HHS Public Access

Author manuscript
Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
Author Manuscript

Published in final edited form as:

Soc Sci Res. 2018 May ; 72: 207–224. doi:10.1016/j.ssresearch.2018.02.003.

Mothers' Work Patterns and Children's Cognitive Achievement:

Evidence from the India Human Development Survey
Kriti Vikrama, Feinian Chenb, and Sonalde Desaib,c
aDepartment of Sociology, National University of Singapore, Singapore
bDepartment of Sociology, 2112 Art-Sociology Building, University of Maryland, College Park, MD
20742
Author Manuscript

cNational Council of Applied Economic Research, New Delhi, India

Abstract
As female labor force participation increases globally, the relationship between maternal
employment and children's development prospects remains unclear. Using data from the India
Human Development Survey (2005), we investigate the link between maternal employment and
children's arithmetic and reading achievement. We develop a work pattern typology that goes
beyond standard measures of employment and captures work intensity and its compatibility with
child-rearing in a transitional economy. We find that the relationship between maternal
employment and children's outcome is not unidimensional. For example, children of self-
employed mothers are not disadvantaged compared to those with stay-at-home mothers, but
Author Manuscript

maternal employment in salaried jobs or wage work outside the home is negatively associated with
cognitive skills in children. However, this negative association is reversed at higher levels of
maternal education, suggesting greater access to resources and flexibility associated with better
jobs mitigating the negative aspects of maternal employment posed by time constraints.
Additionally, maternal employment is associated with maternal involvement in schoolwork and
financial investment in academic activities, providing evidence that both time and resources
devoted to children's education are significant.

1. Introduction
As increasing female labor force participation becomes a global trend, a great deal of public
and scholarly attention has been paid to the effects of maternal employment on children's
Author Manuscript

well-being around the world. Among various indicators of children's welfare, including
health, education, development, and social functioning, children' cognitive growth has been
the research focus for many social scientists, as it is critically related to development in later
adulthood and has a long term impact on educational achievement and socioeconomic

Corresponding Author: Kriti Vikram, Department of Sociology, National University of Singapore, AS1 #04-28, 11 Arts Link,
Singapore 117570. [email protected].
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our
customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of
the resulting proof before it is published in its final citable form. Please note that during the production process errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Vikram et al. Page 2

attainment over the life course. However, despite myriad studies on the topic, the direction
Author Manuscript

and the strength of the relationship between maternal employment and children's
development prospects remains a topic of debate (Vandell and Ramanan, 1992; Parcel and
Menaghan, 1994; Ruhm, 2004; Johnson et al., 2012).

First, maternal employment potentially has both positive and negative consequences for
investments in children. Employment may impose demands on women's time that compete
with childcare or allocation of time devoted to children's education, but at the same time, it
may result in additional income, reduce fertility and create a sense of self-efficacy that may
increase overall investments in children (Desai and Jain, 1994). For example, it has been
suggested that income in the hands of women is more likely to result in child-focused
investments than income in the hands of men (Dwyer and Bruce, 1988; Engle, 1993).
However, the stress or fatigue associated with balancing employment, childcare, and
household chores may negatively affect women's interactions with children. While maternal
Author Manuscript

employment may shape direct investments in children, the role of these investments in
shaping children's education and learning outcomes depends on a variety of contexts,
including inputs from school and childcare arrangements. Finally, in the context of
developing countries, women's employment takes many forms, including work in the
informal sector and a combination of a range of income-generating activities, including
some that are compatible with childcare and others that are not (Lloyd, 1991; Mason and
Palan, 1981).

How the positive and negative influences of maternal employment are balanced remains an
empirical question. Tracing these diverse influences is particularly important in rapidly
developing economies such as India. Using data from the India Human Development Survey
(IHDS) of 2004-05, we investigate the link between maternal employment, mothers'
Author Manuscript

contributions to children's school activities, and children's cognitive development for school-
going children aged 8 to 11 years. This is a critical period for the development of cognitive
ability and achievement, even more so than early adolescence (Guo, 1998; Boardman et al.,
2002). The intellectual and academic development of children at this age has strong
implications for future performance and is closely associated with socioeconomic disparities
later in life.

The main contributions of this paper include the development of a work pattern typology
that goes beyond the standard measures of employment (for example, working versus not
working or occupational classification) to highlight the diversity of women's experiences in
income-generating activities in a changing economy, particularly within the agricultural
sector. We use rich survey data containing direct assessments of child outcomes and specific
Author Manuscript

parental inputs into children's education (for example, involvement with school activities) to
trace the mechanisms linking maternal employment with arithmetic achievement. While the
effect of social capital at home and school has been examined in Western literature (see
Dufur et al., 2013), such analysis is rare for developing countries.

Specifically, we ask the following research questions: 1) Does maternal employment

negatively affect children's cognitive achievement? 2) If so, does this reduction depend on
the type of work mothers do? Is some type of work less compatible with child-specific

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 3

investments than others? 3) Does maternal education moderate the impact of employment on
Author Manuscript

cognitive achievement? 4) Do women's work patterns affect their investment in children in

terms of the time and resources devoted to children's education? To what extent does this
reduction (if any), translate into child outcomes? Our work stands to shed light on the
intricate process of how maternal employment influences childhood development. We expect
that the effect of maternal employment on children's development outcome is not
unidimensional, and the answer to the question cannot be a simple yes or no. Both positive
and negative effects of maternal employment are theoretically plausible, but the context and
nature of work is of primary importance. Particularly in the context of a developing country,
where the type of employment is diverse and non-standard type of work is pervasive, we
believe it is essential to properly characterize the nature of maternal employment and how its
effect may manifest through the time and resources devoted to children's educational
process.
Author Manuscript

Maternal Employment and Child Outcomes—While assessing the relationship

between maternal employment and child outcomes, it is important to focus on the following
four issues: (1) nature of work, (2) time investments in children, (3) financial investments in
children, and (4) role of education. Each has received considerable attention in the global
literature. In the following sections, we explore how India's unique social and cultural
conditions shape these four issues.

Nature of Work
Literature in the Western context draws heavily from developmental psychology and
sociology, highlighting the importance of the mother–child relationship and the mother's
parenting style in influencing attachment, work–family conflict and role incompatibility, as
Author Manuscript

well as the role of economic and social contexts (see review by Brooks-Gunn et al., 2010;
Lucas-Thompson et al., 2010).

Empirical literature on maternal employment and child well-being is extensive but generally
inconclusive, and the question of whether maternal employment has detrimental, beneficial,
or no effect seems to hinge on a range of factors, including, to name only a few: the
developmental context; children's life stage and measurement of child outcomes; the
definition and timing of maternal employment; parenting styles; the economic well-being of
the family; the quality of childcare arrangements; other family/household characteristics
(Brooks-Gunn et al., 2010; Johnson et al., 2012; Zick et al., 2001; Beyer, 1995; Goldberg et
al., 2008; Lucas-Thompson et al., 2010).

Among these constellations of factors and mechanisms, how mothers' income-generating

Author Manuscript

work is characterized is of tremendous importance in developing country settings. Although

considerable research has been conducted on maternal employment and its impact on
cognitive outcomes (e.g. Vandell and Ramanan, 1992; Brooks-Gunn et al., 2002, 2010;
Lucas-Thompson et al., 2010; Johnson et al., 2012) less attention has been paid to how
maternal employment is measured. Standard measures of work, for example, working versus
not working, full-time versus part-time work, formal versus informal work, or standard
occupational classifications, often fail to accurately assess mothers' workloads. In rural, less

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 4

industrialized settings, a large majority of women are engaged in income-earning activities

Author Manuscript

outside the formal sector and often combine multiple activities. Without taking into account
these multiple activities, it is difficult to gauge the time demands on mothers and the
subsequent implications for children's development.

Previous research has either treated maternal employment as a dichotomous variable or has
captured the intensity of work in terms of the hours spent on the job, with the common
distinctions being between working full-time and part-time (Brooks-Gunn et al., 2002;
Waldfogel et al., 2002; Cooksey et al., 2009). Certain researchers have looked beyond these
categories of maternal employment to focus on the nature of employment, such as its
complexity (Parcel and Menaghan, 1994; Joshi et al., 2009). In the developing regions of the
world, there is an increased emphasis on capturing the complexity of women's employment
by distinguishing between formal and informal sector work (Lloyd, 1991; Desai and Jain,
1994; Donahoe, 1999). Short et al. (2002) have further nuanced the understanding of how
Author Manuscript

maternal employment influences child outcomes by demonstrating how wage versus non-
wage work influences the time spent on childcare. Few researchers have considered non-
wage workers, but Short et al. (2002) demonstrate the importance of recognizing multiple
economic activities and their consequences for both work–childcare compatibility and work
intensity in China. In India, as in China, the majority of women live in rural areas and
engage in economic activities outside the wage and salary sector, such as agricultural work
on family-owned farms or sideline activities such as animal care or running a family
business. With the rising rates of non-farm employment for men (Coppard, 2001), women
often end up picking up the slack - working on both family farms and as farm laborers.

Time Investments in Children

Because of the zero sum nature of time use, it is often assumed that a mother's employment
Author Manuscript

leads to some reduction in the time she will spend with her children, but this view has been
heavily contested. For instance, research in the US often suggests that a mother's
employment does not result in any decreased time with her children; instead, it leads to a
rearrangement of her schedule that affects only her leisure time (Bianchi, 2000; Sayer et al.,
2004). Other research suggests that mothers may compensate for the possible negative
effects of employment by increasing their level of shared activities or improving the quality
of the time they spend with their children (Moorehouse, 1991; Bryant and Zick, 1996).

It is possible, however, that the stress or fatigue associated with employment may affect
women's interactions with children negatively. Arguably, maternal employment outside the
home weakens the family's social capital, which includes the frequency of interactions and
the strength of bonds developed between parents and children (Coleman, 1988). Mothers'
Author Manuscript

labor force participation is thought to contribute to a reduction in time spent with children,
limiting the role of the family in the socialization process and therefore potentially leading
to behavior problems or negative childhood outcomes (Parcel and Menaghan, 1993).

Research on developed countries tends to focus on maternal time constraints in the context
of alternative childcare and often argues that high quality childcare arrangements may make
up for some of the negative effects of maternal absence. In this paper, we do not focus on the
nature of childcare but on direct investments in children's education. Our exclusion of

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 5

childcare reflects the unique situation in India. India has few organized childcare programs,
Author Manuscript

particularly for school aged children. Although there are some crèche facilities in
metropolitan cities, most are limited to infants and pre-school age children. Most of the
childcare is typically provided by the extended family or older siblings with self-care being
common for primary school age children. In our sample, almost 43 per cent of the
households with employed mothers are living in joint families, that is, residing with the
extended family, which frequently includes paternal grandparents and other relatives. In
nuclear households, older brothers and sisters may act as caregivers, and families frequently
live adjacent to other family members who may provide childcare.

However, the Indian school system has increasingly begun to demand greater parental
involvement in children's education. Children, even those in kindergarten and grade 1,
receive daily homework, i.e., school-curriculum based activities to be undertaken at home,
and these require parental supervision. Recent legislation has made it mandatory for
Author Manuscript

government schools to organize regular Parent Teacher Association meetings, and private
schools have followed suit. Thus, in families where mothers face competing demands on
their time, time investments in children may decline, with a negative impact on children's
learning outcomes.

Financial Investments in Children

Women's employment raises a family's income, thus increasing the financial resources
available to children. Maternal employment may also increase the share of resources devoted
to children. Research on intra-household bargaining notes that men and women frequently
have different expenditure preferences, and women's participation in income-generating
activities allows them to increase child-specific investments (Agarwal, 1997; Dwyer and
Bruce, 1988). Hence, maternal employment could bring additional resources to the family,
Author Manuscript

enhance a mother's autonomy and sense of well-being, increase the children's resources, and,
consequently, benefit the children (Beyer, 1995; Vandell and Ramanan, 1992).

One of the ways greater access to financial resources may have a positive impact on children
may be via enrollment in a private school. While free universal schooling up to grade 8 is
available to all Indian children, studies document that the quality of school instruction is
often poor, characterized by teacher absenteeism (Chaudhury et al., 2006), apathy (The
Probe Team, 1999), and discrimination (Deshpande, 2011). Given this situation, more
parents are seeking admission to private schools for their children, and private school
enrollment in rural India had jumped from 10 per cent in 1993-94 (Shariff, 1999) to over 30
per cent by 2014 (ASER Centre, 2015). In urban areas, more than half the children attend
private schools (Desai et al., 2009b). Since children in private schools exhibit slightly better
Author Manuscript

educational performance than those in government schools (Desai et al., 2009b;

Muralidharan and Kremer, 2008), maternal employment may improve children's educational
outcomes by facilitating their enrollment in private schools.

In addition, private coaching is viewed as providing children with an extra advantage. It

refers to after-school private lessons where children are taught the school curriculum and
helped with specific problems or subjects. It is increasingly seen as an integral part of
overall education of the child but is within the reach of a select few – 19 per cent of our

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 6

sample. However, there is considerable variation in the quality and kind of help students
Author Manuscript

typically receive via private coaching. Some receive tutoring the year round; others are
tutored for a short period, such as before exams. Some receive private coaching from school
teachers for additional payment, while others attend coaching classes. Students spend more
time studying and pursuing curriculum based activities at home when they have private
tutoring lessons (Desai et al., 2010).

Role of Education
The role of women's education in shaping their labor force participation is especially
noteworthy. Educational achievement among women in India is quite low, with the 2011
census estimating female literacy at 65.5% (Chandramouli and Registrar General, 2011).
Even though Indian women have made significant gains in education over time, these gains
have not translated into associated gains in employment. In fact, education has a U-shaped
Author Manuscript

relationship with women's labor force participation, with the least educated and the most
educated women being most likely to work (Klasen and Pieters, 2012). Srivastava and
Srivastava (2010) also show that female labor force participation is higher for illiterate
women than for those with higher levels of education; this trend reverses for women with
technical or vocational education or college education in India. Klasen and Pieters' (2012)
analysis of urban female labour force participation in India finds that below secondary
education, female employment is distributed widely across sectors and types of employment
(paid and self-employment) but is driven by economic necessity; the large majority of highly
educated (secondary and above) women are more likely driven by pull factors and are
employed in higher quality, skilled, service sector jobs. Given this, our study is likely to
capture employment at the two ends of the education distribution.

Other Determinants of Women's Employment in India

Author Manuscript

The issue of whether women participate in the workforce and the kind of work they do is
closely linked to their education, household conditions, and regional factors that shape the
opportunity and availability of work. These factors also have an independent effect on child
outcomes, such as education, making it important to control for certain factors when
examining the relationship between maternal employment and child outcomes. A number of
background factors affecting women's labor force participation in India deserve particular
attention (Klasen and Pieters, 2012). For example, starting with the classic work by Jacob
Mincer (Mincer, 1962), a great deal of literature has documented the inverse relationship
between a husband's income and a wife's labor force participation. In addition, given the
large extended families common in India, the incomes of other household members may
affect women's labor force participation (Desai et al., 2010; Jain and Banerjee, 1985;
Author Manuscript

Sharma, 1980). Finally, as we discuss in a later section, caste, religion, and place of
residence also shape women's employment and children's educational outcomes.

We control for these observable factors in our analyses. Even though we account for
selection into employment by controlling for a wide range of controls, there would be some
unobserved characteristics, such as ability or motivation that we cannot control; as these are
likely to influence both women's employment as well as children's cognitive scores. Hence,
there is a need for caution when interpreting our results.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 7

Research Significance and Hypotheses—We make significant contributions to the

Author Manuscript

study of maternal employment and child development on several fronts. First, instead of
using the conventional classification of work (work versus non-work, or wage work versus
non-wage work), we use a work typology that is more appropriate for the development
context of India. Our focus on different types of work is particularly important because
different activities have different claims on a woman's time. For example, a woman with a
salaried job is most likely to work away from home and may have little flexibility in
scheduling. At the same time, her schedule is often fixed and predictable, making it easy to
plan activities around work. For someone who engages in agricultural wage work or animal
husbandry, the location of work is typically close to home and probably offers flexibility in
scheduling. However, the work hours can be long and physically demanding. For women
who combine several types of work activities, the work hours can be extended, and the
fatigue and stress induced by juggling these multiple activities can take a toll on their time
Author Manuscript

and energy, leaving little room for involvement in academic activities with children. Having
several jobs could also indicate a precarious economic situation which compels women to
engage in different types of employment to make ends meet. The negative consequences of
maternal employment are likely to be apparent in such situations. Based on the above
argument, we formulate the following:

H1a: Maternal employment, especially of mothers with employment outside their homes and
those with multiple jobs, will disadvantage children in math and reading skills as compared
to those children with stay-at-home mothers.

H1b: Self-employment will offer mothers greater flexibility in scheduling activities, and
their children will not be more disadvantaged than those with stay-at-home mothers.
Author Manuscript

Second, instead of testing all possible relevant mechanisms whereby maternal employment
can affect child development, we focus on a dimension that is often overlooked in the
literature (mostly due to the unavailability of data), that is, a mother's actual involvement
with school-related work, in terms of supervision of homework and involvement in parent-
teacher associations (PTA). We hypothesize that this involvement is an important pathway
for mothers to influence their children's cognitive achievements. Conversely, then, we also
hypothesize:

H2: Time constraints associated with different forms of maternal employment will
lead to lower participation in children's academic activities such as supervision of
homework and involvement in parent teacher meetings.

Third, we postulate that women's participation in the workforce is associated with increased
financial resources being made available to children, thereby increasing their educational
Author Manuscript

achievements.

H3: Maternal employment will be associated with increased financial resources

available to children, thereby increasing their educational achievements.

Fourth, secondary and higher education is associated with skilled and more remunerative
jobs, whereas jobs at the lower end of the educational distribution are primarily driven by
economic necessity. Employment at the lower end may not be beneficial to children as it

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 8

takes away parental time and offers few economic resources for investment in schooling. At
Author Manuscript

the upper end of the educational distribution, jobs may offer greater flexibility and higher
remuneration, enabling mothers to invest in children's academic advancement. For mothers
with higher education, maternal employment would enable better cognitive outcomes,
compared to employed mothers with lower education:

H4: The impact of maternal employment on children's cognitive scores will be

moderated by maternal education.

2. Data and Methods

2.1 Data and Sample
The IHDS (2004-05) is a nationally representative survey of 41,554 urban and rural
households. It covers all the states and union territories of India, with the exception of
Author Manuscript

Andaman and Nicobar and Lakshadweep Islands (Desai et al., 2008). The households are
spread across 33 states and union territories, 384 districts, 1503 villages, and 971 urban
blocks located in 276 towns and cities. This survey was jointly organized by researchers
from the University of Maryland and the National Council of Applied Economic Research,
New Delhi, and funded by the National Institute of Child Health and Human Development.
The survey was translated into 13 Indian languages and administered by pairs of local
interviewers; women respondents were interviewed by women interviewers whenever
possible. See Desai et al. (2010, pp. 214–216) for detailed information on sampling.

The survey was specially designed to capture the diversity of women's work patterns, and
the interviewers were trained to ensure that women's work in household enterprises and
family farms was appropriately captured, along with their labor market work. Most
importantly, the IHDS incorporates the direct measurement of reading, writing, and
Author Manuscript

arithmetic skills of children aged 8–11 years, along with parental inputs into children's
education. Conducting educational assessments in developing countries—particularly India
—is difficult for a variety of reasons: children's abilities vary tremendously and an
instrument must capture children at both ends of the distribution; tests must be translated
into many different languages with similar difficulty levels; the instrument must be simple
and intuitive so that interviewers can administer it easily and it would not frighten children
who are not used to standardized tests. The IHDS research team worked with Pratham, a
non-governmental organization that has worked in the field of elementary education for
several years. They have developed simple assessment tools to measure the effectiveness of
their training programs and administered these tools to over 250,000 children in a
nationwide survey reported in the Annual Status of Education Report 2005 (Pratham, 2005).
The tests were developed in 12 Indian languages, as well as in English, and children were
Author Manuscript

asked to take the test in whichever language they found the most comfortable. The test data
are unique, particularly because they are combined with a wealth of household
characteristics.

We focus on arithmetic and reading skills attained by children between the ages of 8 and 11
years. We do not use the writing measure; it was recorded as a dichotomous measure and did

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 9

not capture variations in performance to the same extent as other tests. More information on
Author Manuscript

arithmetic and reading skills is provided in the following section.

Our working sample includes school-going children between the ages of 8 and 11 years,
who were administered the cognitive tests and where information about their parents and
their household is available. The sample included a total 10,043 children belonging to 8,732
households. We adjusted for clustering at the household level by using the cluster command
in STATA.

From an original sample of 41,554 households, 17,061 contained children ages 8-11. Of
these, 12,306 were administered the arithmetic test and 12,356 were administered the
reading test. The survey shows a substantial non-response on learning tests. Primary school
enrollment in India is almost universal; in this survey, about 8.5 per cent of the children aged
8 to 11 are currently out of school. An assessment of missing information of reading and
Author Manuscript

arithmetic scores reveals that 82 per cent of those not in school did not even attempt the test.
Children were required to be willing participants, and it is likely that out-of-school children
did not feel confident to take the test. Even when they did take the test, a majority could not
read numbers; for instance, only 20 per cent could identify numbers and less than 2 per cent
could do subtraction or division. For reading, the performance was slightly better but still
quite low; only 11.4 per cent of the out-of-school sample could read words; 6.8 per cent
could read paragraphs and another 6.5 per cent could read a short story. Home schooling is
rare in the Indian context, and it is likely that out-of-school children receive no form of
instruction. Those who were out-of-school but able to read could have been recent drop-outs.
Consequently, we restricted our analysis to the currently enrolled children, 91.5 per cent of
the total.
Author Manuscript

Since the household survey was the main focus of this study, the administration of the
reading and arithmetic skills was left to the end. We suspect that household fatigue as well as
interviewer fatigue played a small role in the missing data. Additionally, children were not
tested because the interviewers were explicitly instructed to obtain parental consent as well
as assent from children and were asked not to pressure the children who were reluctant. We
have carried out Heckman selection analysis to evaluate if selection is an issue and whether
this selection makes a substantive difference to our results. We ran two step Heckman
selection models on both arithmetic and reading scores, treating the outcome variables as a
continuous variable (See appendix for Heckman selection results). We used three
instruments for selection: rainy season, number of interviews carried out in a primary
sampling unit (PSU), and whether the child was enrolled in school. It is easier to interview
children during peak rainy seasons (July and August in India) as they are less likely to be in
Author Manuscript

school as compared to the rest of the year when season does not obstruct travel to schools.
The number of interviews within a PSU should also be a good instrument as the survey
teams would be required to be in a village/town for much longer if they needed to interview
a large number of households. This longer time span would give them the opportunity to
make several visits to contact children and complete the cognitive tests.

Because our outcome variable is ordinal, we are unable to carry out Heckman selection
model on the original analysis1. By treating our dependent variables as continuous instead of

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 10

ordinal, we run the Heckman selection models, which show that the missing data is selective
Author Manuscript

but largely due to enrolment in schools. The substantive results for maternal employment
remain unchanged even if we take selection into account (tables I and II in the appendix).
The coefficients are not directly comparable but the significance tests yield the same result,
therefore our interpretation does not change.

After excluding children who were not enrolled in schools or any other academic institution
and who had missing data on dependent, independent, and control variables, the final
analytic sample comprised 10,043 children. Table III in the Appendix shows the steps taken
to arrive at the final sample.

The decline in sample size due to the missing data on the independent and control variables
is minimal (1.3 per cent). However, 7.65 per cent of the cases had missing values on
attendance in parent–teacher meetings. This is not surprising, as some schools in India,
Author Manuscript

especially in the rural areas, may not have parent teacher associations or parents may not be
familiar with their existence. In our sample, 90 per cent of the missing data on PTA
participation is for rural areas. We coded the missing values to represent 0 and included a
variable to indicate missing values on PTA attendance.

2.2 Dependent Variables–Arithmetic and Reading Skills

We measure arithmetic skills as a four-category ordinal variable: the child 1) has no
recognition of written numbers; 2) can read numbers; 3) can subtract a two-digit number
from another two-digit number; and 4) can correctly divide a three-digit number by a one-
digit number. We code it as 0, 1, 2, and 3, respectively, in the analysis, with a higher score
indicating higher arithmetic skill. Close to a quarter of the children can perform division,
while less than 20 per cent have no recognition of written numbers.
Author Manuscript

The reading skills are measured through five categories: the child 1) cannot read at all; 2)
can read letters; 3) can read words; 4) can read a short paragraph; and 5) can read a short
story. Since in this school-going sample, 80 per cent of the children tested had completed at
least Standard 1, and 60 per cent had completed Standard 2; therefore, they were generally
expected to be able read at least a simple paragraph with three sentences; 33.5 per cent can
read a one-page story, and 22.4 per cent can read simple two to three sentence paragraphs
but not a one-page story. However, 8 per cent cannot recognize letters and about 10 per cent
can recognize letters but not read words.

2.3 Typology of Work Patterns

The IHDS is one of the rare surveys in India to collect information on both income and
Author Manuscript

employment. The survey began by asking about different sources of household income. The
respondents were then questioned about which household members participated in each of
those income-generating activities over the last one year.

1Heckman models have normality assumptions that ordinal logit models are unable to meet. Using maximum-likelihood probit models
with sample selection entails that we treat our dependent variable as dichotomous which leads to loss of important information on
cognitive scores. Instead, we perform linear regression by treating our outcome variables as continuous instead of ordinal and compare
our results with estimates from the Heckman selection model.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 11

We broadly create four mutually exclusive categories of types of employment:

Author Manuscript

• Self-employed women, including women who exclusively work on their own

farms (4.6per cent), household business (2.74 per cent), or animal husbandry
(16.33 per cent);

• Women employed outside their homes, including women exclusively employed

insalaried work (3.55 per cent), non-farm wage work (2.58 per cent), and farm
wage work(6.54 per cent);

• Women holding multiple jobs, either self-employed or outside the home: two
jobs (20 percent) and three or more jobs (7.5 per cent) over the year preceding
the survey; and

• Those not employed in income-generating activities (36.2 per cent, residual)

Author Manuscript

This combination of information from different streams of activity presents a holistic picture
of the work undertaken by the women during the preceding year. It simultaneously captures
the nature and intensity of maternal work. The work schedules of women employed outside
their homes are likely to be less flexible than those of self-employed women, while the time
and investment in children may be seriously limited if women combine two or more jobs.

2.4 Mother's Involvement with School-related Activities

We include three variables that represent the pathways that potentially mediate the impact of
maternal employment on children's cognitive outcomes: whether women participate in
school PTAs, whether women supervise their children's homework, and whether children
attend private schools and/or private coaching lessons.
Author Manuscript

2.5 Control Variables

Women's participation in the workforce and the kind of work in which they engage is closely
linked to a number of factors that shape the opportunity and availability of work. Some of
these deserve particular attention in the Indian context, especially as they can have an
independent effect on children's cognitive scores.

Maternal Education—We include the education of the mother as a dummy category

indicating whether or not she has completed at least eight years of schooling (secondary
education). In our view, eight or more years of education may indicate qualification for
skilled jobs in a predominantly informal labor market. The education of the mother is of
interest, as she is in a position to influence the learning of her children by reviewing material
taught in school, helping prepare for exams, taking an interest in school activities, and
Author Manuscript

participating in parent–teacher meetings, among others. We also interact secondary

education with maternal employment categories to assess if the impact of employment is
moderated by education.

Housing—We include an index of housing amenities ranging from 0 to 6, including piped

indoor water, separate kitchen, flush toilet, wall type, floor type and roof type, as these
reflect a household's long-term economic well-being. We select this indicator because it is
less likely to be associated with a woman's earnings; she most likely married into this

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 12

particular dwelling. A more detailed household asset score might be associated with
Author Manuscript

maternal employment as a woman is likely to contribute to other aspects of the standard of

living.

Household Income Excluding Maternal Earning—We combine income from all

sources to develop a composite income assessment for the household. From the total family
income, we subtract the income of the mother to exclude the effect maternal earnings may
have on cognitive scores. When women were employed either on household farms,
household businesses or animal care at home, the data only provided earnings from that
source for the entire household. However, the survey has a comprehensive measurement of
the amount of time (the average number of hours each day and the number of days spent on
this work during the past year) spent on each income-earning activity by each individual in
the household. We used this information to determine the proportion of time spent by
women compared to the rest of the family and estimated their income using this information.
Author Manuscript

Caste and Religion—We include the caste and religion of the family as controls. India is
predominantly a Hindu nation, albeit with a substantial degree of religious diversity. In the
Hindu religion, Hindu society is classified into four varnas or castes: Brahmin (priest and
teacher), Kshatriya (ruler and warrior), Vaishya (trader), and Shudra (servant). The two
groups lowest in the social hierarchy, dalits and adivasis, are outside the caste system. The
dalits, originally called ‘untouchables’ are also referred to as the ‘Scheduled Castes’ (SCs);
the adivasis or tribals are called the ‘Scheduled Tribes’ (STs)2. The caste stratification
system has tended to dominate Indian society, with a substantial percentage of Muslims,
Christians, and Sikhs identifying with the caste groups defined by Hindu tradition.

Caste and religion shape social relations in India in myriad ways. Anthropological literature
Author Manuscript

has suggested that women's ‘immurement’ is the hallmark of higher caste status (Srinivas
1977). Immurement or seclusion may be practiced in diverse ways, including through
veiling or purdah, familial control of women's physical movement, even not allowing women
to go unescorted in public spaces (Desai et al., 2010). These constraints may reduce
women's ability to work in public spaces (such as construction work) or to travel to offices.
Muslim women are less likely to participate in the labor force than their Hindu sisters (Das,
2005; Desai and Andrist, 2010). Caste and religion also seem closely linked to children's
school performance (Desai et al., 2009a; Deshpande, 2011). Children from socially
disadvantaged groups such as the SCs and STs are less likely to complete schooling and
perform worse than their higher-caste counterparts on learning achievement tests (Desai and
Kulkarni, 2008). In contrast, Brahmins have a strong intergenerational advantage, reflected
in their higher academic achievement (e.g. Goyal, 1989).
Author Manuscript

We treated forward castes (22.3 per cent), i.e., higher castes excluding Brahmins, as the
reference category. The caste categories include Brahmins (5.5 per cent), other backward
castes (OBCs) (41.7 per cent), SCs (24 per cent), and STs (6.6 per cent). We treated Hindus
as the reference category (81.6 per cent) for religious groups; the other categories were

2‘Scheduled Castes’ and ‘Scheduled Tribes’ are referred to as such because a list of the castes and groups was drawn and enumerated
in two separate schedules of the Constitution of India.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 13

Muslims (13 per cent), Christians (1.8 per cent), Sikhs (1.7 per cent) and the rest (1.9 per
Author Manuscript

cent).

Child-level Controls—We include children's age and sex. We expect a child's age to be
positively associated with his or her cognitive scores, and since girls are disadvantaged in
India, we expect them to have lower scores (Barcellos et al., 2014). We also include the
number of years the child has completed in school.

Household Structure—We include information on the number of siblings under the age
of 5 years in the household, as having siblings suggests competition for parents' attention.
Additionally, we include the presence of teenagers at home (not necessarily the siblings of
the child but other relatives as well) who can be in a position to help with school work
and/or relieve the pressure on younger children to help with household chores. Girls may
also be called upon to help the mother in her domestic duties. Accordingly, we include
Author Manuscript

interaction terms of these household structure variables with the gender of the child to
determine whether girls were expected to shoulder an additional burden of care work.

The presence of grandparents had been included as a control as they may be considered
additional caregivers in the family. It was not significant in the analysis; perhaps because
grandparents may not be educated enough to help children with school work, among other
things. Accordingly, we did not include it in the final analysis.

Regional Controls—Regional diversity in India shapes the labor market opportunities for
women and the performance of school systems. Urban women are less likely to be employed
(National Sample Survey Office, 2013), but urban children have better educational
opportunities and get higher scores on educational tests (Desai et al., 2010). Rural areas
Author Manuscript

offer limited and variable access to educational institutions. Therefore, each model includes
controls for the state of residence, measured by a series of dummy variables as well as
whether the place is urban.

2.6 Methods
Since our outcome variables are ordinal (math and reading skills), we use ordinal logit
regression in our analysis. We start with a model that includes mothers' work patterns and
the control variables (Model 1)3. In Model 2, we add interaction terms between maternal
employment typologies and completion of secondary education. We then add household SES
variables, household income and quality of housing, as potential confounding variables
(Model 3). Next we add two pathway variables: mother's involvement with school-related
work, in terms of the supervision of homework and involvement in PTAs (Model 4). In
Author Manuscript

Model 5, we add enrollment in private school and private coaching. We expect the maternal
employment effect to be attenuated after the introduction of these intervening variables. The
results from the ordered logit models for arithmetic and reading scores are presented in

3Variance inflation factor (VIF) and condition number both reveal that there are no multicollinearity problems in our equation
specifications. Mean VIF of all the independent and control variables included in the analysis at 1.86 and condition index is 27.29.
Condition number was calculated using the command ‘coldiag’ in STATA. According to Belsley, Kuh, and Welsch (1980), a condition
number of 30 or higher is suggestive of multicollinearity.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 14

Tables III and IV respectively. We present the regression coefficients (log odds) in the table;
Author Manuscript

these can also be expressed as the odds ratio of the chance of attaining higher skill level.

3. Results
We start with a bivariate analysis of the mothers' work pattern categories, arithmetic and
reading scores, and their involvement in school-related activities and investments in private
schooling and coaching (Table II). Our findings show children of stay-at-home mothers
perform better on arithmetic and reading tests than children of employed women. Stay-at-
home mothers are also more involved in children's school activities and make greater
financial investments in education. As the t-test results show, the children of employed
women are significantly different from those of stay-at-home mothers.

Contrary to what was hypothesized, private school enrollment and additional coaching are
Author Manuscript

lower among employed mothers. Interestingly, women who are employed outside their
homes are generally more involved in children's school work than those who are self-
employed. However, they are still less likely to send their children to private school or to
find coaching for them than those who are self-employed. Further, women who hold
multiple jobs have the lowest level of PTA participation and spend the least time on home-
based school work. The stay-at-home mothers have the highest level of education, most
economic resources, spend the most amount of time on school-related activities, and are
most likely to send their children to private schools and coaching. Could the effect of
employment be confounded by education or economic resources?

Model 1 shows that compared to children with stay-at-home mothers, children whose
mothers have outside employment and those with multiple jobs have lower math scores
(Table III). Compared to children with stay-at-home mothers, children whose mothers have
Author Manuscript

outside employment have lower reading scores (Table IV). At the same time, children with
self-employed mothers do not suffer a disadvantage over those with stay-at-home mothers.
In Model 2, we explore if the association between women's employment and children's
cognitive scores varies by her education level. The children of women employed in jobs
outside the household or who hold multiple jobs continue to be at a disadvantage over those
with stay at home mothers only if the women have less than secondary schooling in both
reading and arithmetic skills. The children of mothers with higher education who are
employed outside the household have an advantage in reading and arithmetic skills. This
suggests that higher education enables mothers to get jobs that do not disadvantage their
children, possibly because of higher remuneration, job security or flexibility offered by these
jobs. However, it could also be indicative of the effort and resources that more highly
educated employed women can bring to bear on their children's academic progress.
Author Manuscript

Since maternal employment is closely associated with economic resources, we introduce two
family SES variables into the models: household income excluding mother's earnings and
the quality of their housing (Model 3). Our findings show the coefficients for maternal work
patterns are reduced in magnitude for reading and arithmetic scores, suggesting that part of
the maternal employment effect is associated with family resources. As expected, children
living in families with higher incomes and better housing have greater cognitive scores. The

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 15

disadvantage associated with holding multiple jobs in reading scores is explained by SES
Author Manuscript

variables, as it is no longer significant in the model. This indicates that combining multiple
jobs is a strategy adopted by households who are in financial distress. However, this
disadvantage continues to persist as far as arithmetic scores are concerned.

At the same time, the coefficients for mothers' work patterns remain significant for children
with mothers with outside employment for both higher and lower educated women. Children
of higher educated women employed in jobs outside the household exhibit a positive
association with reading and arithmetic scores. The relationship is reversed for children of
women with lower education who are employed outside their households.

Mediating Pathways
In Model 4, we introduce two pathway variables measuring mothers' involvement with
children's school work. The model shows that involvement with PTAs and helping children
Author Manuscript

with homework are both strongly associated with children's arithmetic and reading skills.
Helping with homework is associated with a 36 per cent increase in the odds of having better
math skills (log odds of 0.305). Parental involvement in PTAs is associated with a 45 per
cent increase in the odds of having better math skills (log odds of 0.369). Similarly, helping
with the children's homework is associated with a 32 per cent increase in the odds of having
better reading skills (log odds of 0.280), and involvement in PTAs increases the odds by 40
per cent (log odds of 0.340).

Adding these two variables explains away the effect of mothers' employment on reading
scores; suggesting the employment effect is mediated by the availability of mothers' time.
There is evidence to show that reading frequency is strongly affected by the availability of
reading material at home, which, in turn, is associated with reading ability (McQuillan and
Author Manuscript

Au, 2001). Our results also highlight the importance of supervised reading.

However, this is not true for arithmetic score where maternal employment outside the
household continues to be significant. The children of higher educated women who are
employed outside their homes show a positive association with math scores, but the children
of lower educated women working outside their homes continue to be disadvantaged in math
skills. The association of math scores with multiple jobs is no longer significant, suggesting
that lower educated women with multiple jobs may face severe time constraints, limiting
their involvement with academic activities.

In Model 5, we see that private school enrollment and private tutoring affects children's
arithmetic skills positively and seems to explain the negative association between math
scores and outside employment among lower educated women. Even the positive effect for
Author Manuscript

higher educated women with outside employment is reduced in magnitude, suggesting the
positive role of private schooling and coaching in children's skill development. Employed
mothers seem to be directing greater financial resources towards quality of education as this
is captured by private education and additional coaching.

Having mothers who have completed more than eight years of schooling is also associated
with an advantage in the final model, suggesting the involvement of these mothers in

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 16

academic activities that go beyond assistance with schoolwork, involvement in PTAs, or the
Author Manuscript

use of financial resources to promote educational activities. More educated mothers possibly
bestow their children a life course advantage by having greater ‘educational’ interactions
with their children in the pre-school years and providing assistance not captured by our
models.

Among the control variables, child's age has a positive association with math and reading
skills. Girls have lower scores than boys. The presence of younger siblings (under the age of
5) at home is negatively associated with arithmetic and reading scores of children. The
presence of teenagers at home (not necessarily the siblings of the child but other relatives in
the age group) is also negatively associated with arithmetic and reading scores, indicating
that the higher the number of children in the household, the lower the performance on
cognitive tests. However, the presence of teenagers at home is beneficial for girls. Girls may
be more likely to seek academic help from their older and more educated siblings or other
Author Manuscript

relatives in the household. It is also possible that the presence of teenagers relieves the
pressure on younger girls, between the ages of 8 and 11 years, to perform household chores.
Children from disadvantaged castes, scheduled castes and tribes, perform poorly on the
cognitive tests as compared to children belonging to forward castes. Muslims have poorer
scores as compared to Hindus.

4. Discussion and Conclusion

Much of the research on the relationship between maternal employment and child outcomes
in the United States and Europe has focused on whether or not the mother works, the age of
the child when she began working, and whether she works full-time or part-time. A study on
India allows us to contribute to previous work by adding a focus on the type of work
Author Manuscript

performed. Employment in a transitional economy typically features a range of activities,

including office work, agricultural or non-agricultural daily labor, work on family farm, and
work in family business. Several studies on the conflict between women's work and fertility
in developing countries note that when women have greater control over work settings (for
example, on a family farm or in a family business), the conflict between work and child
rearing is less than it is for women who are wage workers in the formal sector (Ho, 1979;
Lloyd, 1991; Mason and Palan, 1981). Moreover, it has recently been recognized that in
developing economies, many women occupy multiple locations in the employment
landscape, often combining paid work with home-based employment (Short et al., 2002).
These women may be particularly vulnerable to severe time pressures as they try to juggle
multiple responsibilities.

Research on women's employment in developing countries and its implications for childcare
Author Manuscript

has alluded to challenges in this sphere and to potential implications for children but the lack
of data makes it difficult to examine the consequences for child outcomes. Our study fills
this niche by focusing on employment characteristics in a transitional economy, India, and
combining this with a focus on children's cognitive outcomes.

The analytical results clearly demonstrate that the relationship between maternal
employment and children's math and reading skills depends on the nature of mothers' work.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 17

The great majority of working women in India are engaged in the informal sector and are
Author Manuscript

often undercounted, making it essential to capture the complexity of their economic

activities. While working mothers in India are less likely to be involved with school-related
activities, and their children are disadvantaged in math and reading skills, this simple
dichotomy does not capture the diversity of maternal employment experiences. The children
of women who are self-employed and work on their own farms, in household businesses, or
in sideline activities such as animal care, perform at the same level as those of stay-at-home
mothers. At the same time, the children of women who work outside their homes or who
hold more than one job show deficits in math and reading skills.

In an attempt to capture the diverse landscape of India, we segregate this relationship by the
education status of the mother. The results are starkly different for employed women at
higher levels of education, suggesting that employed women, who are presumably in better
paying jobs with greater benefits and flexibility, are able to give their children an advantage
Author Manuscript

in both reading and arithmetic skills. In other words, how employment impacts children's
development is India is complex – it is hard to predict by sector data alone how employment
might impact a child's cognitive scores, because jobs within a sector can be varied.

We parse the maternal employment effect by maternal education (with higher education
defined as more than eight years of education) with the underlying assumption that higher
educated women are more likely to have access to better jobs. By introducing interaction
terms between education and employment typologies, we find that higher educated women
who are working in jobs outside their household positively affect their children's outcomes.
However, employment outside the household is a heterogeneous category, which includes
salaried work and wage work in agriculture and non-agricultural sectors. We even tried to
assess if one of these sectors was driving the results by including separate variables for
Author Manuscript

employment in each sector but we did not find a significant result. Arguably, categorization
by sector cannot capture the varied working conditions of women in India. It is likely that
higher educated women within each sector of employment have access to better, more
flexible and higher paying jobs.

Overall, our analytical results suggest that maternal employment influences child
development through various mechanisms. First, our findings are in line with the time
constraint perspective. It is not about any decrease in time, but about time specifically
related to school activities central for children's cognitive development. Our work typology
acknowledges the heterogeneity of women's employment in India. In the case of work that is
more flexible in nature, such as family farm work or animal care, it may not be so difficult to
combine work and child-rearing activities. In the case of wage or salaried employment, the
Author Manuscript

inflexible schedule and location of work may restrict the availability of time for mothers for
activities such as PTA meetings held in school during work hours. Further, for women who
combine multiple jobs, the long and extended hours of work could be too physically
demanding, resulting in less involvement in their children's school work. Finally, children of
educated women, especially those in engaged in employment outside the household, show
an advantage in their cognitive outcomes. The results suggest that these mothers attempt to
adjust their schedule to accommodate tasks such as attending PTA meetings and assisting
with homework. Research in the US suggests that working mothers try to spend quality time

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 18

with children by cutting back on leisure, sleep, and time spent on personal care (Bianchi,
Author Manuscript

2000). Our analysis identifies maternal involvement in school-related work as an important

mechanism that links maternal employment and child outcomes.

This analysis presents two distinct narratives – one is a story of time constraints and
disadvantages associated with employment at the lower end of the educational distribution
where the mother is more likely pushed into the labor force due to economic necessity and is
engaged in casual, low-skill work (Srivastava and Srivastava, 2010). Kabeer (2012)
discusses the phenomenon of ‘time poverty’ where poor women, especially in rural areas,
spend long hours in domestic chores such as cooking, collecting water and firewood. This
‘time poverty’ is exacerbated for working mothers. In such a scenario, not only does
employment take time away from childcare but the lack of resources leaves children bereft
of any other benefits.
Author Manuscript

At the other end of the education distribution, maternal employment is characterized by

increased investments in children's academic activities. The time constraints of well-
educated employed mothers do not interfere with academic activities, and their employment
permits greater investment in private schooling and additional coaching.

Second, economic resources clearly matter. Controlling for family economic resources
considerably reduces the negative associations of maternal employment. Research has
shown that scholarly culture at home, measured by the number of books available at home,
has a strong and positive influence on children's education (Evans et al., 2010). We also see
that well-to-do families substitute mothers' time with private classes and tutors, essentially a
trade-off for less time spent with children. These additional resources result in better
academic performance.
Author Manuscript

These results highlight the dilemma facing Indian public policy. With increasing crowding in
agriculture (Papola, 2012), salaried work remains the primary avenue of increasing women's
income but it also imposes great time constraints. Additionally, the juggling of multiple
roles, especially when the mother is employed outside the home, takes a toll on children's
development. These conflicts can be addressed in two ways. First, time demands on mothers
can be reduced by promoting time-saving technologies and services, such as increased
access to gas stoves to reduce time in firewood gathering and cooking, improved
transportation systems to reduce commuting time, and expanded reliable access to electricity
to allow evening supervision of children's homework. Second, demands for supervision of
schoolwork at home can be reduced and parent-teacher meetings can be scheduled for
evenings or weekends. There is a clear recognition that the Indian school system has failed
to provide high quality education to children, leaving parents and private tutors to pick up
Author Manuscript

the slack (Pritchett, 2013; Desai et al., 2009b). Improvements in the educational system may
reduce the reliance on parental input to improve children's educational outcomes.

The study is not without its limitations. First, we focused on school-going children aged 8 to
11 years. With the IHDS data, it is not possible to document the effect of maternal
employment in early childhood, often identified as a critical period. Second, our study is
cross-sectional, thus limiting our ability to establish time order and make causal inferences.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 19

We are thus cautious in our interpretations and explain the mechanisms as associational
Author Manuscript

rather than causal. Third, we use step-by-step regression models to examine the mediating
effects of homework supervision, PTA involvement, and private school enrollment, but we
have been unable to account for other relevant mechanisms, given the lack of data on them.
Fourth, while we have tried to control for many theoretically relevant variables that affect
both maternal employment and child outcomes, unobserved heterogeneity in terms of who
works and who does not may bias our results.

In sum, we caution against making easy assumptions about the implications of maternal
employment on children's cognitive development. In a developing economy like India,
where the nature of work is complex, it is important to consider the heterogeneity of work
experiences. Our study suggests investment in school-related activities as one important way
through which maternal employment may affect children's development outcomes. Future
research should investigate other mechanisms, such as parental social networks and
Author Manuscript

connections that may facilitate school family interactions, maternal work conditions, and the
role of other caregivers, as all of these could potentially shape children's cognitive outcomes.

Acknowledgments
An earlier version of this paper was presented at the 2013 Annual Meeting of the Population Association of
America at New Orleans, LA. We are grateful to Reeve Vanneman and Liana Sayer for their helpful comments. We
gratefully acknowledge support from the Eunice Kennedy Shriver National Center for Child Health and Human
Development grant R01HD041455 and R01HD046166, and R24-HD041041, Maryland Population Research
Center.

Appendix
Table I
Author Manuscript

Comparison of estimates (Heckman selection model, OLS and Ordinal logit) for arithmetic
scores.

Heckman Equation For comparison of estimates

Selection equation Main equation OLS analysis Ordinal Logit

Not missing on
Math Math score Math score Math score
Self-employed women (Ref:
0.096** (0.035) -0.034 (0.026) -0.009 (0.024) -0.014 (0.053)
Unemployed women)
Women employed outside their
0.114** (0.042) -0.147*** (0.032) -0.122*** (0.031) -0.245*** (0.067)
households
Multiple jobs 0.122*** (0.035) -0.068** (0.026) -0.034 (0.025) -0.072 (0.055)
Self-employed women*higher education -0.096 (0.080) 0.036 (0.060) 0.012 (0.057) 0.053 (0.126)
Author Manuscript

Women employed outside their

0.043 (0.097) 0.264*** (0.071) 0.279*** (0.067) 0.634*** (0.152)
households*higher education
Multiple jobs*higher education 0.185* (0.094) -0.003 (0.065) 0.032 (0.061) 0.065 (0.135)
Log of household income excluding
-0.006 (0.006) 0.017*** (0.005) 0.017*** (0.004) 0.039*** (0.010)
maternal earning
Housing Index 0.005 (0.008) 0.086*** (0.006) 0.090*** (0.005) 0.194*** (0.012)
Maternal Education (Secondary and
-0.009 (0.036) 0.191*** (0.027) 0.190*** (0.026) 0.420*** (0.057)
above)

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 20
Author Manuscript

Heckman Equation For comparison of estimates

Years of education -0.002 (0.009) 0.214*** (0.007) 0.224*** (0.006) 0.501*** (0.015)
Sex of the child: Female (Ref: Males) -0.015 (0.022) -0.126*** (0.016) -0.135*** (0.015) -0.293*** (0.034)
Age of the child -0.009 (0.012) 0.073*** (0.009) 0.059*** (0.009) 0.128*** (0.019)
Number of children under 5 0.027 (0.016) -0.061*** (0.012) -0.056*** (0.011) -0.121*** (0.025)
Number of teenagers -0.009 (0.012) -0.034*** (0.009) -0.036*** (0.009) -0.081*** (0.019)
Brahmin (Ref: Forward caste) -0.057 (0.055) 0.173*** (0.040) 0.163*** (0.038) 0.367*** (0.086)
Other backward classes -0.024 (0.031) -0.064** (0.023) -0.071** (0.022) -0.175*** (0.047)
Scheduled caste 0.011 (0.035) -0.251*** (0.026) -0.254*** (0.025) -0.564*** (0.055)
Scheduled tribe -0.073 (0.050) -0.219*** (0.039) -0.244*** (0.037) -0.559*** (0.082)
Muslim (Ref: Hindus) -0.129*** (0.036) -0.186*** (0.027) -0.222*** (0.026) -0.472*** (0.057)
Christian 0.141+ (0.081) 0.024 (0.062) 0.056 (0.059) 0.062 (0.130)
Author Manuscript

Sikh 0.091 (0.091) -0.029 (0.068) -0.005 (0.065) -0.062 (0.140)

Others -0.005 (0.088) 0.008 (0.064) 0.011 (0.061) 0.004 (0.135)
Urban area 0.105*** (0.031) 0.076*** (0.022) 0.093*** (0.021) 0.224*** (0.047)
Rainy season 0.121* (0.047)
Number of interviewers -0.001 (0.002)
Child in school 1.754*** (0.048)
Mills Lambda -0.556*** (0.051)
Constant -1.498*** (0.147) 0.510*** (0.107) 0.205* (0.099)
Constant cut1 1.001*** (0.218)
Constant cut2 2.985*** (0.219)
Constant cut3 4.597*** (0.221)

Observations 17,028 17,028 12,306 12,306

Author Manuscript

R-squared 0.319

Standard errors in parentheses p<0.001,

**
p<0.01,
*
p<0.05;
state controls not shown. Source: IHDS 2004-05.
Table II

Comparison of estimates (Heckman selection model, OLS and Ordinal logit) for reading
scores.

Heckman Equation Comparison of Estimates

Selection equation Main equation OLS Ordinal logit

Author Manuscript

Not missing on
Reading Reading score Reading score Reading score
Self-employed women (Ref:
0.095** (0.035) -0.044 (0.035) 0.001 (0.032) 0.020 (0.053)
Unemployed women)
Women employed outside their
0.128** (0.042) -0.194*** (0.043) -0.144*** (0.040) -0.205** (0.066)
households
Multiple jobs 0.128*** (0.035) -0.024 (0.036) 0.038 (0.033) 0.072 (0.054)
Self-employed women*higher education -0.077 (0.080) 0.151+ (0.081) 0.118 (0.074) 0.132 (0.130)

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 21
Author Manuscript

Heckman Equation Comparison of Estimates

Women employed outside their
0.026 (0.097) 0.337*** (0.096) 0.356*** (0.088) 0.620*** (0.160)
households*higher education
Multiple jobs*higher education 0.183+ (0.095) 0.057 (0.089) 0.117 (0.080) 0.184 (0.144)
Log of household income excluding
-0.006 (0.006) 0.025*** (0.006) 0.024*** (0.006) 0.041*** (0.009)
maternal earning
Housing Index 0.007 (0.008) 0.102*** (0.008) 0.111*** (0.007) 0.179*** (0.012)
Maternal Education (Secondary and
-0.008 (0.036) 0.231*** (0.037) 0.229*** (0.034) 0.465*** (0.059)
above)
Years of education 0.002 (0.009) 0.282*** (0.009) 0.301*** (0.008) 0.511*** (0.015)
Sex of the child: Female (Ref: Males) -0.017 (0.022) -0.084*** (0.022) -0.101*** (0.020) -0.154*** (0.034)
Age of the child -0.008 (0.012) 0.083*** (0.012) 0.060*** (0.011) 0.119*** (0.019)
Number of children under 5 0.029+ (0.017) -0.098*** (0.016) -0.088*** (0.015) -0.129*** (0.025)
Author Manuscript

Number of teenagers -0.007 (0.012) -0.052*** (0.013) -0.055*** (0.012) -0.091*** (0.019)
Brahmin (Ref: Forward caste) -0.049 (0.055) 0.200*** (0.054) 0.186*** (0.049) 0.439*** (0.090)
Other backward classes -0.014 (0.031) -0.055+ (0.031) -0.062* (0.028) -0.101* (0.048)
Scheduled caste 0.016 (0.035) -0.230*** (0.035) -0.232*** (0.033) -0.383*** (0.055)
Scheduled tribe -0.063 (0.050) -0.219*** (0.052) -0.256*** (0.048) -0.412*** (0.079)
Muslim (Ref: Hindus) -0.133*** (0.036) -0.144*** (0.037) -0.206*** (0.034) -0.309*** (0.056)
Christian 0.157+ (0.081) -0.053 (0.084) 0.009 (0.078) 0.074 (0.132)
Sikh 0.091 (0.091) -0.044 (0.092) -0.003 (0.085) 0.008 (0.150)
Others -0.008 (0.088) 0.078 (0.088) 0.082 (0.080) 0.179 (0.133)
Urban area 0.110*** (0.031) 0.070* (0.030) 0.103*** (0.028) 0.179*** (0.047)
Rainy season 0.115* (0.047)
Number of interviewers -0.001 (0.002)
Author Manuscript

Child in school 1.758*** (0.048)

Mills Lambda -0.963*** (0.066)
Constant -1.534*** (0.148) 1.462*** (0.145) 0.917*** (0.129)
Constant cut1 0.351 (0.217)
Constant cut2 1.572*** (0.216)
Constant cut3 2.786*** (0.217)
Constant cut4 3.961*** (0.218)

Observations 17,028 17,028 12,356 12,356

R-squared 0.296

Standard errors in parentheses p<0.001,

**
p<0.01,
*
p<0.05;
Author Manuscript

state controls not shown. Source: IHDS 2004-05

Table III
Final Sample Calculation

Total sample Dropping cases

Children in the ages of 8 to 11 in the sample 17,036
Sample with valid cases on the dependent variable 12,288 4748

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 22
Author Manuscript

Total sample Dropping cases

Eliminating children not in school 12,041 247
Eliminating cases who are not children of women who answered the empowerment 10,352 1,689
questionnaire which contained the question on help with homework
Eliminating the missing cases on homework assistance 10043 309
Final sample 10043

Source: India Human Development Survey (IHDS) 2004-05

References
Agarwal, Bina. “Bargaining” and Gender Relations: Within and Beyond the Household. Feminist
Economics. 1997; 3(1):1–51.
ASER Centre. ASER 2014: Annual Status of Education Report. ASER Centre; New Delhi: 2015.
Author Manuscript

Barcellos, Silvia H., Carvalho, Leandro S., Lleras-Muney, Adriana. Child gender and parental
investments in India: are boys and girls treated differently? American Economic Journal: Applied
Economics. 2014; 6(1):157–189. [PubMed: 24575163]
Beyer, Sylvia. Maternal Employment and Children' s Academic Achievement: Parenting Styles as
Mediating Variable. Developmental Review. 1995; 15(2):212–253.
Bianchi, Suzanne M. Maternal Employment and time with children: Dramatic change or surprising
continuity? Demography. 2000; 37(4):401–414. [PubMed: 11086567]
Boardman, Jason D., Powers, Daniel A., Padilla, Yolanda C., Hummer, Robert A. Low Birth Weight,
Social Factors, and Developmental Outcomes among Children in the United States. Demography.
2002; 39(2):353–368. [PubMed: 12048956]
Brooks-Gunn, Jeanne, Han, WenJui, Waldfogel, Jane. Maternal Employment and Child Cognitive
Outcomes in the First Three Years of Life: The NICHD Study of Early Child Care. Child
Development. 2002; 73(4):1052–1072. [PubMed: 12146733]
Brooks-Gunn, Jeanne, Han, WenJui, Waldfogel, Jane. First-Year Maternal Employment and Child
Development in the First Seven Years. Monographs of the Society for Research in Child
Author Manuscript

Development. 2010; 75(2):7–9. [PubMed: 25152543]

Belsley, David A., Kuh, Edwin, Welsch, Roy E. Regression diagnostics: identifying influential data
and sources of collinearity. John Wiley & Sons; 1980.
Bryant, W Keith, Zick, Cathleen D. An Examination of Parent–Child Shared Time. Journal of
Marriage and the Family. 1996; 58(1):227–237.
Chandramouli, C. Registrar General. Provisional Population Totals. Government of India; New Delhi:
2011. Census of India 2011.
Chaudhury, Nazmul, Hammer, Jeffrey, Kremer, Michael, Muralidharan, Karthik, Rogers, F Halsey.
Missing in Action: Teacher and Health Worker Absence in Developing Countries. Journal of
Economic Perspectives. 2006; 20(1):91–116. [PubMed: 17162836]
Coleman, James S. Social Capital in the Creation of Human Capital. American Journal of Sociology.
1988; 94:S95–S120.
Cooksey, Elizabeth, Joshi, Heather, Verropoulou, Georgia. Does Mothers' Employment Affect
Children's Development? Evidence from the Children of the British 1970 Birth Cohort and the
Author Manuscript

American NLSY79. Longitudinal and Life Course Studies. 2009; 1(1):95–115.

Coppard, Daniel. NRI Report No 2662. Chatham, UK: Natural Resources Institute; 2001. The Rural
non-farm Economy in India: A Review of the Literature. Available at: http://r4d.dfid.gov.uk/
Output/189327/Default.aspx [accessed 2 December 2015]
Das, Maitreyi B. Muslim women's low labour force participation in India: Some structural
explanations. In: Hasan, Zoya, Menon, Ritu, editors. A Minority: Essays on Muslim Women in
India. Oxford University Press; New Delhi: 2005. p. 189-221.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 23

Desai, Sonalde, Jain, Devaki. Maternal Employment and Changes in Family Dynamics: The Social
Context of Women's Work in Rural South India. Population and Development Review. 1994;
Author Manuscript

20(1):115–136.
Desai, Sonalde, Dubey, Amaresh, Joshi, BL., Sen, Mitali, Shariff, Abusaleh, Vanneman, Reeve.
ICPSR22626ev1 University of Maryland and National Council of Applied Economic Research,
New Delhi [producers], 2007. Ann Arbor, MI: Inter-university Consortium for Political and Social
Research. [distributor]; 2008. India Human Development Survey (IHDS) [Computer file].
2008-07-30
Desai, Sonalde, Kulkarni, Veena. Changing Educational Inequalities in India in the Context of
Affirmative Action. Demography. 2008; 45(2):245–270. [PubMed: 18613480]
Desai, Sonalde, Adams, Cecily D., Dubey, Amaresh. Segmented Schooling: Inequalities in Primary
Education. In: Thorat, Sukhadeo, Newman, Katherine S., editors. Blocked by Caste:
Discrimination and Social Exclusion in Modern India. Oxford University Press; New Delhi:
2009a. p. 230-52.
Desai, Sonalde, Dubey, Amaresh, Vanneman, Reeve, Banerji, Rukmini. Private Schooling in India: A
New Educational Landscape. In: Bery, SumanBossworth, Barry, Panagariya, Arvind, editors. India
Author Manuscript

Policy Forum 2008-09. Vol. 5. Sage Publications; New Delhi: 2009b. p. 1-38.
Desai, Sonalde, Dubey, Amaresh, Joshi, BL., Sen, Mitali, Shariff, Abusaleh, Vanneman, Reeve.
Human Development in India: Challenges for a Society in Transition. Oxford University Press;
New Delhi: 2010.
Desai, Sonalde, Andrist, Lester. Gender scripts and age at marriage in India. Demography. 2010; 47(3):
667–687. [PubMed: 20879683]
Deshpande, Ashwini. The grammar of caste: economic discrimination in contemporary India. Oxford
University Press; New Delhi: 2011.
Donahoe, Debra A. Measuring Women's Work in Developing Countries. Population and Development
Review. 1999; 25(3):543–576.
Dufur, Mikaela J., Parcel, Toby L., Troutman, Kelly P. Does Capital at Home Matter More than Capital
at School? Social Capital Effects on Academic Achievement. Research in Social Stratification and
Mobility. 2013; 31(2013):1–21.
Dwyer, Daisy, Bruce, Judith. A Home Divided: Women and Income in the Third World. Stanford
University Press; Stanford, CA: 1988.
Author Manuscript

Engle, Patrice L. Influences of Mothers' and Fathers' Income on Children's Nutritional Status in
Guatemala. Social Science & Medicine. 1993; 37(11):1303–1312. [PubMed: 8284697]
Evans, Mariah DR., Kelley, Jonathan, Sikora, Joanna, Treiman, Donald J. Family Scholarly Culture
and Educational Success: Books and Schooling in 27 Nations. Research in Social Stratification
and Mobility. 2010; 28(2):171–197.
Goldberg, Wendy A., Prause, JoAnn, Lucas-Thompson, Rachel, Himsel, Amy. Maternal Employment
and Children's Achievement in Context: A Meta-analysis of Four Decades of Research.
Psychological Bulletin. 2008; 134(1):77. [PubMed: 18193996]
Goyal, Santosh. Social Background of Officers in the Indian Administrative Service. In: Frankel,
Francine R., Rao, Madhugiri SA., editors. Dominance and State Power in Modern India: decline of
a social order. Vol. 2. Oxford University Press; Delhi: 1989. p. 425-33.
Guo, Guang. The Timing of the Influences of Cumulative Poverty on Children's Cognitive Ability and
Achievement. Social Forces. 1998; 77(1):257–287.
Ho, Teresa J. Time Costs of Child Rearing in the Rural Philippines. Population and Development
Author Manuscript

Review. 1979; 5(4):643–62.

Jain, Devaki, Banerjee, Nirmala, editors. Tyranny of the Household: Investigative Essays on Women's
Work. Shakti; New Delhi: 1985.
Johnson, Rucker C., Kalil, Ariel, Dunifon, Rachel E. Employment patterns of less-skilled workers:
Links to children's behavior and academic progress. Demography. 2012; 49(2):747–772. [PubMed:
22246798]
Joshi, Heather, Cooksey, Elizabeth, Verropoulou, Georgia. Working Paper 2009/1. London: Centre for
Longitudinal Studies Institute of Education, University of London; 2009. Combining Childrearing
with Work: Do Maternal Employment Experiences Compromise Child Development.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 24

Kabeer, Naila. Women's economic empowerment and inclusive growth: labour markets and enterprise
development. International Development Research Centre. 2012; 44(10):1–70.
Author Manuscript

Klasen, Stephan, Pieters, Janneke. Vol IZA Discussion Paper 6395. Bonn, Germany: IZA; 2012. Push
or Pull? Drivers of Female Labor Force Participation During India's Economic Boom. Available at:
https://ssrn.com/abstract=2019447 [accessed 10 June 2017]
Lloyd, Cynthia B. The Contribution of the World Fertility Surveys to an Understanding of the
Relationship between Women's Work and Fertility. Studies in Family Planning. 1991; 22(3):144–
161. [PubMed: 1949098]
Lucas-Thompson, Rachel G., Goldberg, Wendy A., Prause, JoAnn. Maternal Work Early in the Lives
of Children and Its Distal Associations with Achievement and Behavior Problems: A Meta-
Analysis. Psychological Bulletin. 2010; 136(6):915. [PubMed: 20919797]
Mason, Karen O., Palan, VT. Female Employment and Fertility in Peninsular Malaysia: The Maternal
Role Incompatibility Hypothesis Reconsidered. Demography. 1981; 18(4):549–575. [PubMed:
7308536]
McQuillan, Jeff, Au, Julie. The Effect of Print Access on Reading Frequency. Reading Psychology.
2001; 22(3):225–248.
Author Manuscript

Mincer, Jacob. Labor Force Participation of Married Women: A Study of Labor Supply. In: Lewis, H
Gregg, editor. Aspects of Labor Economics. Princeton University Press; Princeton: 1962. p. 63-97.
Moorehouse, Martha J. Linking maternal employment patterns to mother-child activities and children's
school competence. Developmental Psychology. 1991; 27(2):295.
Muralidharan, Karthik, Kremer, Michael. Public and Private Schools in Rural India. In: Chakrabarti,
Rajashri, Peterson, Paul E., editors. School Choice International: Exploring Public Private
Partnerships. MIT Press; Cambridge, Mass: 2008. p. 91-110.
National Sample Survey Office. Vol. 552. National Statistical Organization, Government of India; New
Delhi: 2013. Employment and Unemployment Situation among Major Religious Groups in India,
NSS 66th Round. Available at: http://www.minorityaffairs.gov.in/publications/employment-and-
unemployment-situation-among-major-religious-groups-india-nsso-66th-round
Papola, TS. ISID Working Paper Series 2012/2. Institution for Studies in Industrial Development; New
Delhi: 2012. Structural changes in Indian economy: Emerging patterns and implications.
Parcel, Toby L., Menaghan, Elizabeth G. Early Parental Work, Family Social Capital, and Early
Author Manuscript

Childhood Outcomes. American Journal of Sociology. 1994; 99(4):972–1009.

Pratham. Annual Status of Education Report. Pratham Documentation Center; New Delhi: 2005.
Pritchett, Lant. Rebirth of Education: Schooling Ain't Learning. Center for Global Development;
Washington DC: 2013.
Ruhm, Christopher J. Parental employment and child cognitive development. Journal of Human
Resources. 2004; 39(1):156–192.
Sayer, Liana C., Bianchi, Suzanne M., Robinson, John P. Are Parents Investing Less in Children?
Trends in Mothers' and Fathers' Time with Children. American Journal of Sociology. 2004; 110(1):
1–43.
Shariff, Abusaleh. India Human Development Report. Oxford University Press; New Delhi: 1999.
Sharma, Ursula. Women, Work and Property in North-West India. Tavistock Publications; London:
1980.
Short, Susan E., Chen, Feinian, Entwisle, Barbara, Fengying, Zhai. Maternal Work and Childcare in
China: A Multi- Method Analysis. Population and Development Review. 2002; 28(1):31–57.
Author Manuscript

Srinivas, Mysore Narasimhachar. The Changing Position of Indian Women. Man. 1977; 12(2):221–38.
Srivastava, Nisha, Srivastava, Ravi. Women, work, and employment outcomes in rural India. Economic
and Political Weekly. 2010; 45(28):49–63.
The Probe Team. Public Report on Basic Education in India. Oxford University Press; New Delhi:
1999.
Vandell, Deborah L., Ramanan, Janaki. Effects of Early and Recent Maternal Employment on Children
from Low-income Families. Child Development. 1992; 63(4):938–949. [PubMed: 1505249]
Waldfogel, Jane, Han, WenJui, Brooks-Gunn, Jeanne. The Effects of Early Maternal Employment on
Child Cognitive Development. Demography. 2002; 39(2):369–392. [PubMed: 12048957]

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 25

Zick, Cathleen D., Bryant, WKeith, Österbacka, Eva. Mothers' Employment, Parental Involvement,
and the Implications for Intermediate Child Outcomes. Social Science Research. 2001; 30(1):25–
Author Manuscript

49.
Author Manuscript
Author Manuscript
Author Manuscript

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Vikram et al. Page 26

Table I
Weighted Descriptive Statistics
Author Manuscript

Mean Std. Dev Range

Arithmetic test score 1.553 1.021 0-3
Reading test score 2.563 1.325 0-4
Stay-at-home mothers 0.362 0.481 0-1
Self-employed women 0.237 0.425 0-1
Women employed outside their homes 0.127 0.333 0-1
Women with multiple jobs 0.275 0.446 0-1
Attending parent-teacher meetings 0.444 0.497 0-1
Missing cases on parent-teacher meetings 0.060 0.237 0-1
Supervision of homework 0.382 0.486 0-1
Private schooling and coaching 0.451 0.625 0-2
Author Manuscript

Log of household income excluding maternal earning 9.794 1.835 0-13.84

Housing index 2.436 1.835 0-6
Maternal education (Higher than 8 years of education) 0.168 0.374 0-1
Years of education for the child 3.049 1.532 0-9
Sex of the child (female) 0.471 0.499 0-1
Age of the child 9.451 1.065 8-11
Number of children under 5 0.463 0.741 0-4
Number of teenagers in the household 0.557 0.859 0-7
Urban 0.253 0.435 0-1
Caste groups: Brahmin 0.055 0.227 0-1
Forward castes 0.223 0.417 0-1
Other backward classes 0.417 0.493 0-1
Author Manuscript

Dalit (lowest caste) 0.240 0.427 0-1

Adivasis (tribal) 0.066 0.248 0-1
Religions: Hindu 0.816 0.387 0-1
Muslim 0.130 0.337 0-1
Christian 0.018 0.132 0-1
Sikhs 0.017 0.130 0-1
Other religions 0.019 0.135 0-1

Source: IHDS 2004–5 data. N = 10,043

Author Manuscript

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Table II
Mean of Arithmetic Scores and Parental Involvement across Maternal Employment Categories

Stay-at-home mothers Self-employed women Probˆ Women employed outside Probˆ Women in multiple jobs Probˆ
Vikram et al.

Reading Scores 1.756 1.518 *** 1.430 *** 1.372 ***

Arithmetic Scores 2.771 2.452 *** 2.440 *** 2.442 ***
PTA Participation 0.548 0.384 *** 0.419 *** 0.369 ***
Help with homework 0.503 0.308 *** 0.384 *** 0.286 ***
Private school/coaching enrollment 0.675 0.432 *** 0.298 *** 0.242 ***

Source: IHDS 2004–5 data. N = 10,043;

ˆ
As compared to stay-at-home mothers

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
Page 27
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Table III
Multivariate Ordinal Regression on Arithmetic Scores

(2) Maternal employment- (5) Adding Private

VARIABLES (1) Control Model (3) Socio-economic status (4) Maternal Involvement
education interactions Enrollment
Vikram et al.

Self-employed women (Ref: Unemployed women) -0.103 (0.091) -0.094 (0.101) -0.024 (0.100) 0.018 (0.097) 0.044 (0.094)

Women employed outside -0.335*** (0.092) -0.424*** (0.104) -0.269** (0.104) -0.224* (0.102) -0.140 (0.101)

Multiple jobs -0.314*** (0.094) -0.323** (0.101) -0.202* (0.100) -0.154 (0.098) -0.096 (0.094)

Self-employed women*higher education -0.152 (0.187) -0.135 (0.187) -0.149 (0.188) -0.144 (0.187)

Women employed outside*higher education 0.630** (0.192) 0.564** (0.192) 0.478* (0.197) 0.416* (0.206)
Multiple jobs*higher education -0.060 (0.204) 0.035 (0.202) -0.046 (0.208) -0.028 (0.211)

Attendance in PTA 0.376*** (0.066) 0.302*** (0.064)

Missing on PTA -0.032 (0.121) -0.048 (0.123)

Supervision of home work 0.307*** (0.064) 0.271*** (0.064)

Private school/coaching 0.585*** (0.052)

Log of household income excluding maternal earning 0.034 (0.020) 0.033 (0.019) 0.026 (0.019)

Housing Index 0.164*** (0.020) 0.144*** (0.020) 0.110*** (0.020)

Maternal Education (Secondary and above) 0.755*** (0.070) 0.727*** (0.087) 0.557*** (0.089) 0.441*** (0.090) 0.357*** (0.091)

Years of education 0.513*** (0.024) 0.513*** (0.024) 0.500*** (0.024) 0.501*** (0.024) 0.518*** (0.024)

Sex of the child: Female (Ref: Males) -0.405*** (0.059) -0.405*** (0.059) -0.398*** (0.059) -0.414*** (0.059) -0.376*** (0.060)

Age of the child 0.135*** (0.030) 0.136*** (0.030) 0.148*** (0.031) 0.151*** (0.031) 0.139*** (0.030)

Urban 0.020 (0.073)

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
0.467*** (0.066) 0.453*** (0.066) 0.213** (0.073) 0.160* (0.073)

Brahmin (Ref: Forward caste) 0.549*** (0.163) 0.542*** (0.162) 0.487** (0.165) 0.479** (0.160) 0.465** (0.155)

Other backward classes -0.157* (0.073) -0.158* (0.073) -0.119 (0.073) -0.094 (0.073) -0.080 (0.074)

Scheduled caste -0.551*** (0.086) -0.553*** (0.086) -0.485*** (0.087) -0.443*** (0.086) -0.380*** (0.085)

Scheduled tribe -0.561*** (0.118) -0.557*** (0.118) -0.420*** (0.119) -0.390** (0.120) -0.328** (0.121)

Muslim (Ref: Hindus) -0.480*** (0.099) -0.485*** (0.099) -0.447*** (0.099) -0.407*** (0.098) -0.358*** (0.098)

Christian 0.356 (0.187) 0.339 (0.186) 0.365* (0.184) 0.318 (0.179) 0.374* (0.189)
Sikh 0.289 (0.172) 0.295 (0.173) 0.223 (0.173) 0.200 (0.171) 0.216 (0.185)
Others 0.242 (0.188) 0.239 (0.189) 0.236 (0.190) 0.244 (0.190) 0.240 (0.191)
Page 28
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

(2) Maternal employment- (5) Adding Private

VARIABLES (1) Control Model (3) Socio-economic status (4) Maternal Involvement
education interactions Enrollment

Number of children under 5 -0.157*** (0.047) -0.157*** (0.047) -0.134** (0.047) -0.126** (0.046) -0.124** (0.044)

Number of teenagers -0.170*** (0.050) -0.171*** (0.050) -0.184*** (0.051) -0.171*** (0.051) -0.150** (0.051)
Vikram et al.

Number of teenagers * girl 0.147* (0.066) 0.146* (0.066) 0.132 (0.068) 0.129 (0.067) 0.129 (0.066)

Constant cut1 0.248 (0.272) 0.222 (0.273) 0.984** (0.344) 1.338*** (0.339) 1.317*** (0.340)

Constant cut2 2.237*** (0.273) 2.212*** (0.274) 2.994*** (0.345) 3.368*** (0.340) 3.379*** (0.341)

Constant cut3 3.793*** (0.275) 3.770*** (0.276) 4.569*** (0.349) 4.957*** (0.344) 5.002*** (0.345)
Observations 10,043 10,043 10,043 10,043 10,043

Robust standard errors in parentheses; state controls not shown; adjusted for clustering in households.
***
p<0.001,
**
p<0.01,
*
p<0.05.

Source: IHDS 2004-05.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
Page 29
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Table IV
Multivariate Ordinal Regression on Reading Scores

(2) Maternal employment- (5) Adding Private

VARIABLES (1) Control Model (3) Socio-economic status (4) Maternal Involvement
education interactions Enrollment
Vikram et al.

Self-employed women (Ref: Unemployed women) -0.083 (0.080) -0.096 (0.086) -0.033 (0.085) 0.009 (0.085) 0.045 (0.084)

Women employed outside -0.284** (0.094) -0.353*** (0.105) -0.218* (0.103) -0.170 (0.103) -0.075 (0.103)

Multiple jobs -0.162 (0.084) -0.179* (0.088) -0.067 (0.089) -0.015 (0.088) 0.051 (0.087)

Self-employed women*higher education 0.025 (0.199) 0.038 (0.203) 0.017 (0.206) 0.025 (0.198)

Women employed outside*higher education 0.517** (0.200) 0.445* (0.197) 0.355 (0.198) 0.263 (0.204)

Multiple jobs*higher education 0.014 (0.219) 0.110 (0.222) 0.057 (0.233) 0.061 (0.240)

Attendance in PTA 0.340*** (0.060) 0.258*** (0.059)

Missing on PTA -0.158 (0.118) -0.175 (0.116)

Supervision of homework 0.281*** (0.064) 0.236*** (0.064)

Private school/coaching 0.643*** (0.051)

Log of household income excluding maternal earning 0.035* (0.016) 0.036* (0.015) 0.029 (0.015)

Housing Index 0.144*** (0.019) 0.126*** (0.020) 0.088*** (0.020)

Maternal Education (Secondary and above) 0.874*** (0.074) 0.814*** (0.089) 0.656*** (0.092) 0.550*** (0.093) 0.460*** (0.096)

Years of education 0.512*** (0.023) 0.511*** (0.023) 0.498*** (0.023) 0.500*** (0.023) 0.521*** (0.023)

Sex of the child: Female (Ref: Males) -0.230*** (0.059) -0.230*** (0.059) -0.223*** (0.059) -0.236*** (0.060) -0.192** (0.060)

Age of the child 0.148*** (0.029) 0.149*** (0.029) 0.160*** (0.030) 0.164*** (0.029) 0.151*** (0.029)

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
Urban 0.386*** (0.063) 0.378*** (0.063) 0.169* (0.069) 0.122 (0.069) -0.027 (0.069)

Brahmin (Ref: Forward caste) 0.562*** (0.141) 0.555*** (0.141) 0.509*** (0.142) 0.491*** (0.139) 0.480*** (0.136)
Other backward classes -0.056 (0.072) -0.057 (0.072) -0.024 (0.072) -0.004 (0.072) 0.016 (0.073)

Scheduled caste -0.388*** (0.082) -0.388*** (0.082) -0.327*** (0.083) -0.295*** (0.082) -0.237** (0.082)

Scheduled tribe -0.390*** (0.107) -0.386*** (0.107) -0.254* (0.109) -0.228* (0.110) -0.175 (0.110)

Muslim (Ref: Hindus) -0.279*** (0.082) -0.283*** (0.083) -0.248** (0.083) -0.214** (0.083) -0.152 (0.083)

Christian 0.166 (0.202) 0.155 (0.202) 0.160 (0.202) 0.112 (0.197) 0.175 (0.201)
Sikh 0.327 (0.260) 0.328 (0.260) 0.272 (0.256) 0.251 (0.248) 0.245 (0.232)
Others 0.278 (0.184) 0.274 (0.184) 0.263 (0.183) 0.258 (0.183) 0.284 (0.183)
Page 30
 Author Manuscript Author Manuscript Author Manuscript Author Manuscript

(2) Maternal employment- (5) Adding Private

VARIABLES (1) Control Model (3) Socio-economic status (4) Maternal Involvement
education interactions Enrollment

Number of children under 5 -0.146*** (0.039) -0.146*** (0.039) -0.126** (0.039) -0.117** (0.038) -0.117** (0.038)

Number of teenagers -0.190*** (0.041) -0.190*** (0.041) -0.204*** (0.041) -0.192*** (0.041) -0.178*** (0.041)
Vikram et al.

Number of teenagers * girl 0.154** (0.058) 0.153** (0.058) 0.144* (0.058) 0.138* (0.057) 0.138* (0.057)

Constant cut1 -0.275 (0.275) -0.294 (0.276) 0.433 (0.323) 0.780* (0.325) 0.725* (0.323)

Constant cut2 0.943*** (0.273) 0.924*** (0.274) 1.659*** (0.322) 2.016*** (0.323) 1.978*** (0.321)

Constant cut3 2.194*** (0.274) 2.176*** (0.275) 2.922*** (0.323) 3.290*** (0.325) 3.277*** (0.322)

Constant cut4 3.364*** (0.278) 3.348*** (0.279) 4.104*** (0.327) 4.481*** (0.330) 4.498*** (0.327)
Observations 10,043 10,043 10,043 10,043 10,043

Robust standard errors in parentheses; state controls not shown; adjusted for clustering in households.
***
p<0.001,
**
p<0.01,
*
p<0.05.

Source: IHDS 2004-05.

Soc Sci Res. Author manuscript; available in PMC 2019 May 01.
Page 31