parts of the theory it is obvious which of the authors provided the basis
theoretical framework, but it is very well thought out. The chapters I was most
interested in were heavily influenced by Chafetz, although I prefer Chafetz' writing
style and deeper explanations in her original works.
13


The Integrated Theory starts with formulating four blocks of fundamental causal
conditions for constructing a complex theory. These are:
1. Theories of the gender organization of production theories that put
emphasis on economic positions of men and women.
2. Theories of the gender organization of reproduction. These theories focus
on childbirth and parenting and their effects on gender psychodynamics and
culture or women's economic activities.
3. Theories of sexual politics, investigating erotic relationships and their
connection to social power.
4. The structure of political economy, which comprises of background
conditions for the three types of gender theories. I appreciate that the theory
contains references to economy, especially political economy. Quite often
we see social science disciplines focused on very narrow subjects where a
broader picture would be more appropriate.
14


For my research, the first two blocks of theories will be the most important. The
Figure 2.1 shows different factors influencing them.
Figure 2.1: Gender Organization of Production (Collins, Chafetz, Blumberg,
Coltrane, and Turner 1993).
Compatibility of productive and reproductive labor is major factor in gender labor
segregation. Pregnant, breastfeeding, or caring-for-children females adjust,
according to the Integrated Theory, their activities (both productive and
reproductive) to make them mutually compatible. Child care labor is most
15


compatible with housework. This makes women in many cases spatially separated
from paid labor opportunities.
There are several conditions that influence productive/reproductive compatibility,
for example demographic and technological aspects of reproduction. Demographic
factors would include birth rate and the timing of births, technological ones bottle-
feeding of infants. Social conditions affect the distribution of reproductive labor -
essentially who takes care of children. Care could be collectivized, there could be
wet nurses employed etc. Cultural factors are also part of the Reproductive Labor
block.
Three change factors in the modem times influenced how much time men and
women spend in reproductive labor (a key factor in the demographic conditions).
These three factors are:
1. More favorable health and medical conditions reduced the death
rate.
2. Birth control technology reduced the birth rate.
3. Sterilized baby bottle made it easier for women to participate in the
labor force. This factor is especially important, because a mother
16


does not have to be present to feed a baby and could be at a different
location generating money resources.
Social conditions raise or diminish the importance of these factors.
From the brief overview of the relevant part of the Integrated Theory of Gender
Stratification it is obvious that child care plays an important role in reproductive
labor and that it influences productive labor conditions. Child care provided by
someone other than a child's mother frees the mother for labor force participation
and, consequently, enables her to access economic resources.
My thesis topic fits in the Compatibility of Productive/Reproductive Labor block. I
proposed that different types of child care arrangements would influence women's
earnings differently, because they would increase or decrease the compatibility of
productive and reproductive labor in different ways.
17


CHAPTER 3
REVIEW OF THE LITERATURE
Womens Employment and Child Care
In the study in which Waldfogel (1998) ascribes part of the family gap to maternity
leave, she writes:
"Child care is likely to be especially critical as a remedy for the family gap
because of the effect it could have in allowing women to maintain work
experience and job tenure, as well as continuity of employment over
childbirth. There is a great deal of evidence that child care costs and supply
constrain women's labor supply... The potential effects of measures to shift
the costs of child care from mothers to society and to expand the availability
of child care particularly care for infants, merit further research."
Scientists seem to support the notion that child care influences the access to
economic resources for women as evidenced for example in the studies of Peter
Cattan (1991) or Gordon and Chase-Landsdale (2001). Other authors conclude that
child care constraints and/or unavailability prevent large numbers of mothers from
holding a job at all (Kimmel 1998, Cattan 1991, Brayfield 1992). Press (1998) in
her dissertation states that limitations on labor force activity as a result of child care
concerns are important factors in explaining poverty. Press explores how "multiple
jeopardy" influences paid work, families, poverty, and social inequality in multi-
18


racial Los Angeles. By multiple jeopardy Press means gender, race, and class
inequality for women of color. Her research focuses mainly on issues of poverty
and she considers child care one of the most important factors that can limit
women's labor force participation. Using the data form the Los Angeles Study of
Urban Inequality and logistic regression Press determined that child care problems
may increase the risk of poverty on average by 25%.
Womens Earnings
How much money women with children make is a dependent variable in my
research. Jane Waldfogel (1997) came to a conclusion, that there is a wage gap
between mothers and other women, even after controlling for human capital.
Waldfogel calls this difference a "family gap." In her study of 2133 employed
women between the ages of 34 and 44 Waldfogel created a pooled data set with a
sample size of over 30 000 woman-year observations. She used the National
Longitudinal Survey of Youth Women. Part of the family gap Waldfogel
discovered was attributed to working part-time, but even after controlling for part
time status, human capital, and unobserved heterogeneity, there still was a four
percent penalty for having one child and almost twelve percent for having two or
more children. Waldfogel speculates that "work and family conflict" may
19


negatively influence wages. The term "work and family conflict" corresponds, in
my view, with the concept of compatibility of productive and reproductive labor as
presented in the Theoretical Perspectives chapter of my thesis.
Waldfogel's study, however, did not deal with the possible influence of child care
on earnings. Is it possible that part of the family gap could be attributed to child
care arrangements?
Budig and England (2001) proposed to explain the residual wage gap by a potential
employer discrimination, statistical discrimination, or a decreased productivity
(and/or effort) of mothers, that are tired from taking care of their children, as
described in greater detail in the preceding chapter.
Waldfogel (1998) suggests that the length, organization, and structure of the
maternity leave can explain about 40% the family gap. The author compared
female-to-male earnings ratios, gender and family gaps with maternity or parental
leaves in selected industrialized countries (depending on available data, she used a
list of 7-15 countries.) Waldfogel concludes that maternity leave coverage had a
positive effect on women's wages and explains this effect by the fact, that maternity
20


leave enables women to the same work arrangement so they not only keep their job,
but also retain experience and seniority. Waldfogel also points to the importance of
child care as a remedy for the family gap.
Child Care Choice
Some of the most important attributes of child care are costs (price), quality, type,
and availability of care. Availability would include what I would call for a lack of
official definitions space and time factors. Space factors encompass number of slots
open for children and perhaps availability of babysitters, within a certain distance
from their place of residence or parents' work. Time factors are characterized as
days and hours of operation of any available child care arrangements. Satisfactory
child care situation will be the result of intersection of the above mentioned factors
and parents' preferences, which may be based on cultural factors and child rearing
philosophy.
Hertz and Ferguson (1996) conducted a qualitative study of 95 dual-earner couples
living in Eastern Massachusetts on child care choice and argued that child care
decisions are influenced more by race, social class and beliefs about mothering than
by cost. They, however, studied mostly middle- and upper middle-class families
21


and even the families the researchers considered working class had income between
forty thousand and one hundred thousand a year. The authors emphasized that
different arrangements suited different families with some families preferring care
by a relative (this included the mother or the father). Black families were reporting
attempts to find multiracial settings. Both black and white parents put importance
on appropriate developmental and emotional surroundings. This finding supports
Waites (1991) conclusion that not only structure, but also process and outcome of
care are important when parents are making decisions about their childrens care.
Hertz (1997) reports that families employ basically three approaches to child care:
the mothering approach, the parenting approach, and the market approach. Hertz
interviewed 95 dual-earner couples about their choices of day care arrangements.
Mothering meant that after a child's birth a wife's status as a mother becomes the
central focus of the family and couples adjust their job activities to maintain their
standard of living (husband increases his working hours or wife adjusts her hours
and job to accommodate the family). For the parenting approach couples family
comes first; both husband and wife try to restructure their work and make sacrifices
in order to be active parents. In these couples men participate in traditionally
female parenting tasks and share them. The market approach in Hertz'
22


interpretation means that couples value and attempt to purchase professional child
care services that they prefer over "old-fashioned folk wisdom." Their marriages
are more equal, but they do not change the organization of gender roles. The
parenting approach appears to be the most recent development in child care
attitudes, according to the author.
Riley and Glass (2002) studied child care preferences among working mothers of
infants. They interviewed the mothers during pregnancy and six months after their
children were bom. The types of care Riley and Glass considered were only two:
care by relatives and care by others. Maybe for the study of care preferences this
operationalization is appropriate, but in the scope of my thesis it would be too
simplified. The data for this study came from a regional Midwest sample of 324
pregnant women working at least 20 hours a week. Interviews were conducted with
a subsample of 247 women that had returned to work six months after giving birth
(76% of the original sample). Using logistic regression Riley and Glass (2002)
concluded that the most powerful determinants in achieving the match between
preferred care and care actually used were employment schedules (fewer hours
worked, evening or night shifts). The authors explain that this finding means that
some parents changed their working hours to get the type of child care they wanted.
23


Interestingly, higher family income did not help the parents get their preferred
mode of child care.
Henly and Lyons (2000) conducted an exploratory study of 57 urban, low income
mothers of children under the age of 13 working in entry-level jobs. Using a semi
structured interview the authors determined that low-income employed parents
disproportionately used informal child care arrangements, mainly because of their
affordability and convenience. In these aspects informal arrangements were
considered advantageous, but they were often inconsistent and unreliable and might
negatively influence job performance and stability.
Child Care Costs and Quality
Child care costs are one of the largest expenses for American households with
working parents. They rank after housing, food, and taxes and the rank does not
change with different income levels (Brayfield and Hofferth 1995). Brayfield and
Hofferth concluded that both cultural and economic factors play role in the
spending on child care of employed women with young children. They utilized the
data from the 1990 Child Care Survey. Mothers earnings and not familial financial
resources determined whether a family pays for child care. The actual cost of pay
24


was related to both earnings and income and higher-income families paid more for
care. The authors, however, assumed that quality of care rises with price, so then-
statement that it is the cost of care, mother's wage, and family income that will
determine what type and quality of care is purchased loses its potency (concerning
the quality of care). The research on quality of child care related to its price is
inconclusive. Brayfield and Hofferth recommend further investigation into the
cultural factors influencing child care.
Hofferth and Wissoker (1992) investigated the choices in child care as related to
price and type of care. They suggest that price is a critical variable in child care
choice (Hofferth and Wissoker 1992). Using the 1985 wave of the NLSY data set,
the authors also determined that mothers and families with higher incomes are
more likely to choose center care over other types of care. Four categories of child
care were considered: care by the father, care by another relative, care by a sitter in
the child's or the sitter's home, and care in a day-care center or a nursery school.
Apparently the categories were collapsed form the original NLSY variables,
because the NLSY recognizes more than four types of care as I will discuss later. I
would not recommend to combine the care by a sitter in the sitter's home and in the
25


child's home into one category, because there is a considerable difference in costs
and/ or possibly quality of the care between a child care provided by a nanny for
one child (for example), and a home that cares for many children at the same time.
Michalopoulos, Robbins and Garfinkel (1992) present a structural model in which
the decision to purchase market child care is made simultaneously with the
employment decision. The purpose of authors' study was to determine what would
happen (in terms of costs and effects) if federal child tax credit changed and more
generous subsidies were available to mothers, mainly in the form of refundable
credits. Michalopoulos et al. used the data from the 1984 Survey of Income and
Program Participation (SIPP). Using the utility maximization model the authors
determined that labor supply of women would increase with the introduction of
refundable credits, but not as much as the increase in costs for child care. Overall
quality of child care would not change very much. The paper concentrates mostly
on the economics of child care.
Hofferth and Wissoker later revised their paper (Chaplin, Hofferth, and Wissoker
1996), but they kept the original assumptions. The revision involved lowering the
original negative price effects on choice of child care. In their first study the
26


authors also stated that parents do not consistently select high quality care. Quality
of care was measured as the child/staff ratio. Quality of care is frequently discussed
as being difficult to measure, and there is not a common agreed-upon
operationalization in the literature. The availability of a relative living within 30
minutes or in the same household increased the likelihood of using a relative for
care. Hofferth and Wissoker did not, however, consider any measure of the
availability of other modes of care. This omission is understandable given the
limitations of the data set. Availability of other types of care would have to be
imported from a different data set.
Barrow (1999) analyzed women's decisions to return to work after the birth of their
children and concluded that the cost of child care plays major role in that decision.
The author cites "economic theory" in explaining that women with lower child care
costs are more likely to return to work. The economic theory model that Barrow
uses is based on an assumption that women are trying to maximize their utility by
getting back to labor force and the cost of child care acts as a "tax" on her earnings.
Barrow does acknowledge possibilities of other (unmeasured) influences, such as a
"taste for work." She worked with data from different data sets: NLSY, SIPP, and
Current Population Survey. It is possible that women with lower family income do
27


not have the luxury to choose if they return to work force or not. Consequently,
because they earn less money than other women do, they can only afford to pay a
certain amount for day care, regardless of their preferences for quality, convenience
etc.
Perhaps the most complete consideration of quality of child care comes in the
research of Linda Waite, Arleen Leibowitz, and Christina Witsberger (1991). The
authors explored the child care selection process, how high quality of care is
defined and its relation to costs. Waite et al. (1991) first overviewed what
constitutes high quality care. Low ratio of children to adults, small groups of
children and early childhood training of teachers are all considered to be positive
influences on children's well-being and development, even though the actual
numbers depend on the age of the child. Waite, Leibowitz, and Witsberger created
an index of child care quality:
1. Care had to meet the Federal Interagency Day Care Requirements
for group size for the child's age, that were in effect at that time.
2. Requirements for number of adult care givers per child for children
of that age also had to be met.
28


3. The provider had to have early childhood training.
Waite et al. (1991) used data from the NLSY 1985 wave. There are six types of
child care distinguished:
1. In the child's own home by parents.
2. In the child's home by others.
3. In the home of a relative.
4. In the home of a non-relative.
5. Nursery school.
6. Organized child care center.
The researchers write that the study's results were somewhat surprising to them:
Parents were not paying more for high-quality care as measured by their index.
Families with more resources also did not obtain higher quality care. Women with
higher earnings or education also did not get higher quality of care for their
children. The results did not vary for younger and older preschoolers.
29


This unexpected finding is explained theoretically from the literature on quality of
healthcare that distinguishes between structure, process, and outcome of care. The
licensing requirements commonly used to measure quality of child care are
structural, but parents may choose care on other bases process or outcome of care.
Process characteristics would include secure and loving environment, and outcome
would indicate emotional and intellectual development of children. I suppose that
these characteristics are difficult to routinely quantify for every child care situation.
30


CHAPTER 4
METHODOLOGY
Study Design
In line with Hubers (1990) reasoning, to gain greater equality in society, women
need to have more resources. In current American society probably the most
common resources-generating activity accessible to women is paid employment.
The ability to engage in paid employment depends on many factors, many of which
in turn depend on gender division of labor, e.g., organization and cost of child
bearing and child rearing, and domestic labor.
Related to these factors, child care is, in my opinion, the most important, because it
is the only factor aside from pregnancy and childbirth requiring direct involvement
of human beings that is not as yet replaceable by technology. Thus child care will
directly influence women's access to resources.
I believe that the relationship between child care and women's earnings is bi-
directional and that the importance of child care as suggested in the Integrated
Theory of Gender Stratification (Collins, Chafetz, Blumberg, Coltrane, and Turner
31


1993) stretches beyond the domestic division of labor. Consequently, women's
earnings are the key to women's access to (economic) resources, provided, of
course, women are allowed to keep them.
Women's earnings depend in the first place on demand for their labor, in the second
place on supply-side variables that enable women to work. One of these variables
for mothers is child care. A woman cannot usually work outside her home without
child care and she cannot secure child care without having the resources
(sometimes considerable) to do so, usually provided by the results of her paid
employment. Child care constraints then possibly influence work factors, e. g., the
type of work, hours worked, and, consequently, the earnings.
If there are differences in earnings among women with children using childcare,
there will likely be even bigger differences between earnings of mothers and
childless women. This proposition fits well into the integrated theory of gender
stratification, as previously discussed.
Based on what the literature review indicates, I decided to examine if there is any
effect of child care arrangements on earnings of working women with children.
32


Research Problem. Questions and Hypotheses
The research problem is to investigate the presumed effect of child care
arrangements on income in the population of working women with children under
the age of 5. Can different types of child care arrangements be used to predict
income?
Null Hypothesis: There is no association between womens income and type of
child care arrangement used.
Alternate (working) hypothesis: There is an association between womens income
and type of child care arrangement used.
Dependent variable: Income, operationalized as dollar amount of income from
wages, salary, and tips in a calendar year (1988).
Independent variable: Types of child care arrangements for one child.
(Operationalization will be described in detail later.)
33


From the literature review I knew the relationship between child care and income
was likely to be reciprocal, that is, income can influence the nature of child care as
well as child care affecting the income. To research both these possibilities I
examined the child care effect on earnings first and then turned to the effect of
earnings (and possibly other variables) on child care arrangements.
Choice of Dataset
One of the most useful sources I found for evaluating data sets dealing with child
care was Raley, Harris, and Rindfuss' article "The Quality and Comparability of
Child Care Data in U. S. Surveys", published in 2000 in Social Science Research.
The article reviews the National Longitudinal Survey of Youth (1986 and 1988),
the Survey of Income and Program Participation (1986,1987, 1988, and 1990), the
National Child Care Survey, and the National Survey of Families and Households.
All of these are large national studies, even though they were not designed for
identical purposes.
The authors of the study concluded that four main sources of differences in child
care data were:
34


1. Season of the year when the interview is conducted.
2. Screening questions determining who is asked about child care.
3. Survey design the population represented.
4. The way the questions are asked.
The questions asked in different surveys differ, but I was surprised to learn that
even the season of the year when the interviews were conducted had a bearing on
the answers, especially when asking about a short period of time, as some questions
do.
The authors report that the level of disagreement across the studies in primary care
type is troublesome and they discuss some of the ways in which they differ. They
caution that the restricted cohorts of the NLSY can produce different descriptive
results compared to SIPP (Survey of Income and Program Participation), but they
do recommend using the NLSY for the investigation of the determinants of child
care use and type for children of employed mothers. I decided to use the NLSY
dataset for several reasons:
35


1. NLS Y is a long running longitudinal study, so there is an option to
repeat the analysis for the same respondents in different years, or to
design a more complicated study at a later date.
2. NLSY is a nationally representative sample.
3. NLSY has links between the mothers work and child care.
4. NLSY has an excellent technical support and a vast online
bibliography and other resources. I consider this a great advantage
for any kind of project.
Description of the Dataset
The NLSY79 is a nationally representative sample of 12,686 young men and
women who were 14-22 years old when they were first surveyed in 1979. Data
were collected yearly from 1979 to 1994, and biennially from 1996 to the present.
The latest year of data available for analysis is 2000.
36


The main data collection on types of child care arrangements occurred in the years
1982-86, 1988, 1992, and 1994-2000. A limited number of child care questions
were also contained in the fertility series in 1987, 1989, and 1990. A special Child
Care Supplement was administered to 347 mothers in 1989, but it has not been
repeated since. I decided against using this supplement for analysis for two reasons:
1. Sample size is quite small and if there were any missing data, it would get
even smaller.
2. One-time nature of the supplement would make it impossible to repeat the
analysis with data from different years and/or longitudinally. I wanted to
leave this option open for possible use later.
The NLS Y79 User's Guide notes that there are differences in universes of
respondents, reference children, kinds of questions asked, and reference periods
across survey years. In the initial survey years the focus of the questions was on
collecting information on child care arrangements used over the past month for
only the youngest children in the household. In 1986 andl988 the questions were
37


extended to all children in the household. In some years (1983-1986 and 1988)
information was gathered on both primary and secondary child care arrangements.
The 1982-2000 child care questions are contained within "Child Care" area of
interest in the NLSY79 data set.
The NLSY79 has three subsamples:
1. A cross-sectional sample of 6,111 respondents designed to be
representative of the non-institutionalized civilian segment of young
people living in the United States in 1979 and bom between January
1,1957, and December 31,1964.
2. A supplemental sample of 5,295 respondents designed to
oversample civilian Hispanic, black, and economically
disadvantaged non-black/non-Hispanic youth living in the United
States during 1979 and bom between January 1,1957, and
December 31, 1964.
38


3. A sample of 1,280 respondents designed to represent the population
bom between January 1, 1957 and December 31,1961, and who
were enlisted in one of the four branches of the military as of
September 30, 1978.
Respondent's sample type can be identified by using variable R01736.
I decided to use one woman (non-institutionalized), with at least one child of the
age of five or under, who worked in the four weeks preceding the date of the
interview, as a unit of analysis. The universe for the analysis was: all women with
above mentioned characteristics interviewed in 1988. The year was chosen because
of the ages of the women and the children: women, who were agesl4-22 in 1979,
would be 22-31 in 1988, in their prime childbearing age and therefore very likely to
have preschool children in the household who would need childcare if the mother
was working.
I also chose the ages of the children based on my review of the literature: I wanted
to rule out school-age children, because in many cases the school is the primary
care provider and also, higher percentage of older children provides self-care
39


(Raley, Harris, and Rindfuss 2000). I assumed that in those situations, child care
would lose some of its (if it, in fact, has any) effects with regard to womens
earnings, because it would not act as a limiting factor as much as it does when
children are pre-school age.
I selected the cases for analysis by identifying 44 variables of interest in the
original NLSY79 dataset and extracted them. The NLSY79 is structured so that
every variable in every year has a unique number, i.e., one variable does not have
the same number in different years. The list of extracted variables is in Appendix
A. The codebook printout for the variables is in Appendix B. For future reference it
is worth noting that each extracted subset had to include variable R00001.00 that
assigns an identification number to respondents. I selected all the cases in which a
woman had one or more children ages five or under and worked in the past four
weeks. Final number of cases selected was 1185.
Dependent Variable: Income
After running simple correlations it was obvious that there was one outlier case
with an extremely high income in my sample. (Income in the NLS Y is
operationalized as income from wages, salary, and tips in a year.) Moreover, it was
40


not possible to determine what the actual income was, because all cases with
income of over one hundred thousand dollars a year were truncated. Selecting the
case out also removed all the missing values of income. There was also one outlier
in spouses income, which I identified and deleted. There were 1136 cases left.
Independent Variable: Child Care Arrangements
In the original dataset there are two basic variables pertaining to arrangements for
care of first and second child. One of them describes the person who takes care of
the child, the other one the place where this care is administered. Description of the
variables is in Appendix C.
My plan was to develop a variable that would encompass information from both
variables.
I did not consider the issue of cost at this time, because the same arrangement could
have different costs for different women in different parts of the country. Who
takes care of children is probably the most important factor in the childrens
development, but where the care takes place is also important.
41


Originally, I considered care in child's own home more flexible, because if a person
(relative or non-relative) is available to come to the mothers home, he or she will
probably be more willing to adjust his or her schedule to the mothers
requirements, while group care centers (in private homes or commercial centers or
schools) are very strict in this regard. On the other hand, some mothers might be
using care by relatives because they have no other choice and cannot afford their
preferred mode of care (Riley and Glass 2002).
Other authors (Henly and Lyons 2000) report that informal care (care by relatives)
is less reliable than commercial day care center care. In light of these conflicting
findings I decided that I could not rank different types of child care arrangements,
but could distinguish among in-home/out-of-home care, relatives and commercial
care arrangements.
The 17 categories in the first child care variable could easily be combined into
broad general categories of
1. Care by relative.
2. Care by non-relative.
42


3. Commercial day care etc.
The place of care variable is related only to the first 8 codes of the child care type
variable. I recoded it into care in childs home and care in other place and then
combined the two child care arrangement variables.
Resulting variable identifies the nature of the child care arrangements in the last
four weeks for the first and second child (Most women in the sample did not have
more than two children aged five and under.):
1 Relative in childs home
2 Relative outside childs home
3 Non-relative in childs home
4 Non-relative outside childs home
5 Commercial day care etc.
I believe this classification better describes the underlying logic of my argument
than the original two separate variables because it gives me the means to compare
43


the effects of different arrangements on income and includes the place where the
care is provided.
Even though 1988 was one of the few years when information was collected also
on secondary child care arrangements, I wanted to focus on primary (or principal)
care. Primary child care is the care that enables mothers to work. From the
frequency tables in Appendix D it is obvious that majority of women in the sample
used only one child care arrangement per child.
44


CHAPTER 5
RESULTS
Descriptive Statistics
To get the descriptive statistics I ran the Frequencies function in the SPSS.
Appendix D contains the resulting tables. The sample is described here by looking
at a number of demographic characteristics of the selected respondents. Median age
of respondents is 27,17.3 % of them are Hispanic, 26 % Black and 56.7 % non-
Black, non-Hispanic. About three quarters (78.1 %) received high school diploma
or equivalent, median highest grade completed is 12. Median size of a respondents
family is 4, median number of all children in the household, including step- and
adopted children is 2. Respondents median income is reported as $10000 for 1988,
hours worked 1820. Net family income (median) is $27700 and the median of
spouses income is $20000.
Examination of marital status shows that 71.4 % in our sample are married, 13.9 %
never married, and 14.7 % were formerly married. Most of the respondents live in
urban areas (76.3 %). Over 83 % of women have one or two children living in the
household. Median age of the first (NLS Y79 assumes the first child is the oldest
45


one) child is three; of the second child it is one year old. The children spend 39.5
hours in child care/week (median), parents (respondent, spouse or partner) pay
$45/week (median) for child care, if they pay for child care at all (in 39.7 % they do
not.) In over 80 % of the cases the principal (primary) child care used is the only
child care arrangement used. This number is higher for the second than the first
children (85.3 v. 81.3).
Table 5.1 compares the type of child care arrangement for child 1 to child care
arrangement for child 2. It is apparent that using a particular child care arrangement
for the first child makes it more likely to use the same kind of arrangement for the
second child. Highest percentages of use for both children occur for "relative in-
home" and "commercial day care" arrangements, around 95%.
46


Table 5.1
Child care
arrangement
for child 1
Relative
In-home
Relative
outside
home
Non-relative
In-home
Non-relative
outside home
Commercial
day care etc.
Total Count
Child Care Arrangement for Child 1 by Child Care Arrangement for Child 2
(Reported in Percentages)
Child care arrangement for child 2
Relative in-home Relative outside home Non-relative in-home Non-relative outside home Commercial day care etc. Total
94.1% .9% 3.1% 28.8%
88.9% 3.1% 3.5% 21.8%
.7% 83.3% 5.1%
1.7% 71.9% .9% 14.1%
5.3% 8.5% 16.7% 21.9% 95.7% 30.2%
152 117 30 96 115 510


Statistical Analysis
Multiple regression was the main statistical method used in the analysis, but I also
used ANOVA and ETA cross tabulation.
ANOVA
ANOVA was used to get a view of differences in income between groups of
women using different child care arrangements. Women using commercial day care
had the highest mean income ($13275/year), women using relative in the child's
home had the lowest one ($8937/year). The ANOVA tables are in Appendix E.
For child one: F (4,989) = 10.176, p = .000. There is 0% probability that one would
find the mean income for the five child care categories would differ as much as
observed just by chance. I concluded that the type of child care arrangement has an
effect on income (at least in a statistical sense).
The Levene's test of homogeneity of variances is significant, the variances are
significantly different and the assumption of equal variances is violated. However,
because the overall F is significant, it is possible to still use the ANOVA and to run
a Post Hoc test. The Post Hoc test helps to point out which categories of child care
48


arrangements make the most differences in income, since, as the Levene's statistics
revealed, the categories do not make an equal difference. I used the Dunnett C
which indicates that the connection between child care arrangements and income is
concentrated in the relation of lower income and "relative in home" category,
"commercial day care" and higher income than other categories. Results for child
two data are similar.
Cross Tabulation with ETA
Cross tabulation with ETA is useful in cases where one variable is interval and one
nominal data (for the associational approach) and one wants to consider only two
variables. I computed this statistics twice, once for income as a continuous variable
and once for income that was recoded into categories by $5000. The second action
resulted in a table of more manageable length than the first one, the results were
similar. The tables for child one and child two with recoded income are in
Appendix F.
When looking at the table with child care arrangements for child one I observed
that the discrepancies between the counts and expected counts in some categories


of child care arrangements were higher than in others, which indicates higher
likelihood that there is an association between income and child care arrangements.
The categories with highest discrepancies were "relative in home" and "commercial
child care". In "relative in home" category the expected counts were lower; in
"commercial child care" they were higher than the actual counts. I assumed that this
meant that one of the categories would be positively and the other one negatively
associated with income.
The value of the ETA was 0.191, ETA square 0.036, which is consistent with ETA
obtained from ANOVA. In the cross tabulation with ETA in which the income was
left continuous, the value of ETA was 0.199, ETA square 0.039, same as the ETA
from ANOVA. These values indicate that about 4% of the variance in income
could be explained by different types of child care arrangements women use.
The tables for child two do not display such big discrepancies between counts and
expected counts, so the likelihood of an association is lower. ETA square is 0.033,
indicating that the variables share only about 3% of the common variance in
income.
50


Multiple Regression
The purpose of multiple regression is to predict an interval scale dependent variable
from a combination of several interval scale and/or dichotomous independent /
predictor variables.
I wanted to examine if there is any effect of child care arrangements on income.
There are other factors that can also influence income, for example education,
marital status, spouse's income etc. I chose multiple regression so I could include
additional variables into the analysis and compare their influence on income with
the presumed influence of the type of child care. I preferred to use stepwise
regression, in which the computer selects the variable that has the highest bivariate
correlation with the outcome variable and enters it into the equation. Next, the
predictor variable with the highest semipartial correlation is entered etc. Once a
variable is entered that no longer makes a statistically significant contribution, it is
removed (Morgan and Griego 1998).
Bivariate Correlations
The first step in multiple regression is to compute bivariate correlations. The
correlation tables with all variables considered for multiple regression are in
51


Appendix G. The tables show some expected and some unexpected results. Income
was positively correlated with age of the respondent and spouse's income and
negatively correlated with family size and rural residence. It is interesting that
using child care center as a mode of child care is positively correlated with
respondent's income, but in the use of a relative in a child's home the correlation is
negative. That is true for child care arrangements for both child one and child two.
Surprisingly, there seem to be no correlation in my sample between income and
cost of child care, marital status or race. Marital status was recoded twice, once as
married/single and once as never married/ever married. None of the marital status
variables were correlated with income on a statistically significant level.
Stepwise Regression
Because spouse's income showed as significantly correlated with respondent's
income, it was necessary to run each version of the regression twice once with
this variable and once without it to allow for women that were not married and thus
would not stay in the analysis with the "spouse's income" variable.
52


I ran the regressions separately for child one and child two and also for
arrangements "care by relative in child's own home" and "commercial day care".
Resulting regression models are in Appendix H. Excluding other arrangements did
not alter the regression results because they were not statistically significant as
shown in the correlation tables, and when I included them on a trial basis the final
coefficients did not change.
Although the complete regression tables including all coefficients are in Appendix
H, for greater clarity I insert two tables with an overview of the relevant
information. It would not be feasible to include all stepwise regression tables in the
text.
Tables 5.2 and 5.3 show statistically significant predictors of income in the above
described regression models. Variables in the tables are pictured in the order in
which they were entered into the equation. Adjusted R Square is "cumulative", for
each step after the first one it indicates the prediction form the combination of
variables, e.g.; Adjusted R Square when only education is entered is 0.082, for
education and arrangement "relative in home" it is 0.116 and so forth. Plus or
53


minus signs indicate if individual predictors are associated with increased or
decreased income.
Table 5.2: Multiple Regression Models with "Spouse's Income" Variable
Included
Child 1 Child 2
Relative Commercial Relative Commercial
in Home Care in Home Care
Predictors Adj.R2 Predictors Adi.R2 Predictors Adj.R2 Predictors Adi.R2
Education+ 0.082 Education+ 0.082 EducationH- 0.058 Education-)- 0.058
Relative in- 0.116 Ruralness- 0.101 Relative in- 0.099 Spouse inc.-t- 0.073
Ruralness- 0.136 Commerc.+ 0.109 Ruralness- 0.116 Non-white+ 0.094
Table 5.3: Multiple Regression Models with "Spouse's Income" Variable
Excluded
Child 1 Child 2
Relative Commercial Relative Commercial
in Home Care in Home Care
Predictors Adj.R2 Predictors Adi.R2 Predictors Adi.R2 Predictors Adj.R2
Education-i- 0.091 Education-!- 0.091 Education-!- 0.058 Education-!- 0.058
Relative in- 0.117 Ruralness- 0.104 Relative in- 0.078 Age+ 0.075
Ruralness- 0.131 Commerc.-i- 0.112 Age+ 0.093 Ruralness- 0.085
Age+ 0.135 Fam. Size- 0.116 Ruralness- 0.103
Married-t- 0.138 Age+ 0.122
Married+ 0.126
Non-white+ 0.128
Significant predictors of income in all models are: education, child care
arrangement "relative in child's home" and rural residence. Education increases
54


income, "relative in child's home" and rural residence are associated with a
decrease in income. Commercial day care is associated with an increase in income
when used for child one, but it is not statistically significant predictor when utilized
for child two. Spouse's income is significant only when considering commercial
child care for child two, in that case it is associated with an increase in respondent's
income. In the same instance there is also a weak positive influence on income by
"nonwhiteness."
When spouse's income is not considered and, therefore, women who are not
married or did not report their husband's income are included, the picture changes
somewhat, but not substantially. Education, "relative in child's home" arrangement
and rural residence are always significant, commercial child care is significant only
in child one cases. In all four models age becomes statistically significant. The
coefficients are similar for child one and child two child care arrangement
variables.
Percentages of the variations in income as denoted by Adjusted R Square vary for
different arrangements, from 0.8% for commercial day care to 4.1% for relative in
home care (married women).
55


The tolerances of excluded variables in all models are high, which alleviates
concerns about multicollinearity (Morgan and Griego 1998).
The results of multiple regression indicate that type of child care arrangement does
influence income of women with children. We reject the null hypothesis. The
alternate hypothesis is supported by the results of the analysis.
Logistic Regression
When looking at the relationship between child care and income one can easily
agree that income influences child care also. Because type of child care
arrangement as used in previous analysis is a nominal variable, appropriate
statistics to use in this case would be logistic regression. Logistic regression
calculates changes in the log odds of the dependent variable, not changes in the
dependent variable itself (as in multiple regression). Logistic regression does have
many analogies to linear regression. Logit coefficients correspond to b coefficients
in the equation, standardized logit coefficients correspond to beta weights, and a
pseudo R Square statistic is available (Garson 2002).
56


I tried to predict if a woman would use certain type of child care or not. I limited
my analysis to those types of care that were significant predictors of income:
"relative in child's home" and "commercial day care" variables.
Independent variables were: income and other variables already used. I used
method=Enter for the analysis, which is considered better when testing theory than
the stepwise method.
Computing the logistic regression was similar to the multiple regression. I created
models with and without the spouse's income variable and for the two forms of
care: "relative in the child's home" and "commercial day care."
Six out of the eight models were significant (using Chi-square). Chi-square
assesses the overall model, but does not indicate which independent variables are
more important than others. The two models that were not significant were
supposed to predict the usage of commercial child care for child two.
A measure called -2 Log likelihood should not be significant for a good-fitting
model, because it signalizes "badness of fit." The Wald statistics tests for
57


significance of individual logistic regression coefficients for each independent
variable, but the SPSS output simplifies the work of the researcher by listing the
significance directly.
The most important tables in the logistic regression analysis are the classification
tables. The columns list the predicted values of the dependent variable, the rows the
actual (observed) values. However, overall percentages in logistic regression may
be misleading. At a closer look one can discern that the models predicted better
who would not use certain type of child care arrangement than who would use it.
The complete results of the logistic regression for all models are in Appendix I. To
Tables 5.4 and 5.5 show only the significant predictors of specific child care
arrangements.
Table 5.4. Significant Predictors for Specific Types of Child Care
Arrangements (Spouse's Income Included)
Child 1 Child 2
Relative in Home Commercial Care Relative in Home Commercial Care
Predictors Sig. Predictors Sig. Predictors Sig. Predictors Sig.
Income .000 Age .000 Income .000
Ruralness .020 Income .027 Fam. Size .001
Fam. Size .000
58


Table 5.5. Significant Predictors for Specific Types of Child Care
Arrangements (Spouse's Income Excluded)
Child 1 Child 2
Relative in Home Commercial Care Relative in Home Commercial Care
Predictors Sig. Predictors Sig. Predictors Sig. Predictors Sig.
Age .003 Age .000 Income .001
Income .000 Income .009 Fam. Size .000
Fam. Size .000
In all models income was a significant predictor of child care arrangements
considered. Age of the respondent and family size also appeared, although not in
every logistic regression model. When trying to measure the strength of the
association, one does not have an R Square in the logistic regression. Cox and
Snell's R Square and its modification, Nagelkerke's R Square are part of the SPSS
output, but their interpretation is problematic, and, I believe, not universally agreed
upon.
The SPSS converts the logit coefficients into odds ratios Exp (B). When the
independent variable increases one unit, the odds that the dependent variable equals
one increase by a factor of ten, when other variables are controlled. Thus the odds
ratio does give some idea about the strength of the association. The odds ratios vary
(for the variables deemed significant in the models) from .568 for rural residence
59


(relative in child one's home, spouse's income included) to 1.778 for family size
(relative in child two's home, spouse's income excluded). Although income shows
as significant in all significant models, the odds ratio for income is 1.000, which
would indicate, that a change in value of income is not associated with change in
the odds of the dependent variable. I have no decisive explanation for this
phenomenon other than that the income is a continuous variable and the one dollar
increase or decrease might not be large enough to register a change when we are
dealing with a dichotomous child care variable. The significance means that the
coefficient is significantly different from zero. I did not find a text that would
address this specific situation; it could also be a false negative, because the Wald
statistics used for determining the significance is prone to errors (Garson 2002).
60


Summary
Results of several statistical analyses indicate that there is, indeed, an association
between type of child care arrangement women use and their income. This
relationship is reciprocal, child care appears to influence income and, in turn,
income influences child care women use. Overall, the differences in child care
arrangements are associated with about four percent variation in women's income.
Women, who use relatives in child's own home for care, have lower income than
other women; women using commercial child care have income slightly higher.
The results were consistent among different statistical methods: linear regression
(stepwise), ANOVA, and cross tabulation with ETA.
61


CHAPTER 6
DISCUSSION AND IMPLICATIONS
Discussion and Conclusions
In my study I proposed that child care arrangements would have an effect on the
income of working women with young (preschool age) children. I used multiple
linear regression to predict income from the nature of child care in combination
with a group of other variables that are known to influence income (education,
urban or rural residence, race, marital status, family size, spouse's income).
Regression, as well as ETA statistics' results support the hypothesis that type of
child care arrangement women use has an effect on how much money they make,
explaining a small percentage of the variation in each of the models.
When looking at how child care influences income one can see that using relative
in child's own home is associated with decreased income, while using commercial
child care is associated with slightly higher income.
62


Limitations of the study must be acknowledged in order to enable consideration of
its implications. The validity of the study was probably influenced by the way the
child care variables were operationalized. There is no standardized or generally
agreed upon way to operationalize different types of child care arrangements and
the rather crude measures I used played their role. Using secondary data also makes
operationalization difficult and influences validity of the research. My study was
cross-sectional; analyzing longitudinal data might provide a clearer picture of the
job trajectories and "child care trajectories" of working women. Longitudinal study
would also solve the time reference problems with decision making: Do
employment decisions or child care decisions come first or are they simultaneous?
Previous research on the topic has been inconsistent. The variation in income
explained by different child care arrangements was small, but I believe the issue
warrants further research to clarify the relationships between child care and
women's earnings.
Literature I reviewed does not specify in which ways different child care
arrangements could influence the income and with the data I have available I
cannot conclusively determine the mechanism working here.
63


If I were to speculate, I would look at reasons why mothers choose certain mode of
care, specifically care by relatives in the child's home or commercial day care.
Riley and Glass (2002) report that mothers do not always use their preferred type of
care and sometimes they do not have a choice and have to use "what is available."
The author also stresses that earnings of mothers returning to the labor force and
who have young children might not be a strong enough market force to establish a
true market choice in child care arrangements. In addition, care by relatives is not
marketable, so the economic and social implications of market behavior in this case
are complicated.
At the same time Henly and Lyons (2000) describe how informal child care, which
includes care by relatives, is often unreliable or unstable and causes problems at
work for women that use it. Henly and Lyons also note that there has not been
much research investigating informal arrangements, with which statement I agree. I
suppose that repeated problems with child care could influence a position a woman
would be able to hold and can also diminish chances for promotions and transfers.
Thus the reliability or stability or lack of them in a certain type of child care
arrangement could influence women's income.
64


Budig and England (2001) found a five percent per child wage penalty, even after
controlling for experience and other factors. The authors used the human capital
theory as their theoretical basis and did not take child care into account. I proposed
that at least part of this penalty can be due to child care factors. Care by relatives in
child's own home, which is widely used, seems to explain about four percent of
variability in income and it does influence the income in a negative way.
Without discounting the human capital theory I proposed that child care factors are
important in determining income of working women, because they influence the
compatibility of productive and reproductive labor as outlined in the Integrated
Theory of Gender Stratification (Collins, Chafetz, Blumberg, Coltrane, and Turner
1993). Due to the lack of previous research and/or theoretical framework I did not
attempt to predict the direction or the precise mechanism of this influence, but the
results of my analysis support the concept of child care arrangements having an
effect on income. It is possible that the compatibility of productive/reproductive
labor is higher while using commercial child care centers than when utilizing other
modes of care and significantly lower for the "relative in home" arrangement.
65


As for commercial child care, with which the women's income is somewhat higher,
the reliability of the arrangement can also play a role here. I have not found any
studies focusing on reliability of different child care arrangements. Hofferth (1992)
writes that when parents were asked about their child care choices, "Of those who
felt quality of care was important, 6 out of 10 mentioned a characteristic of the
provider, such as a warm and loving style, reliability, training, or experience as
most important."
Waite, Leibowitz and Witsberger (1991) point to structural, processual, and
"outcome of care" variables that can be important in child care choice. Reliability
would, in my opinion, belong to the processual variables group. Parents may
consciously choose a form of care because of its reliability, among other factors,
but the reliability of a certain type care chosen for other reasons (cost, availability)
can be influencing women's income.
Logistic regression showed that income influences types of child care arrangements
used. This result is in line with my assumption that the relationship between child
care and income is reciprocal one influences the other. The most obvious
implication of this finding would be to assume that women with higher incomes
66


can afford more expensive child care arrangements and that is why they are using
commercial child care centers. In the sample I used, however, the cost of child care
was not correlated with income at all.
Implications of the Study and Recommendations
Commercial day care arrangements are associated with higher income, while using
relative in the child's home is associated with decreased income. The mechanism of
this process could not be determined at this time, nor was it my goal to do so. One
of possible explanations could be connected to reliability (or stability) of different
child care arrangements. If further research determines that this is, indeed, the case,
it would then be logical to encourage and support (even in monetary form or in the
form of tax credits and subsidies) the use of child care centers. If women's income
would rise after switching from informal arrangements to organized group centers,
we would have discovered a simple, albeit expensive, way to enable women access
to economic resources that, in turn, would hopefully lead to lowering gender
inequality and to improvement of women's standing in society.
It would be interesting to duplicate my study in several years when the next NLS Y
cohort, the NLSY97 is old enough to be employed and have young children. A
67


large study in itself would be a longitudinal analysis of changes in employment and
changes in child care arrangements over several years. It could also be useful to
repeat Budig and England's (2001) study of the wage penalty for motherhood with
child care arrangements included in the analysis. All this proposed research would
be complicated and time consuming, but not impossible.
I would also recommend that researchers studying child care issues from any angle
pay more attention to the processual characteristics of child care.
The career trajectory seen in current American society might be called a "male"
model, because the conditions for successful career do not allow for
accommodation of reproductive labor. Yet as more families choose the parenting
approach to child care, as opposed to the mothering or market approaches (Hertz
1997), fathers will face the same adjustments in their careers that women face now.
Hopefully we will then start to create a universal, "human" model/approach to both
careers (productive labor) and child rearing (reproductive labor).
During the course of this research it surprised me very much how little research has
been done on child care in connection with women's employment and how
68


inconsistent the reports were. Yet child care is of highest concern to families with
children. I hope I contributed, however little it might have been, to greater
understanding of this topic.
69


APPENDIX
APPENDIX A. LIST OF EXTRACTED VARIABLES
National Longitudinal Surveys of Youth 1979 Public 2002-04-09 : Extract
Data File Name is: complete tagset.DAT
Created on Wednesday, August 27, 2002
Extract run begun at: 22:26:25.00
Extract run finished at: 22:27:39.00
Data Record is Fixed, Length is: 113 Characters
Total Number of Variables per Case on complete tagset.DAT is: 44
Total Number of Cases is: 12686
National Longitudinal Surveys of Youth 1979 Public 2002-04-09 DB
Investigator Extract Page 1 of 1
Variable Minimum Label Maximum N Mean
R0000100 1 ID# (1-12686) 79 12686 1 5 (5) 12686 6343.500
R0006500 0 HGC BY RS MOTHER 79 20 6-7 (2) 11878 10.869
R0006900 1 OCC OF LNGST JOB 78 STEP/MOTHERS 3D 984 8-10 (3) 79 7117 556.778
R0007900 0 HGC BY RS FATHER 79 20 11 12 (2) 10880 10.947
R0008300 1 OCC OF LNGST JOB 78 STEP/FATHERS 3D 984 13 15 (3) 79 9015 493.829
R0149900 1 ATND R DEC PURSUE CAREER DELAY FAM 4 16-17 (2) 79 2548 3.004
R0214700 1 RACL/ETHNIC COHORT /SCRNR 79 3 18-19 (2) 12686 2.434
R0214800 1 SEX OF R 79 2 20-21 (2) 12686 1.495
R2503000 0 OCC M-RCNT SP 3D 88 996 22 24 (3) 5354 533.940
R2509800 1 HIGHEST DGR RCVD 88 8 25 26 (2) 8564 1.788
R2525400 1 OCC @ CUR/M-RCNT JOB 70 3D CP88 984 27 29 (3) 8989 477.458
R2654500 0 ICHK R WORK/ON AAF LAST 4 WKS F 88 1 30-31 (2) 3246 0.595
R2655600 1 PRNCLP C-CARE CHILD#1 LAST 4 WKS F 17 32 33 (2) 88 2655 8.128
70


R2655800 1 PRNCLP C-CARE CHILD#2 LAST 4 WKS F 88 17 34 35 (2) 1573 7.965
R2657200 1 PLACE PRNCLP ARNGMT (1-6) CHILD# 1 F 88 3 36 37 (2) 1401 1.565
R2657300 0 #HRS/WK CHILD#1 PRNCLP C-CARE F 88 996 38 40 (3) 2027 35.795
R2657500 1 #CHILD CARED PRNCLP C-CARE CHILDil F 88 80 41 42 (2) 1655 4.861
R2657600 1 #ADLT SPVSNG PRNCLP C-CARE CHILD#1 F 88 20 43 44 (2) 1668 1.511
R2657900 1 PLACE PRNCLP ARNGMT (1-6) CHILD#2 F 88 3 45 46 (2) 832 1.481
R2658000 1 #HRS/WK CHILD#2 PRNCLP C-CARE F 88 996 47 49 (3) 1205 35.732
R2658200 1 #CHILD CARED PRNCLP C-CARE CHILD#2 F 88 60 50 51 (2) 982 4.556
R2658300 1 #ADLT SPVSNG PRNCLP C-CARE CHILD#2 F 88 41 52 53 (2) 991 1.489
R2662700 0 PERSON RESP C-CARE CHILD#1 SPCL TRNG 88 1 54-55 (2) 499 0.499
R2662800 0 PRNCLP C-CARE ONLY 1 USED CHILD#1 F 88 1 56 57 (2) 2623 0.826
R2663100 0 PERSON RESP C-CARE CHILD#2 SPCL TRNG 88 1 58 59 (2) 279 0.427
R2663200 0 PRNCLP C-CARE ONLY 1 USED CHILD#2 F 88 1 60-61 (2) 1533 0.855
R2670800 0 ICHK CHILD#2 LIVING IN HH FEMALE 88 1 62 63 (2) 2656 0.572
R2672300 0 ANY PMT MADE BY R/SP/PRTNR F 88 1 64 65 (2) 2631 0.420
R2672400 1 AMT/WK PD C-CARE BY R/SP/PRTNR F 88 996 66 68 (3) 1097 55.413
R2 672500 0 ANY NONCASH PMT BY R/SP/PRTNR F 88 1 69-70 (2) 2629 0.052
R2672600 0 R/SP LOSE WRK 4 WKS NO C-CARE F 88 1 71-72 (2) 2627 0.032
R2673200 0 AGE BIO CHILD#1 LAST BDAY F 88 14 73 74 (2) 2302 2.976
R2673900 0 AGE BIO CHILD#2 LAST BDAY F 88 13 75 76 (2) 790 1.692
R2870110 1 FAMILY SIZE 88 15 77 78 (2) 10465 3.054
R2870600 1 EMPLOYMENT STATUS RECODE 88 8 79-80 (2) 10465 1.951
R2871000 0 MARITAL STATUS 88 6 81-82 (2) 10461 0.839
R2871101 0 HIGHEST GRADE COMPLTD (REV) 88 20 83 84 (2) 10424 12.704
R2871300 23 AGE OF R 0 INT DATE 88 32 85 86 (2) 10465 27.088
R2872700 0 RS CURRENT RESIDENCE URBAN/RURAL 88 1 87-88 (2) 9976 0.794
R2877600 0 # OF BIO/STEP/ADPT CHILDREN IN HH 88 9 89 90 (2) 10465 0.907
R2971400 0 TOT INC WAGES AND SALRY P-C YR 89 566028 91 96 (6) 10326 14724.278
R2973600 0 TOT INC SP WAGE AND SALRY P-C YR 89 427830 97 102 (6) 4986 18541.856
71


R3074000
0
R3075100
0
TOT NET FAMILY INC P-C YR 89 8723 47279.064
1057448 103 109 (7)
# OF HRS WRKD IN P-C YR 89 10394 1612.245
4992 110 113 (4)
72


APPENDIX B. CODEBOOK ENTRIES FOR EXTRACTED VARIABLES
National Longitudinal Surveys of Youth 1979 Public 2002-04-09
Codebook
This Report Was Created on August 27, 2002 at 22:27.
R00001.00 [CASEID] IDENTIFICATION CODE
Survey Year: 1979
NOTE: THIS NUMBER WAS ASSIGNED ACCORDING TO THE
RECORD NUMBER OF EACH RESPONDENT ON THE TAPE.
THE FIRST RESPONDENT WAS ASSIGNED ID#1, THE SECOND
WAS ASSIGNED ID#2, ETC.
ACTUAL NUMBER
Refusal(-1)
Don't Know(-2)
TOTAL =========>
Interview(-5)
Min: 1
Max: 12686
Lead In: None.
Default Next Question: R00001.10
0
0
12686 Valid Skip(-4) 0 Non-
0
Mean: 6343.5
R00065.00 [S01Q16] HIGHEST GRADE COMPLETED BY R'S MOTHER
Survey Year: 1979
WHAT IS THE HIGHEST GRADE OR YEAR OF REGULAR SCHOOL THAT YOUR
MOTHER
EVER COMPLETED?
132 0 NONE
73


24 1 1ST GRADE
87 2 2ND GRADE
183 3 3RD GRADE
173 4 4TH GRADE
198 5 5TH GRADE
421 6 6TH GRADE
260 7 7TH GRADE
801 8 8TH GRADE
698 9 9TH GRADE
999 10 10TH GRADE
1107 11 11TH GRADE
4817 12 12TH GRADE
364 13 1ST YR COL
561 14 2ND YR COL
178 15 3RD YR COL
647 16 4TH YR COL
101 17 5TH YR COL
92 18 6TH YR COL
14 19 7TH YR COL
21 20 8TH YR COL OR MORE
0 95 UNGRADED
11878
Refusal (-1) 3
Don't Know(-2) 527
Invalid Skip(- 3) 251
TOTAL Interview(-5) =-> 12659 Valid Skip 0
Lead In: R00064. 00[Default]
Default Next Question: R00066.00
R00069.00 [S01Q19A] OCCUPATION OF LONGEST JOB IN 1978, R'S
MOTHER/STEPMOTHER
(CENSUS 3 DIGIT)
Survey Year: 1979
WHAT KIND OF WORK WAS SHE DOING? (IF MORE THAN ONE KIND OF WORK
PROBE:
DURING 1978, WHAT KIND OF WORK DID SHE DO THE LONGEST?) WHAT WERE
74


SOME OF HER MAIN ACTIVITIES OR DUTIES? (PROBE FOR 2 MAIN
ACTIVITIES)
SEE ATTACHMENT 3 FOR DEFINITIONS
842 1 to 195: 001-195 PROFESSIONAL, TECHNICAL AND
KINDRED
315 201 to 245: 201-245 MANAGERS,OFFICIALS AND
PROPRIETORS
363 260 to 285: 260-285 SALES WORKERS
1760 301 to 395: 301-395 CLERICAL AND KINDRED
166 401 to 575: 401-575 CRAFTSMEN,FOREMEN AND KINDRED
1 580 to 590: 580-590 ARMED FORCES
1368 601 to 715: 601-715 OPERATIVES AND KINDRED
72 740 to 785: 740-785 LABORERS, EXCEPT FARM
12 801 to 802: 801-802 FARMERS AND FARM MANAGERS
147 821 to 824: 821-824 FARM LABORERS AND FOREMAN
1698 901 to 965: 901-965 SERVICE WORKERS, EXCEPT
PRIVATE HOUSEHOLD
373 980 to 984: 980-984 PRIVATE HOUSEHOLD
0 0 : 00 NONE
990: 990 SAME AS PRESENT JOB
995: 995 DID NOT WORK
996: 996 NEVER WORKED
7117
Refusal(-1) 3
Don't Know(-2) 80
Invalid Skip(-3) 273
TOTAL =======> 7473
Interview(-5) 0
Valid Skip(-4)
5213
Non-
Min:
Max:
1
984
Mean:
556.78
Lead In: R00068.00[Default]
Default Next Question: R00070.00
R00079.00 [S01Q23] HIGHEST GRADE COMPLETED BY R'S FATHER
Survey Year: 1979
LET'S GO BACK TO YOUR FATHER NOW. WHAT IS THE HIGHEST GRADE OR YEAR
OF
75


REGULAR SCHOOL THAT YOUR FATHER EVER COMPLETED?
199 0 NONE
48 1 1ST GRADE
102 2 2ND GRADE
215 3 3RD GRADE
208 4 4TH GRADE
190 5 5TH GRADE
499 6 6TH GRADE
306 7 7TH GRADE
950 8 8TH GRADE
568 9 9TH GRADE
754 10 10TH GRADE
616 11 11TH GRADE
3694 12 12TH GRADE
286 13 1ST YR COL
558 14 2ND YR COL
162 15 3RD YR COL
899 16 4TH YR COL
151 17 5TH YR COL
209 18 6TH YR COL
56 19 7TH YR COL
210 20 8TH YR COL OR MORE
0 95 UNGRADED
10880
Refusal(-1) 3
Don't Know(-2) 1150
Invalid Skip(-3) 423
TOTAL > 12456 Valid Skip
Interview(-5) 0
Lead In: R00078.00[Default]
Default Next Question: R00080.00
R00083.00 [S01Q26A] OCCUPATION OF LONGEST JOB IN 1978, R'S
FATHER/STEPFATHER
(CENSUS 3 DIGIT)
Survey Year: 1979
WHAT KIND OF WORK WAS HE DOING? (IF MORE THAN ONE KIND OF WORK
PROBE:
76


DURING 1978, WHAT KIND OF WORK DID HE DO THE LONGEST?) WHAT WERE
SOME
OF HIS MAIN ACTIVITIES OR DUTIES? (PROBE FOR 2 MAIN DUTIES)
SEE ATTACHMENT 3 FOR DEFINITIONS
1025 1 to 195: 001-195 PROFESSIONAL, TECHNICAL AND
KINDRED
1219 201 to 245: 201-245 MANAGERS,OFFICIALS AND
PROPRIETORS
410 260 to 285: 260-285 SALES WORKERS
387 301 to 395: 301-395 CLERICAL AND KINDRED
2310 401 to 575: 401-575 CRAFTSMEN,FOREMEN AND KINDRED
142 580 to 590: 580-590 ARMED FORCES
1732 601 to 715: 601-715 OPERATIVES AND KINDRED
602 740 to 785: 740-785 LABORERS, EXCEPT FARM
236 801 to 802: 801-802 FARMERS AND FARM MANAGERS
250 821 to 824: 821-824 FARM LABORERS AND FOREMAN
697 901 to 965: 901-965 SERVICE WORKERS, EXCEPT
PRIVATE HOUSEHOLD
5 980 to 984: 980-984 PRIVATE HOUSEHOLD
0 0 : 00 NONE
0 990 : 990 SAME AS PRESENT JOB
995: 995 DID NOT WORK
996: 996 NEVER WORKED
9015
Refusal(-1)
Don't Know(-2)
Invalid Skip(-3)
TOTAL ======>
Interview(-5)
Min:
Max:
1
984
10
206
464
9695
0
Mean:
Valid Skip(-4)
493.83
2991
Non-
Lead In: R00082.00[Default]
Default Next Question: R00084.00
R01499.00 [S17Q03H] ATTITUDE TOWARD R'S DECISION TO PURSUE
CAREER, DELAY
FAMILY
Survey Year: 1979
77


SEE R(1492.) ASK FEMALE R'S ONLY: YOU DECIDED TO PURSUE A FULL TIME
CAREER AND DELAY STARTING A FAMILY
154 1 STRONGLY DISAPPROVE
4 64 2 SOMEWHAT DISAPPROVE
1149 3 SOMEWHAT APPROVE
781 4 STRONGLY APPROVE
2548
Refusal(-1) 0
Don't Know(-2) 62
Invalid Skip(-3) 61
TOTAL > 2671 Valid Skip(-4) 10015
Interview(-5) 0
Lead In: R01498.00(Default]
Default Next Question: R01500.00
R02147.00 [*Created] R'S RACIAL/ETHNIC COHORT
Survey Year: 1979
2002 1 HISPANIC
3174 2 BLACK
7510 3 NON-BLACK, NON-HISPANIC
12686
Refusal(-1) 0
Don't Know(-2) 0
TOTAL ========= > 12686 Valid Skip(-4
Interview(-5) 0
NOTE: AUGUST 26 i, 1980 DENNIS GREY
0
DESCRIPTION:
COHORT=3;
IF R(1736.) =4 R(1736.)=8 R(1736.)=H R(1736
COHORT=l;
Non-
FROM SCREENER
Non-
)=14 THEN
78


IF R(1736.)=17 R(1736.)=20 THEN C0H0RT=1;
IF R(1736.)=3 R(1736.)=7 R(1736.}=10 R(1736.)=13 THEN
C0H0RT=2; '
IF R(1736.)=16 R(1736.)=19 THEN C0H0RT=2;
R (2147.)=COHORT;
Lead In: R02146.00[Default]
Default Next Question: R02148.00
R02148.00 [*Created] SEX OF R
Survey Year: 1979
6403 1 MALE
6283 2 FEMALE
12686
Refusal(-1) 0
Don't Know(-2) 0
TOTAL =========> 12686 Valid Skip(-4) 0 Non-
Interview (-5 ) 0
NOTE: AUGUST 26, 1980 DENNIS GREY
MARCH 1, 1986 NORC
DESCRIPTION:
IF (R(1736.)>0 & R(1736.)<=4) (R(1736.)>=9 & R(1736.)<=11) THEN
SEX=1;
IF R(1736.)>=15 & R(1736.)<=17 THEN SEX=1;
IF (R(1736.)>=5 & R(1736.)<=8) (R(1736.)>=12 & R(1736.)<=14) THEN
SEX=
IF R(1736.)>=18 & R(1736.)<=20 THEN SEX=2;
79


R (2148.)=SEX;
SEX CODE CHANGED ON 42 CASES
Lead In: R02147.00[Default]
Default Next Question: R02149.00
R25030.00 [Q2.5] OCCUPATION OF MOST RECENT SPOUSE, (CENSUS 3
DIGIT)
Survey Year: 1988
DURING 1987, WHAT KIND OF WORK DID YOUR (MOST RECENT)
(HUSBAND/WIFE)
DO? WHAT WERE (HIS/HER) MAIN ACTIVITIES OR DUTIES?
SEE ATTACHMENT 3, INDUSTRY AND OCCUPATION CODES
726 1 to 195:
KINDRED
506 201 to 245:
PROPRIETORS
263 260 to 285:
773 301 to 395:
664 401 to 575:
149 580 to 590:
661 601 to 715:
271 740 to 785:
12 801 to 802:
47 821 to 824:
520 901 to 965:
PRIVATE HOUSEHOLD
26 980 to 984:
5 0 : 00
0 990 : 990
643 995 : 995
88 996 : 996
001-195 PROFESSIONAL, TECHNICAL AND
201-245 MANAGERS,OFFICIALS AND
260-285
301-395
401-575
580-590
601-715
740-785
801-802
821-824
901-965
SALES WORKERS
CLERICAL AND KINDRED
CRAFTSMEN,FOREMEN AND KINDRED
ARMED FORCES
OPERATIVES AND KINDRED
LABORERS, EXCEPT FARM
FARMERS AND FARM MANAGERS
FARM LABORERS AND FOREMAN
SERVICE WORKERS, EXCEPT
980-984 PRIVATE HOUSEHOLD
NONE
SAME AS PRESENT JOB
DID NOT WORK
NEVER WORKED
5354
Refusal(-1) 10
Don't Know(-2) 50
Invalid Skip(-3) 120
TOTAL > 5534
Interview(-5) 2221
Valid Skip(-4)
4931
Non-
80


Min: 0 Mean: 533.94
Max: 996
Lead In: R25029.00[Default]
Default Next Question: R25031.00
R25098.00 [Q3.9A] HIGHEST DEGREE EVER RECEIVED
Survey Year: 1988
(HAND CARD B) WHAT IS THE NAME OF THE HIGHEST DEGREE YOU HAVE EVER
RECEIVED?
6031 1 HIGH SCHOOL DIPLOMA (OR EQUIVALENT)
626 2 ASSOCIATE/JUNIOR COLLEGE (AA)
587 3 BACHELOR OF ARTS DEGREE (BA)
922 4 BACHELOR OF SCIENCE (BS)
178 5 MASTER'S DEGREE (MA,MBA,MS,MSW)
11 6 DOCTORAL DEGREE (PHD)
49 7 PROFESSIONAL DEGREE (MD,LLD,DDS)
160 8 OTHER
8564
Refusal(-1) 1
Don't Know(-2) 0
Invalid Skip(-3) 6
TOTAL ======> 8571 Valid Skip(-4) 1894
Interview(-5) 2221
Lead In: R25097.00[Default]
Default Next Question: R25099.00
R25254.00 [Q5.26A] OCCUPATION AT CURRENT JOB/MOST RECENT JOB
(70 CENSUS 3
DIGIT) CPS ITEM
Survey Year: 1988
WHAT KIND OF WORK WERE YOU DOING FOR THIS JOB? WHAT KIND OF WORK
WERE
81


YOU DOING FOR THE MOST HOURS LAST WEEK?
1356 1 to 195:
KINDRED
966 201 to 245:
PROPRIETORS
415 260 to 285:
1825 301 to 395:
1016 401 to 575:
0 580 to 590:
1274 601 to 715:
617 740 to 785:
14 801 to 802:
113 821 to 824:
1271 901 to 965:
PRIVATE HOUSEHOLD
122 980 to 984:
0 0 : 00
0 990 : 990
0 995 : 995
0 996 : 996
8989
001-195 PROFESSIONAL,TECHNICAL AND
201-245 MANAGERS,OFFICIALS AND
260-285
301-395
401-575
580-590
601-715
740-785
801-802
821-824
901-965
SALES WORKERS
CLERICAL AND KINDRED
CRAFTSMEN,FOREMEN AND KINDRED
ARMED FORCES
OPERATIVES AND KINDRED
LABORERS, EXCEPT FARM
FARMERS AND FARM MANAGERS
FARM LABORERS AND FOREMAN
SERVICE WORKERS, EXCEPT
980-984 PRIVATE HOUSEHOLD
NONE
SAME AS PRESENT JOB
DID NOT WORK
NEVER WORKED
Refusal (-1) 1
Don't Know(-2) 1
Invalid Skip(-3) 35
TOTAL ========> 9026 Valid Skip (-4) 1439 Non-
Interview (-5) 2221
Min: 1 Mean: 477.46
Max: 984
Lead In: R25253.00[Default]
Default Next Question: R25255.00
R26545.00 [Q10.2] INT CHECK R WORKED OR ON ACTIVE ARMED
FORCES DUTY IN
LAST 4 WEEKS? (FEMALE)
Survey Year: 1988
I: REFER TO CALENDAR ROWS A AND B. HAS RESPONDENT WORKED OR BEEN ON
ACTIVE DUTY IN THE PAST 4 WEEKS?
82


1932 1 YES
1314 0 NO
3246
Refusal(-1) 0
Don't Know(-2) 0
Invalid Skip(-3) 2
TOTAL > 3248 Valid Skip(-4) 7217
Interview(-5) 2221
Lead In: R26544.00[Default]
Default Next Question: R26546.00
R26556.00 [Q10.6A1] PRINCIPAL CHILD CARE ARRANGEMENT FOR 1ST
CHILD DURING
LAST 4 WEEKS (FEMALE)
Survey Year: 1988
(HAND CARD AA) DURING THE LAST 4 WEEKS, WHAT WAS (CHILD) USUALLY
DOING
OR HOW WAS (CHILD) USUALLY CARED FOR DURING MOST OF THE HOURS THAT
YOU
[(WORKED/PARTICIPATED IN YOUR ACTIVITY/ACTIVITIES) (USED
CHILDCARE)]?
322 1
552 2
4 3
4 4
17 5
211 6
22 7
292 8
195 9
80 10
35 11
1 12
363 iL 13
122 14
97 15
296 16
CHILD'S OTHER PARENT/STEPPARENT
CHILD'S GRANDPARENT
CHILD'S SIBLING UNDER AGE 15
CHILD'S SIBLING AGE 15 OR OVER
OTHER RELATIVE OF CHILD UNDER AGE 15
OTHER RELATIVE OF CHILD AGE 15 OR OLDER
NONRELATIVE OF CHILD UNDER AGE 15
NONRELATIVE OF CHILD AGE 15 AND OVER
CHILD IN DAY CARE CENTER OR GROUP CARE CENTER
CHILD IN NURSERY SCHOOL OR PRESCHOOL
CHILD IN DAY CAMP
CHILD IN OVERNIGHT RESIDENCE CAMP
CHILD IN KINDERGARTEN, ELEMENTARY, OR SECONDARY
CHILD CARES FOR SELF
R'S WORK OR ACTIVITY AT HOME
R CARES FOR CHILD AT WORK OR PLACE OF ACTIVITY
83


42 17 OTHER ARRANGEMENT
2655
Refusal(-1) 0
Don't Know(-2) 0
Invalid Skip(-3) 12
TOTAL > 2667
Interview(-5) 2221
Valid Skip(-4) 7798
Lead In: R26555.00[Default]
Default Next Question: R26557.00
Non-
R26558.00 [Q10.6A2] PRINCIPAL CHILD CARE ARRANGEMENT FOR 2ND
CHILD DURING
LAST 4 WEEKS (FEMALE)
Survey Year: 1988
SEE R(26556.)
235
290
15
5
11
135
13
205
99
31
16
1
153
SCHOOL
45
78
222
19
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
1573
Refusal(-1)
Don't Know(-2)
Invalid Skip(-3)
CHILD'S OTHER PARENT/STEPPARENT
CHILD'S GRANDPARENT
CHILD'S SIBLING UNDER AGE 15
CHILD'S SIBLING AGE 15 OR OVER
OTHER RELATIVE OF CHILD UNDER AGE 15
OTHER RELATIVE OF CHILD AGE 15 OR OLDER
NONRELATIVE OF CHILD UNDER AGE 15
NONRELATIVE OF CHILD AGE 15 AND OVER
CHILD IN DAY CARE CENTER OR GROUP CARE CENTER
CHILD IN NURSERY SCHOOL OR PRESCHOOL
CHILD IN DAY CAMP
CHILD IN OVERNIGHT RESIDENCE CAMP
CHILD IN KINDERGARTEN, ELEMENTARY, OR SECONDARY
CHILD CARES FOR SELF
R'S WORK OR ACTIVITY AT HOME
R CARES FOR CHILD AT WORK OR PLACE OF ACTIVITY
OTHER ARRANGEMENT
0
0
4
84


TOTAL ========> 1577 Valid Skip(-4) 8888 Non-
Interview (-5) 2221
Lead In: R26557.00[Default]
Default Next Question: R26559.00
R26572.00 [Q10.6C1] PLACE OF PRINCIPAL CHILD CARE ARRANGEMENT
(CODES 1-8)
FOR 1ST CHILD (FEMALE)
Survey Year: 1988
WHERE WAS (CHILD) USUALLY CARED FOR UNDER THIS ARRANGEMENT?
633 1 CHILD'S RESIDENCE
745 2 OTHER PRIVATE HOME
23 3 OTHER PLACE
1401
Refusal (-1) 0
Don't Know(-2) 0
Invalid Skip(-3) 44
TOTAL =========> 1445 Valid Skip(-4) 9020
Interview(-5) 2221
Non-
Lead In: R26571.00[Default]
Default Next Question: R26573.00
R26573.00 [Q10.6D1] # OF HOURS PER WEEK 1ST CHILD USUALLY IN
PRINCIPAL CHILD
CARE (FEMALE)
Survey Year: 1988
ABOUT HOW MANY HOURS PER WEEK WAS (CHILD) USUALLY CARED FOR UNDER
THIS
ARRANGEMENT?
996 OVERNIGHT RESIDENCE CAMP
ACTUAL HOURS
1 0
85


566 1 to 9
218 10 to 19
219 20 to 29
211 30 to 39
550 40 to 49
127 50 to 59
21 60 to 69
11 70 to 79
6 80 to 89
3 90 to 99
94 100 to 9999999: 100+
2027
Refusal(-1) 0
Don't Know(-2) 4
Invalid Skip(-3) 270
TOTAL =====> 2301
Interview(-5) 2221
Valid Skip(-4)
8164
Non-
Min:
Max:
0
996
Mean:
35.8
Lead In: R26572.00[Default]
Default Next Question: R26574.00
R26575.00 [Q10.8.1] # OF CHILDREN CARED FOR IN PRINCIPAL CHILD
CARE FOR 1ST
CHILD (FEMALE)
Survey Year: 1988
NOW I WOULD LIKE TO ASK YOU ABOUT OTHER ASPECTS OF (CHILD)'S
CURRENT
MAIN CARE ARRANGEMENT, THAT IS (MAIN CARE PROVIDER IN Q.6A, PAGE
10-
128). INCLUDING (CHILD), HOW MANY CHILDREN ARE CARED FOR TOGETHER,
IN
THE SAME GROUP, AT THE SAME TIME? (DO NOT INCLUDE CHILDREN IN THE
ENTIRE SCHOOL/CAMP/PROGRAM.)
ACTUAL NUMBER
0 0
86


1 to 4
5 to 9
10 to 14
15 to 19
20 to 24
25 to 29
30 to 34
35 to 39
40 to 44
45 to 49
50 to 9999999:
1240
200
75
45
39
19
18
1
5
1
12
1655
Refusal(-1) 0
Don't Know(-2) 14
Invalid Skip(-3) 61
TOTAL =========> 1730
Interview(-5) 2221
Min: 1 Mean:
Max: 80
Valid Skip(-4) 8735
4.86
Lead In: R26574.00[Default]
Default Next Question: R26576.00
Non-
R26576.00 [Q10.9.1] # OF ADULTS SUPERVISING CHILDREN, PRINCIPAL
CHILD CARE
FOR 1ST CHILD (FEMALE)
Survey Year: 1988
HOW MANY PEOPLE SUPERVISE [YOUR (CHILD/CHILDREN/THE (# IN Q.8)
CHILDREN IN THAT GROUP]? (PROBE:) HOW MANY TEACHERS ARE IN THAT
CLASS?
ACTUAL NUMBER
0 0:
1223 1
278 2
77 3
49 4
12 5
11 6
87


to 99999: 17+
7
8
9
10
11
12
13
14
15
16
17
2
6
0
3
2
1
1
0
2
0
1
1668
Refusal(-1) 1
Don't Know(-2) 4
Invalid Skip(-3) 57
TOTAL =======> 1730
Interview(-5) 2221
Min: 1 Mean:
Max: 2 0
Valid Skip(-4) 8735
1.51
Lead In: R26575.00[Default]
Default Next Question: R26577.00
Non-
R26579.00 [Q10.6C2] PLACE OF PRINCIPAL CHILD CARE ARRANGEMENT
(CODES 1-8)
FOR 2ND CHILD (FEMALE)
Survey Year: 1988
SEE R(26572.)
441 1 CHILD'S RESIDENCE
382 2 OTHER PRIVATE HOME
9 3 OTHER PLACE
832
Refusal(-1) 0
Don't Know(-2) 0
Invalid Skip(-3) 24
TOTAL > 856 Valid Skip(-4) 9609
Interview(-5) 2221
88


Lead In: R26578.00[Default]
Default Next Question: R26580.00
R26580.00 [Q10.6D2] # OF HOURS PER WEEK 2ND CHILD USUALLY IN
PRINCIPAL CHILD
CARE (FEMALE)
Survey Year: 1988
SEE R(26573.)
996 OVERNIGHT RESIDENCE CAMP
ACTUAL HOURS
0 0
382 1 to 9
123 10 to 19
114 20 to 29
114 30 to 39
314 40 to 49
71 50 to 59
11 60 to 69
8 70 to 79
4 80 to 89
2 90 to 99
62 100 to 9999999: 100+
1205
Refusal(-1) 0
Don't Know(-2) 4
Invalid Skip(-3) 179
TOTAL =======> 1388
Interview(-5) 2221
Min: 1 Mean:
Max: 996
Lead In: R26579.00[Default]
Default Next Question: R26581.00
Valid Skip(-4) 9077 Non-
-35.73
89