social problem

A Contextual Analysis of Gender Disparity in Education in India: The Relative Effects of Son Preference, Women's Status, and Community

Author(s): Sunita Bose

Source: Sociological Perspectives , Vol. 55, No. 1 (Spring 2012), pp. 67-91

Published by: Sage Publications, Inc.

Stable URL: https://www.jstor.org/stable/10.1525/sop.2012.55.1.67

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to Sociological Perspectives

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A strong preference for sons over daughters in Indian families has been documented over centuries (Phillip and Bagchi 1995). A major consequence of such son prefer- ence is selective neglect or discrimination against daughters in nutrition, health care, and education. Even today, discrimination from birth leads to significantly poorer life chances for female children. Previous research shows that girls often receive less nutritious food or fewer calories than boys and are far less likely to receive medical care when they are sick (Bose 2011; Ghosh 1995; Nag 1991; Pande 2003). Selective neglect of young girls is often linked to excess mortality of females at young ages in India (Ghosh 1995; Oster 2009; Velkoff and Adlakha 1998).

The research on education also shows that girls are at a considerable disadvan- tage compared to boys. Girls are much less likely to go to school than boys (King- don 2005). In the event that they do go to school, girls are much more likely to be

A CONTEXTUAL ANALYSIS OF GENDER DISPARITY IN EDUCATION IN INDIA: THE RELATIVE EFFECTS

OF SON PREFERENCE, WOMEN’S STATUS, AND COMMUNITY

SUNITA BOSE State University of New York at New Paltz

ABSTRACT: Multi-level analyses of data on 18,519 families with opposite sex children from NFHS-3 are used to test the impact of maternal son preference and context on the gender differential in education in India. The results show that girls are at a greater educational disadvantage compared to their brothers in families with maternal son preference. Mother’s education is shown to reduce the effect of son preference and to reduce the bias against daughters. Additionally, there is more gender inequality in education in communities with high levels of maternal son preference and low women’s status. Importantly, a family’s location in a community or region with higher women’s status explains much of the effect of maternal son preference on the gender disparity. The importance of context is further solidified by significant interaction effects showing that the negative effect of mother’s education is modified by context. Keywords: gender gap, education, India

Sociological Perspectives, Vol. 55, Issue 1, pp. 67–91, ISSN 0731-1214, electronic ISSN 1533-8673. © 2012 by Pacific Sociological Association. All rights reserved. Please direct all requests for permission to photo- copy or reproduce article content through the University of California Press’s Rights and Permissions website, at http://www.ucpressjournals.com/reprintinfo.asp. DOI: 10.1525/sop.2012.55.1.67.

Address correspondence to: Sunita Bose, 600 Hawk Drive, New Paltz, NY 12561; e-mail: [email protected].

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

68 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

withdrawn from school at an earlier age because they have to help out at home, or because of financial difficulties (Basu 1992). However, while there are numerous studies on the gender disparity in education in India, as well as widespread recogni- tion that son preference leads to gender discrimination, none have examined the role of context in shaping the persistent educational disparity between boys and girls.

Much of the prior research on education only investigates parental and family characteristics as predictors of gender differential. A key factor that contributes to gender discrimination against daughters is women’s status. It has been posited that women with higher status, particularly those with more education, may be less inclined to discriminate against their daughters as a result of having more knowledge and control over resources (Afridi 2010). However, there are no stud- ies that investigate the mechanism through which maternal status operates on son preference and gender discrimination. Of particular importance to this study is the lack of research investigating the link between individual-level women’s status and the community in which they live.

The assumption that shared membership in a community modifies individual behavior is central to this investigation of gender bias in children’s education. The extent of differential treatment may vary across families in different social settings. In the context of gender bias, individual characteristics, such as a lack of education, might inhibit the effective flow of information about gender equality. But commu- nity norms and attitudes in the form of village or neighborhood-level characteristics can also influence ideas and practices beyond the individual characteristics (Kravdal 2004; Parashar 2005). Therefore, a key question here is: what is the relative effect of son preference, mother’s status, and context on educational discrimination against daughters in India? Can community norms overwhelm individual preferences?

Background

Globally, women have recorded significant gains in education, particularly at the lower levels. Women are acquiring primary and secondary education in larger numbers than ever before (Deolalikar 1993; Hossain and Tisdell 2005; Obasi 1997; Worku 2001). However, in spite of these gains, there is a persistent gender gap in educational attainment in developing countries. Information from demo- graphic and health surveys between 1994 and 2005 show that in forty countries less than one-quarter of reproductive age women have completed secondary school (Vadnais, Kols, and Abderrahim 2006). Explanatory research on the gen- der gap typically looks at the parent-child relationship and expectations about the future (Kabeer 2000). Economic theory would predict that parents are more likely to educate their children if they can expect a financial return in the future (King- don 2002). In developing countries, particularly for the poor, an additional factor that needs to be taken into account is the relevance of education for the future of the children. In India, both these perspectives can be useful in understanding the gender disparity in education.

As a result of the low priority placed on women’s education within families, India has one of the lowest female literacy rates in Asia. In addition, a large gender gap persists between the literacy rates for males and females. While substantial gains

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 69

have been made in education in the last few decades, the 2001 census recorded 75.3 percent literate men compared to only 53.7 percent literate women (Census of India 2001). The census records also show that more males are literate than females in every Indian state. Disturbingly, 17 percent of the total districts in India had a gender gap in literacy rate above 30 percentage points (Bhargava 2002).

For many families, education is a way to increase their daughters’ marriage poten- tial (Dhruvarajan 1989; Nambissan 2005; Vlassoff 1996). While higher education for women is often considered unnecessary and suspect, many believe that some edu- cation is beneficial because it makes women better housekeepers, allows them to maintain better household accounts, and helps them bring up children better by being able to supervise their education (Jeffery and Jeffery 1996; Somjee 1989; Vlas- soff 1996). In rural areas, boys want educated wives because they are more socially adept and no longer rustic (Jeffery and Jeffery 1994). However, a little education (most often up to primary school) is often considered sufficient for these purposes.

A look at enrolment patterns of children confirms this attitude among parents. Nationally, about three out of four children aged 6–17 years attend school. But school attendance is significantly higher for boys than girls, particularly in rural areas where attendance rates for girls is about 12 percentage points lower than that for boys (NFHS-3 2007). Data on school attendance also show that the propor- tion of girls attending school decreases with age, while for boys it remains stable (Ghosh 1995; NFHS-3 2007; Velkoff 1998). By ages 15–17 years, about half the boys are in school compared to only about a third of the girls in that age group (NFHS-3 2007). Girls are most often taken out of school to help with family responsibili- ties, or after they reach puberty as a way of protecting their honor. For millions of landless laboring families who travel for eight months out of the year, boys are left behind to continue their schooling but girls are taken along because of concerns over their safety and the general low priority afforded to education for girls (Joshi 1998). In rural areas, girls’ education is also stopped because parents will not allow their daughters to travel beyond the village (Jeffery and Jeffery 1994).

Theoretical Framework

Increased education is associated with better bargaining power and a resultant increase in resource control (Sen 1990). It is thought to improve a woman’s efficiency and ability to deal with the outside world (Amin 1996). Inside the home, increased education may improve women’s status by giving them more decision-making authority, more equal partnership with their spouses, and the ability to overcome or resist some cultural biases and norms (Amin 1996; Mukhopadhyay and Garimella 1998). Studies also show that discrimination against women is decreased by influ- ences that give women more voice and agency within the family (Dreze and Sen 1995). One of these influences is education. Thus, gender discrimination in educa- tion has an impact not only on the women who are denied education by limiting their options, but also on future generations of their daughters.

In general, bias against female children may be seen as a result of strong son preference in India. In turn, a preference for sons usually results from a combina- tion of cultural and economic factors. Marriage usually requires a move of the bride

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

70 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

to her husband’s family, and convention requires that her entire work effort and its returns belong to that family. This expectation often makes parents unwilling to spend scarce resources on educating a daughter given that there will be no future economic returns for her natal family (Banerjee 1998). Culturally, it has been argued that gender disparity exists as a result of prejudice against female education and also against their working outside the home. Education may make girls unsuitable by changing their ability to adjust to their husband’s household (Sopher 1980). While some education is socially valued because it makes girls better wives and mothers, there is no hesitation about stopping her education to teach her household skills given the perceived irrelevance of higher education to her future.

Much of the existing analyses of gender discrimination tend to focus on micro- or individual-level characteristics of families or parents. However, given the important role of community on social behavior in India, a more complete theoret- ical framework for explaining gender disparity should also consider community norms and their intersection with individual-level attitudes as a potential mecha- nism of transmission of bias against daughters. In the following sections I outline a contextual framework that combines micro- (individual-level) and macro- (community-level) explanations of discrimination against daughters in India. The micro-component of the model shows that maternal/family socioeconomic characteristics affect level of son preference, which in turn affects gender discrimi- nation in education. The macro-component shows that community characteris- tics, particularly indicators of community norms about women’s status, affect the micro-level relation between individual characteristics and gender discrimination.

Importance of Individual-Level Characteristics (Micro-Component)

Son Preference. Discrimination against daughters is often thought to be a direct consequence of son preference. For women in particular, sons are a source of immediate and future power and well-being. Culturally, sons are preferred because of religious and social reasons. Sons are necessary for many Hindu rituals and particularly those that ensure the well-being of the soul after death (Kishor 1995). Socially, sons are preferred because they continue the patriline.

Economically, sons are preferred because they provide support in their parents’ old age, are the preferred heirs of property, and are the recipients of dowry upon marriage (Arnold 1997; Das Gupta 1998; Dube 1997). Daughters are a liability because parents incur large expenses for their marriage and do not get any finan- cial return from their daughters’ labors. Marriage can also create a lifelong burden of additional expenses with no reciprocity for a woman’s natal family (Miller 1981; Srinivas 1984). In addition, in societies where females are economically dependent on males and are virtually excluded from mainstream economic activities, there is a greater premium placed on sons (Desai 1998). Women’s access to material and physical comforts are often dependent on their sons. They provide an impor- tant source of insurance against the risk of losing economic support of husbands through widowhood, illness, or divorce. A son gives a woman a permanent link to her husband’s kin since succession and inheritance passes through males. In India, generally only sons have a moral obligation to look after their elderly parents,

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 71

while society frowns upon parents accepting material help from a married daugh- ter. In this context it is not unreasonable to provide additional education and train- ing to sons over daughters to ensure the future well-being of the family.

Women’s Status. The preceding discussion suggests that an analysis of educa- tional discrimination as a consequence of son preference must consider a variety of individual-level characteristics that may impact son preference. In particular, factors that are directly related to women’s agency are thought to have strong effects on female disadvantage (Afridi 2010). By increasing women’s agency and control over resources, women may be able to improve childcare and may also raise the impor- tance assigned to girls (Dreze and Sen 1995). Among other factors, education and productive wage employment is thought to enhance women’s ability to overcome cultural barriers and control resources. Previous research has shown that education can erode traditional sex-based attitudes and develop more egalitarian views (Amin 1996; Mukhopadhyay and Garimella 1998). Education also influences a woman’s position in her family and society and may consequently increase her perceived value of daughters. It exposes women to more opportunities and provides them with control over material resources (Satia and Jejeebhoy 1991). Educated women may also be able to make better use of existing resources in their child care prac- tices. Prior research confirms that mother’s education has a strong positive effect on girls’ schooling and enrolment although not on boys’ schooling (Kingdon 2002; Unni 1998). Based on this research I expect that increasing mother’s education will decrease son preference and lower discrimination against female children.

Another major factor affecting a woman’s worth is her labor force participation. Women’s bargaining power is increased by outside earnings because they improve her chances of survival should her marriage dissolve (Sen 1990). In addition, paid work gives her a clearer perception of her individuality and well-being by provid- ing her with higher perceived contribution to her family’s economic position. Sen (1990) maintains that perceptions are important for defining entitlement to resource within the family. Work improves women’s social standing within the family and the society. Her contribution to family prosperity is more visible, and she has more voice because she is less dependent on others (Dreze and Sen 1995). I expect work- ing mothers to be more willing and able to expend resources on her daughters com- pared to non-working mothers. If paid work is indeed a proxy for higher status, then I also hypothesize that working mothers will exhibit lower levels of discrimination against their daughters as a result of lower levels of son preference.

Importance of Community Characteristics

While economics and culture can explain much of gender discrimination, most researchers tend to focus on individual or family characteristics as explanatory factors in their analyses. A major weakness is the exclusion of the community in the analytical framework. Hindu culture stresses the importance of the collectivity. Some scholars, therefore, see Indian society as one where the individual interest is subordinated to collective interests (Dumont 1970), while others see a distinct and important role for the individual in spite of the collective role (Mines 1992). An important characteristic of community is a commitment to a set of shared values

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

72 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

and norms (Etzioni 1996). This concept of community is particularly important to understand social life in India. Communities set the context for social life in India and individual action is often affected by groups or communities of which the individual is a member.

Relations in India are further characterized by local dependence. Social net- works are close knit and one has obligations dependent on kinship, caste, and neighborhood (Beteille 1974; Das 2001). Members of a village or even an urban neighborhood are so close that they are commonly addressed in kinship idiom. Thus, older neighbors are often called aunt or uncle and are likely to call non-fam- ily members to order when deemed necessary. There is also a high value placed on personal loyalty and ties of personal dependence. Economic, political, and social relations are all dependent on the local community (Das 2001; Dhruvarajan 1989; Srinivas 1998). Communities impose social sanctions against members who do not conform to the shared norms and values of the community with the most severe penalty being exclusion, which is described as “a kind of social excommunication” (Dumont 1980:178).

Given the importance of local relations in India, it is reasonable to expect gen- der discrimination in education to be affected by the community context. In other words, discrimination depends not only on individual characteristics, but also on the setting, the surrounding environment where people live and work. Indeed, the finding that there is considerable regional variation in son preference in spite of the strong patriarchal bias in most of India (Arnold 2001) is consistent with this expec- tation. These differences may be traced to different marriage and kinship patterns in different parts of India (Dyson and Moore 1983 Karve 1965. Areas in the north are culturally less favorable to women than areas in the south. Culture dictates greater value of males as both producers and heirs and undervalues the role of women in the north (Miller 1997). Son preference is strongest in the north, char- acterized by male-centered kinship patterns, where married women face isolation from natal kin due to local exogamy. In this situation, a woman can consolidate her position in her affinal home by giving birth to sons. When women are mar- ginalized, sons and brothers are more valued than daughters or sisters for every individual (Das Gupta 1987). Compared to their northern counterparts, women in the south are more likely to be more valued, more likely to survive, be educated, marry later, and be involved in the productive economy (Malhotra, Vanneman, and Kishor 1995). Therefore, I expect northern women to exhibit the highest level of discrimination against their daughters compared to the other regions.

On a more disaggregated level, characteristics of villages and urban neighbor- hoods should also affect familial relations. In communities where women have higher status, son preference is likely to be lower, and as a result, discrimination against daughters is also likely to be lower. Thus, I expect that the higher the level of female education in the community or the economic contribution of women, the higher the status of women in that area. This should lead to less discrimination against daughters regardless of the characteristics of the child’s family.

Moreover, aside from the direct effect of community on gender discrimination, there should also be an interactive effect of individual and community charac- teristics. Therefore, an analysis of educational discrimination should include an

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 73

examination of how the individual and the community intersect with regard to the decision to invest more in boys’ education than in girls’ education. In par- ticular, the effect of individual-level characteristics such as maternal education or paid work will not influence son preference the same way across communi- ties. In communities with higher women’s status, where daughters are treated relatively well, most families are likely to follow suit regardless of their indi- vidual preferences. For example, it has been shown that in areas of high female labor force participation, boys and girls are treated more equally even when the mother does not work herself (Basu 1992). Alternatively, in predominantly male power structures, women realize and internalize their lower worth and use whatever resources they control to selectively favor sons (Basu 1992). In general, I expect that in communities with higher women’s status, the effects of individ- ual-level work or education may be somewhat diluted compared to communi- ties with lower women’s status. Thus, where female education is the norm, a low or uneducated mother may also be inclined to give children of both sexes access to education because of social pressures. On the other hand, if community-level female education is low, then higher maternal education is expected to have a more equalizing impact on children’s education as a more educated mother may be better able to withstand social pressures against girls’ education. Similarly, areas with a high proportion of women working should value women more than areas where very few women work. Thus, even if an individual mother does not work for pay, the family should exhibit less discrimination against daughters because socially they are valued.

Full Model

Figure 1 presents a diagram of the combined theoretical model with the indi- vidual (micro-) and contextual (macro-) components. Gender disparity in educa- tion among sons and daughters is shown as a consequence of son preference. The micro-component of the model specifies that individual or family characteristics affect gender disparity in education. There are both direct and indirect impacts. The direct effect shows that higher women’s status reduces the educational differ- ential by providing more egalitarian treatment of daughters and sons. The indirect effect operates through maternal son preference as higher women’s status reduces the extent of son preference leading to a further lowering of discrimination against daughters.

The macro-component of the model shows that there are three possible path- ways for community characteristics to affect the gender disparity in education. First, there is a direct effect of community since regions or communities with higher women’s status are likely to exhibit more egalitarianism and lower educational dis- advantage for girls. Second, there is an indirect effect as communities with higher women’s status are likely to show lower levels of son preference, thereby result- ing in lower gender disparity in education. Finally, it is likely that individual and community characteristics intersect to affect the micro-level relationship between maternal characteristics and gender disparity in education to dilute or strengthen the micro-relationship depending on the social environment of the family.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

74 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

DATA AND METHODS

The data for this study are taken from the third Indian National Family Health Sur- vey (NFHS3) conducted in 2005–06. The NFHS3 is a nationally representative survey that interviewed 124,385 women aged 15–49 years. Use of this survey makes it pos- sible to directly test the level of gender discrimination within families by comparing opposite sex siblings. The hierarchical data structure also allows for a multi-level modeling approach by tying individual characteristics to community-level attri- butes. The analysis is based on data on children aged 7–18 years. Level 1 or micro- level variables are individual-level variables while Level 2 or macro-level variables are at the village (in rural areas) or urban block level. The sample consists of families with identifiable children of both sexes in the household dataset. The NFHS3 has 21,991 families with opposite sex children (of the household head or his/her spouse) between the ages of 7 and 18. After accounting for missing data, the final sample size is 18,519 families within 3,688 primary sampling units (PSU) or communities.

Dependent Variable

The most direct measure of gender differential in treatment comes from families with children of both sexes. Families with children of only one sex cannot, by defi- nition, exhibit preferential treatment of one over the other. Therefore, the sample

Micro Component: Individual/family characteristics Women’s status - Mother’s education - Mother’s work Other socio-economic characteristics

Son Preference

Macro Component: Community characteristics Women’s status - female education - female paid work Geographic region Urban/rural location

Gender disparity in education among male and female children

FIGURE 1 Theoretical Framework for Educational Disparity Among Males and Females

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 75

in this study only includes families with children of both sexes. The dependent variable is the gender differential in education between opposite sex siblings. It is measured as the difference in the years of education between a randomly chosen male and a randomly chosen female sibling between 7 and 18 years of age from each family. The selected age range allows for sufficient time for a child to start school, yet keeps the analysis focused on the extent and impact of gender differen- tial among children. It should be noted that the survey only provides information on current residents in the household. Since females are more likely than males to have married and moved out of the household, it is likely that the extent of edu- cational disparity is somewhat understated in this sample. Girls in their mid- to late teens are most likely to have a larger disadvantage compared to their male counterparts, but are more likely to be missing from this dataset than their broth- ers of similar age.

Level 1 Explanatory Variables

A primary individual-level explanatory variable is maternal son preference, a dichotomous variable coded 1 for women with stated son preference and 0 for those without. This measure is derived from respondents’ reports about their perceived ideal sex composition of families. Women were asked to state the ideal number of sons and the ideal number of daughters a family should have. The vari- able is constructed so that women who believe that the ideal family composition should have more sons than daughters are identified as having son preference. The actual level of son preference in this sample is likely slightly underestimated because women who gave non-numeric responses (such as “up to God”) are con- sidered not to exhibit son preference by this measure even though many of them may actually prefer sons to daughters.

There are two other Level 1 explanatory variables, both reflecting women’s sta- tus. As stated earlier, higher women’s status should be associated with lower lev- els of discrimination against daughters. Mother’s education is a continuous variable that measures the number of completed years of education that the mother has. Mother’s work is measured by a dummy variable coded 1 for working mothers and 0 otherwise.

Level 2 Explanatory Variables

There are several macro-level variables that I use in order to account for the social context. First, there is evidence that the gender disparity in education varies by region. The northern states have the lowest literacy rates and the highest gen- der disparity (Bhargava 2002; NHFS-3 2007). For example, in the northern state of Rajasthan, the literacy rate for women aged 15–49 is only 36 percent compared to a male literacy of 74 percent (NFHS-3). In Jharkhand, the gender gap in literacy is 38 percentage points (Census of India 2001). Among the other areas, Kerala (with exceptionally high female literacy rates) and some of the north-eastern hill states have relatively small gender disparities. Moreover, there are regional differences in son preference as the cultural and economic factors that promote discrimination against daughters are more diluted in the south than in the north. I divide India

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

76 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

into three regions: north, south, and other regions. North comprises the north and north-western part of India. South includes the states of Andhra Pradesh, Karna- taka, Kerala, Tamil Nadu, and Maharashtra. Maharashtra is considered an inter- mediate state that is often culturally considered closer to the south than the north (see Sopher 1980 for a detailed discussion), and therefore, I include it in this region. The rest of India falls into the “other” category. Given the kinship structures and relative status of women in each region, I expect the north to exhibit the highest levels of discrimination, the south to exhibit the least, and the other regions to exhibit intermediate levels of discrimination against female children.

I also expect the level of women’s status in the community to impact gender discrimination. Community-level women’s status is operationalized through the levels of female education and paid work within a PSU. In the NFHS3, a PSU is a city block in urban areas or a village in rural areas. The following macro-level vari- ables are calculated by taking individual responses and aggregating them to the PSU level. All aggregations were done on the full sample of women (N = 124,385) rather than on the subsample used for this analysis in order to have a more accu- rate representation of the community. The percentage of literate females in the PSU is used to approximate the level of adult female education in the immediate social environment. The percentage of women working for pay in the PSU approximates the amount of economic contribution made to the area by women.

Finally, I expect the community-level son preference to impact gender differ- entials in education. If the community exerts an influence on families, then com- munities with high levels of son preference are expected to exhibit higher levels of gender discrimination in education, regardless of the individual maternal son preference. The percentage of mothers with son preference is expected to approximate the community-level son preference.

Controls–Level 1

Prior research shows that class is an important predictor of the consequences of son preference (Kishor 1995; Miller 1981). Therefore, the Level 1 control variables include indicators of both social and economic class. Social class is operationalized by caste, while economic class is defined as standard of living. The Constitution of India makes special provision for a list of backward castes and tribes (includ- ing the formerly untouchables) that are called scheduled castes and tribes. In my study, scst is a dummy variable that is coded 1 for scheduled caste or tribe and 0 otherwise. The standard of living index (SLI) is an already constructed variable from data on housing characteristics and household possessions (see NFHS-3 2007 for detailed explanation of variable construction). Two dummy variables, medium sli (1 = medium; 0 = otherwise) and high sli (1 = high; 0 = otherwise), reflect the eco- nomic class of the families under consideration. The reference category is low sli.

The cultural worth of women in India is also dependent upon religion. Religion is comprised of two dummy variables—Hindu (coded 1 for Hindus and 0 other- wise) and Muslim (coded 1 for Muslims and 0 otherwise). Other religions is the reference category. While previous research on education shows that Muslims are at a disadvantage compared to non-Muslims (Husain and Sarkar 2011; Kingdon

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 77

2005), explanations of son preference usually include a description of the need for sons under Hinduism. Therefore, I expect both Hindus and Muslims to exhibit more gender discrimination compared to other religions.

Past research on gender bias against daughters has shown that the number of surviving siblings can affect the extent of bias (Pande 2003). In this study, number of children is a continuous variable that measures the number of living children still at home since other children’s presence can have an impact on the resources expended on the children under consideration. Mother’s age is a continuous vari- able that provides the age of the mother at the time of the survey. After marriage a young woman often loses voice and agency and has limited authority or deci- sion-making ability. However, her status increases with age, particularly if she has sons (Das Gupta 1995). Therefore, I expect older women to have more control over household resources and, as a result, be able to allocate more resources to their daughters. Finally, to make sure that the gender differential in education among siblings is not simply a result of the age difference between them, the analysis includes a control for children’s age difference. This continuous variable is calculated by subtracting the age of the sister from the age of the brother.

Controls–Level 2

The data on educational attainment show that there is a large rural urban dif- ference, with girls in urban areas being much more likely to attend school (Census of India 2001; NFHS-3 2007). In this study, location is operationalized through a dummy variable, urban, that is coded 1 for families in urban areas and 0 otherwise. Urban areas receive more exposure to Western ideals and practices, including more egalitarian views about gender roles. This should lead to lower son prefer- ence in urban areas and lower levels of discrimination against female children.

Analytical Strategy. I begin my investigation by examining the bivariate dif- ferences in gender differential in education between siblings as well as the differ- ences in the potential mediating variables and control variables in families with and without maternal son preference. Next I conduct multilevel regression anal- yses that are organized into several models. An initial baseline model includes the control variables along with the primary explanatory variable, maternal son preference, in order to examine the association between son preference and gen- der disparity in education while controlling for individual and family background factors. Subsequent models add the rest of the explanatory variables to this base model to examine their ability to account for the impact of son preference on gen- der disparity in education among siblings. All models are based on hierarchical linear modeling (HLM) as the appropriate strategy to use with nested, multilevel data (Raudenbusch and Bryk 2002).

RESULTS

As expected, the results indicate greater gender disparity in families with stated maternal son preference. Table 1 provides descriptive statistics for all variables disaggregated by maternal son preference. Of the 18,519 pairs of siblings, about

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

78 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

27 percent (n = 5,067) had mothers who exhibited son preference. The results for the dependent variable show that brothers have significantly higher levels of education than sisters in families where mothers exhibit son preference, with a brother having about .25 extra year of education than his sister (compared to a gender disparity of –.17 in families without maternal son preference).

TABLE 1 Descriptive Statistics for Variables Used in Analyses of Educational Disparity Among Opposite Sex Siblings, National Family Health Survey, India, 2005–06

(N = 18,519)

Maternal Son Preference (N = 5,067)

No Son Preference (N = 13,452)

M SD M SD

Dependent variable Education differencea 0.249 3.335 –0.169 3.386

Explanatory variables Level 1

Female status: Mother’s educationa 1.966 3.412 3.913 4.703 Mother’s worka 0.544 0.498 0.486 0.500

Level 2 Region

North (reference)a 0.597 0.490 0.446 0.497 Southa 0.132 0.339 0.277 0.447 Other region 0.271 0.444 0.277 0.447

Community Pct. son preferencea 30.612 14.985 18.383 13.358 Pct. female literatea 53.361 25.772 65.320 25.280 Pct. female labor forcea 43.796 25.572 39.994 23.739

Control variables Level 1

Scheduled caste/tribea 0.355 0.478 0.316 0.465 Standard of living

Low (reference)a 0.297 0.457 0.210 0.407 Mediuma 0.399 0.490 0.339 0.473 Higha 0.303 0.460 0.450 0.498

Religion: Hindua 0.727 0.445 0.701 0.458 Muslima 0.180 0.384 0.158 0.365 Other (reference)a 0.092 0.289 0.140 0.347

Mother’s agea 36.527 5.501 36.250 5.359 Number of childrena 4.390 1.590 3.797 1.626 Sibling age differencea –0.158 4.120 –0.298 3.934

Level 2 Urbana 0.316 0.465 0.445 0.497

a. Difference between the two groups is statistically significant at the .05 level.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 79

Significant differences between siblings with and without maternal son prefer- ence also emerge among several of the other variables. Among the explanatory vari- ables, the difference in mother’s education by son preference stands out. Mothers who exhibit son preference have an average educational level that is almost half of the educational level of mothers who do not exhibit son preference (2 compared to 3.9 years). Further, the bivariate results show that a higher percentage of working women exhibit son preference relative to those with no son preference (54 percent versus 49 percent). While this is contrary to the expectation that working women have higher status and should exhibit less son preference, it may be a reflection of the economic status of the women who are most likely to work for pay in India. If poor women are the ones most likely to be in the paid work force, then maternal work is not necessarily an indicator of higher status since these women are working out of necessity (in low paying informal sector jobs) rather than by choice.

The differences among the Level 2 variables are consistent with some of the research on son preference. Siblings whose mothers exhibit son preference are more likely to be from northern India. At the community level, families with maternal son preference are more likely to be located in neighborhoods or villages where the average percentage of son preference is almost double that of communities where families without maternal son preference are located. There is also a statisti- cally significant difference in the community-level female literacy among the two groups. The average community-level female literacy is considerably higher for families where mothers do not exhibit son preference. However, as with maternal work, son preference is more evident in communities with more working women, a further indicator that paid work is not necessarily a sign of greater women’s status in India.

Among the control variables, children with maternal son preference are a little more likely to be a member of a scheduled caste or tribe, have a low or medium standard of living, and be from a rural area, compared to children without mater- nal son preference. It should be noted that sibling age difference shows that on average the sister is somewhat older than the brother in families both with and without son preference. It is possible that this difference in sibling ages accounts for the slight educational advantage for girls in families without son preference. At the same time, it is noteworthy that girls are at a disadvantage educationally in spite of being older on average in families with maternal son preference. Siblings in families with maternal son preference are also slightly more likely to be Hindu or Muslim and slightly less likely to belong to other religions compared to siblings in families without maternal son preference.

Tables 2 and 3 provide results from the multivariate, multi-level analyses of the educational difference between opposite sex siblings within a family. Overall, it is clear that context overwhelms maternal characteristics in that it matters more than does son preference in predicting the gender disparity in education. Further, context interacts with mother’s education to modify the effect of this individual- level characteristic on the gender differential in children’s education. The multi- level analyses were conducted in several stages. First, a fully unconditional model on the dependent variable confirmed that educational status varies across the PSUs. Next, I added son preference as an independent variable to show the impact

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

80 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

of son preference on the gender differential in education at the bivariate level. The results confirmed that son preference has a strong positive effect on gender discrimination in education against female children. Daughters in families with maternal son preference are more disadvantaged compared to daughters in fami- lies without son preference (results not presented). The next stage was the formu- lation of a baseline multivariate model.

Model 1 (Table 2) is the base model and includes all the control variables along with maternal son preference to provide a baseline effect of son preference on educational difference. The fixed effects show that son preference has a statistically

TABLE 2 Multilevel Regression Analysis of Gender Disparity in Education between Opposite Sex

Siblings, National Family Health Survey, India, 2005-06 (N = 18,519)

Model 1 Model 2 Model 3

b SE b SE b SE

Fixed effects Explanatory variables Level 1

Son preference 0.190** 0.039 0.165** 0.039 0.168** 0.039 Female status:

Mother’s education –0.033** 0.005 Mother’s work 0.068 0.036

Level 2 Region (ref = North):

South –0.209** 0.047 Other region –0.173** 0.046

Community: Pct. son preference Pct. female literate Pct. female labor force

Control variables Level 1

Scheduled caste/tribe –0.010 0.041 –0.040 0.041 –0.013 0.041 Standard of living (ref = low):

Medium –0.189** 0.046 –0.162** 0.046 –0.199** 0.046 High –0.441** 0.051 –0.316** 0.054 –0.469** 0.051

Religion (ref = other religion): Hindu 0.226** 0.056 0.172** 0.056 0.172** 0.058 Muslim 0.177* 0.073 0.113 0.074 0.140 0.074

Mother’s age 0.009** 0.003 0.010** 0.003 0.009** 0.003 Number of children 0.050** 0.012 0.029* 0.013 0.042** 0.012 Sibling age difference 0.613** 0.004 0.612** 0.004 0.613** 0.004

Level 2 Urban –0.248** 0.041 –0.187** 0.041 –0.234** 0.041

Random effects: Intercept 0.151** 0.025 0.138** 0.025 0.141** 0.025

–2 log likelihood 83,414 83,375 83,397

(Continued)

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 81

significant, positive effect on the gender disparity in education. Among families with maternal son preference, a brother’s educational advantage over his sister is increased by a fifth of a year (b = .19), as compared to families without mater- nal son preference. Among the control variables, families with a medium or high standard of living exhibit less gender disparity than families with a low standard of living. Girls in Hindu and Muslim families are more disadvantaged educa- tionally than their brothers as compared to girls in families belonging to other religions. As expected, sibling age difference has a positive impact on the gender

TABLE 2 Multilevel Regression Analysis of Gender Disparity in Education between Opposite

Sex Siblings, National Family Health Survey, India, 2005-06 (N = 18,519) (Continued)

Model 4 Model 5

b SE b SE

Fixed effects: Explanatory variables Level 1

Son preference 0.092* 0.040 0.083* 0.041 Female status:

Mother’s education –0.013* 0.005 Mother’s work 0.003 0.040

Level 2 Region (ref = North):

South –0.014 0.049 Other region –0.016 0.045

Community: Pct. son preference 0.005** 0.001 0.005** 0.002 Pct. female literate –0.012** 0.001 –0.011** 0.001 Pct. female labor force 0.002 0.001 0.002 0.001

Control variables Level 1

Scheduled caste/tribe –0.045 0.040 –0.053 0.041 Standard of living (ref = Low):

Medium –0.091* 0.046 –0.090 0.046 High –0.232** 0.052 –0.204** 0.055

Religion (ref = other religion): Hindu 0.051 0.056 0.036 0.058 Muslim –0.041 0.073 –0.059 0.075

Mother’s age 0.015** 0.003 0.015** 0.003 Number of children –0.000 0.012 –0.006 0.013 Sibling age difference 0.611** 0.004 0.610** 0.004

Level 2 Urban 0.023 0.043 0.030 0.043

Random effects Intercept 0.073** 0.023 0.074** 0.023

–2 log likelihood 83,180 83,196

*p < .05; **p < .01.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

82 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

differential in education among siblings. The older the brother is than his sister, the more education he has compared to his sister. Urban families exhibit less gen- der discrimination against female children as compared to rural families in terms of their children’s education.

Models 2 through 4 add the other explanatory variables to Model 1 in order to examine the degree to which the impact of son preference on gender disparity in education can be explained by differences in individual-level women’s status, region, and community-level women’s status. Model 2 adds mother’s education and work status to the baseline model. As expected, mother’s education is nega- tively associated with gender discrimination against daughters. For every addi- tional year of mother’s education, the gender differential in children’s education is lowered by .03 year. Thus, children of a high school educated mother exhibit a gender differential that is two-fifths of a year less (–.396 = –.033 × 12) than that for children of an illiterate mother. However, the coefficient for mother’s work is not statistically significant. Importantly, controlling for these two variables related to women’s status diminishes the difference in gender disparity between siblings with and without maternal son preference. The coefficient for son preference is reduced by 13 percent when maternal female status is held constant [.13 = (.190– .165) / .190]. Thus, it appears that part of the reason why maternal son prefer- ence increases the gender disparity in education is that it is associated with lower women’s status.

Model 3 evaluates the hypothesis that regional differences can explain the dif- ference in gender disparity among children with and without maternal son pref- erence. This model adds to the baseline model (Model 1) two dummy variables reflecting regional differences—south and other regions (north is the reference cat- egory). The fixed effects show that families in both the south and the other regions exhibit less gender disparity in education among siblings compared to families in the north. Specifically, families in the south show a reduction in educational dif- ference by 0.21 year, while families in the other regions show a reduction by 0.17 year, in comparison to the educational difference in the north. Interestingly, these regional differences also appear to explain a portion of the difference in gender disparity in families with and without maternal son preference. Controlling for region reduces the coefficient for son preference by 12 percent [.12 = (.190–.168) / .190]. Thus, one reason why families with maternal son preference exhibit greater gender discrimination against their daughters lies in the fact that they are located in regions where women are less valued in general.

Model 4 considers the effect of community-level indicators on gender discrimina- tion in education and its effect on the impact of maternal son preference on gender discrimination. This model adds two community-level indicators of women’s sta- tus—percent of literate females and percent of working females—as well as the com- munity-level son preference. The results are mostly consistent with my expectation that communities with higher women’s status and lower son preference exhibit less gender discrimination in education among brothers and sisters. As expected, families in communities with higher percentages of women reporting son preference exhibit greater gender disparity in education among siblings. Additionally, families in com- munities with higher percentages of literate women show a lower educational dif- ference among siblings, indicating less gender discrimination in these communities.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 83

One way of quantifying this impact is to compare the average community with and without maternal son preference. In communities where 65 percent of the adult women are literate (as in the case of the average community without maternal son preference), the male advantage in years of education is reduced by four-fifths of a year (.78 = .012 × 65). On the other hand, the average community-level female literacy for families with maternal son preference is 53 percent, meaning that the male advantage is lowered by three-fifths of a year (.64 = .012 × 53). Thus, the female disadvantage in the average community with son preference is one-fifth of a year greater than that in the average community without son preference, simply as a result of the effect of lower adult female literacy in those communities.

The other indicator of community-level women’s status, the percent of work- ing women in a community, does not have a statistically significant impact on the gender differential in education. It is noteworthy that controlling for contextual women’s status explains the largest portion of the impact of maternal son prefer- ence on educational differential among opposite sex siblings. Controlling for these contextual characteristics reduces the coefficient for maternal son preference from .190 to .092, a reduction of 52 percent [.52 = (.190 – .092) / .190]. Thus, an important reason why maternal son preference impacts gender disparity in children’s educa- tion is because these families are more likely to be located in villages or neighbor- hoods with low female literacy and high son preference.

The final model, Model 5, includes all of the independent variables and is the full main effects model. The fixed effects show that maternal son preference and education, as well as context, affect the educational difference between brothers and sisters. Daughters are at a greater disadvantage when their mothers exhibit son preference. As expected, educated mothers exhibit less gender discrimination, showing that greater women’s status is associated with lower gender differential in children’s education.

Model 5 also shows that context has some important direct effects on gender discrimination in education within families. Consistent with expectations, com- munities with higher women’s status exhibit less gender discrimination in educa- tion among brothers and sisters. Specifically, families in communities with higher percentages of literate women and lower percentages of women with stated son preference show lower gender gap in children’s education. For example, in com- munities where 65 percent of the adult females are literate, the male advantage in years of education is reduced by .72 year (.72 = .011 × 65).

Importantly, in this complete model, the gender differential in education among children with and without maternal son preference is reduced from .190 to .083, a reduction of 56 percent [.56 = (.190–.083) / .190]. As was apparent from the earlier models, a great deal of this reduction is affected by context, particularly region and community-level women’s status. Maternal education also has a modest impact as a mediating variable. It appears that much of the difference in treatment of opposite sex siblings in families with and without son preference is a result of differences in maternal education, as well as the families’ location in regions and communities with lower women’s status, rather than as a result of stated son pref- erence per se.

The preceding discussion suggests that women’s education is an important pathway for providing more egalitarian treatment of opposite sex children.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

84 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

However, it is possible that the impact of maternal education varies by context. For example, as argued earlier, the impact of individual-level education may be different in a community with high female literacy versus a community with low female literacy. Strong community norms regarding women’s education could make a mother invest in her daughter ’s education regardless of her own educational level. Thus, the effect of maternal education should be weaker in regions and locations where women’s education is normative. In order to fully understand the effects of women’s education, both at the individual and com- munity levels, it is necessary to explore the possible interaction effects of contex- tual variables on individual-level women’s education (see Table 3). The results from Models 1 and 2 in Table 3 show that the effect of mother ’s education on the gender differential in children’s education varies by context. In particular, the negative effect of mother ’s education on the differential in education among brother and sister (i.e., lower gender discrimination) is weaker in regions and communities with higher women’s status. The preceding results show that the greater the mother ’s education, the lower the brother ’s advantage in educa- tion over his sister. However, in communities with greater percentages of liter- ate females, or the south and other regions (compared to the north), this female advantage is weakened. Results from Model 1 show that the effect of mother ’s education as a source of lower gender discrimination against daughters is weak- est in areas where adult female education is high. For example, if we were to compare communities with 25 percent literate females up to 60 percent literate females, a mother with secondary education (10 years) would have the following effect depending on her location:

Percent female literacy: 25%: –.125×10 + (0.002×10×25) = –0.75 50%: –.125×10 + (0.002×10×50) = –0.25 60%: –.125×10 + (0.002×10×60) = –0.05

In other words, in a community with only 25 percent female literacy, daugh- ters of secondary school educated mothers have reduced their educational disadvantage by almost three-quarters of a year compared to their brothers at school. However, this female advantage dwindles as community-level edu- cation increases resulting in almost zero impact on the gender differential in communities with 60 percent literate females. Thus, as expected, the effect of mother ’s education as a source of lower gender discrimination against daugh- ters is the strongest in communities where adult female education is low. Alter- natively, the effect of mother ’s education is weak in communities with high female education as the community effect overwhelms the individual effect in these communities.

Similar effects can be seen from Model 2, where the negative effect of mother’s education on the gender differential is stronger in the north compared to all other regions. For example, results from Model 2 show that daughters of high school educated mothers (compared to those with illiterate mothers) in the north have reduced their educational disadvantage by almost a third of a year over their brothers (–.026 × 12 = –.31), while in the south there is almost no reduction in the gender differential [(–.026 × 12)+(.028 × 12) = .024]. Thus, in the north, where

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 85

female status and education is lower, individual-level maternal education has a strong impact, leading to a gender differential that is more advantageous for females among children of educated mothers.

TABLE 3 Multilevel Regression Analysis of Gender Disparity in Education with Cross-Level

Interaction Effects, National Family Health Survey, India, 2005–06 (N = 18,519)

Model 1 Model 2

b SE b SE

Fixed effects: Explanatory variables Level 1

Son preference 0.083* 0.041 0.081* 0.041 Female status:

Mother’s education –0.125** 0.017 –0.026** 0.006 Mother’s work –0.003 0.040 0.005 0.040

Level 2 Region (ref = North):

South –0.016 0.049 –0.131* 0.064 Other region –0.000 0.045 –0.081 0.055

Community: Pct. son preference 0.005** 0.002 0.004** 0.002 Pct. female literate –0.014** 0.001 –0.011** 0.001 Pct. female labor force 0.002 0.001 0.002 0.001

Control variables Level 1

Scheduled caste/tribe –0.054 0.041 –0.060 0.041 Standard of living (ref = low):

Medium –0.053 0.046 –0.098* 0.046 High –0.173** 0.055 –0.213** 0.055

Religion (ref = other religion): Hindu 0.053 0.058 0.042 0.058 Muslim –0.048 0.075 –0.061 0.075

Mother’s age 0.012** 0.003 0.015** 0.003 Number of children –0.003 0.013 –0.007 0.013 Sibling age difference 0.610** 0.004 0.610** 0.004

Level 2 Urban 0.003 0.043 0.031 0.043

Interaction of maternal education with context Community

Pct. female lit 0.002** 0.000 Region

South 0.028** 0.010 Other region 0.021* 0.010

Random effects: Intercept 0.063** 0.022 0.071** 0.023

–2 log likelihood 83,163 83,200

*p < .05; **p < .01.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

86 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

DISCUSSION AND CONCLUSION

A major long-term consequence of son preference in India is the disparate qual- ity of life and opportunities provided to sons compared to daughters. When a daughter is considered a temporary member of her natal family, it is reasonable to assume that resources will be disproportionately allocated to the favored sons in the long-term interests of the family. Among the many areas of possible dis- crimination against daughters, educational discrimination has a particularly strong impact on future life chances. Education improves a woman’s life, not just by giving her more tools for economic self-sufficiency, but also by improving her status within her household and the larger society. Thus, understanding why edu- cational discrimination persists among the youth in India is key to understanding and improving women’s lives.

While previous research on gender discrimination in education has neglected the effect of context on parents’ attitudes and decision-making, this study shows that it is crucial to understanding why there is a persistent bias against daugh- ters’ education even as educational opportunities are expanding for all in India. In particular, my research provides new insight into the intersection of individual and community characteristics. The results confirm that not only is the level of women’s status in communities and regions an important predictor of gender dis- crimination in education, it also modifies the way that individual-level maternal characteristics impact such discrimination.

A comparison of siblings within families shows that a sister is at a disadvantage in terms of years of education when compared to her brother. More importantly perhaps, the multivariate analysis confirms that maternal son preference has a sig- nificantly positive impact on the gender differential in education. In families with maternal son preference, the female disadvantage in education is almost one-fifth of a year greater than in families without maternal son preference, even after con- trolling for class, religion, sibling age difference, and number of children. Thus, consistent with expectations, families that exhibit son preference appear to expend more resources on their sons’ education than daughters’ education. This result is even more noteworthy considering that the sample consists of children below the age of 18, and therefore, I am only investigating educational opportunities pro- vided through high school or below. It is likely that the discrepancy between male and female education increases as the children reach adulthood.

A key finding of this study is that much of the impact of son preference on gender differential in education can be explained by differences in adult women’s status and context between families with and without son preference. More spe- cifically, my analysis suggests that children whose mothers exhibit son preference are also more likely to have mothers with lower levels of education, and be located in regions and communities with lower women’s status. These individual and community characteristics impact the gender differential in education over and above son preference. For example, not only do mothers’ education and commu- nity-level women’s status have a weakening effect on the gender bias, controlling for these variables reduces the coefficient for son preference by more than half. Thus, in addition to the direct effects of these variables, they also exhibit indirect effects by reducing the impact of son preference on the gender differential.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 87

The importance of female education at the individual and community level is particularly clear from this analysis. Mothers who are educated themselves are likely to display more egalitarian views regarding gender and less likely to exhibit son preference. As a result, they are less likely to discriminate against their daugh- ters. While this is consistent with previous research, my analysis shows that there are additional important effects of community-level women’s education. Higher levels of female literacy in communities indicate that women’s education is more normative in these communities, leading to more egalitarian treatment of sons and daughters. My analysis shows that children without maternal son preference are more likely to reside in communities with higher rates of female literacy. Resi- dence in these communities allows young girls to more easily follow the commu- nity norm and become educated, regardless of their mother’s educational level.

The effect of context on son preference is also evident in regional differences among families with and without son preference. Son preference is more prevalent in the north compared to the rest of India, perhaps as a result of regional cultural differences that place a lower value on women and their natal connections in these areas. These regional differences in son preference carry on to impact the gender differential in education among children. Opposite sex siblings without maternal son preference are more likely to reside in the south and the other regions and, as a result, exhibit more gender egalitarianism with regard to their years of education.

My research also shows how individual and community characteristics intersect to impact the gender differential in children’s education. Specifically, the results show that the effect of mother’s education is stronger in communities with lower women’s education and in regions with lower women’s status. This finding has important theoretical implications as it shows that educated mothers in commu- nities with low women’s status are able to withstand community norms against female education and provide more equal treatment for their daughters compared to their sons.

This study provides the initial stages of understanding the persistent bias in education against daughters in India. At a policy level, this research confirms the need for adult education and literacy drives in communities, particularly those targeting women. However, there are many possible avenues for future research that may be taken in order to expand this study as well as to gain a deeper under- standing of the mechanisms that affect gender discrimination among children in India. First, my analysis of education is based solely on quantity and not quality of education. For example, it is possible that boys may be sent to better schools or receive more time off to do homework. Future research on educational discrimi- nation should consider the resources spent on educating males versus females. Parents may be more likely to spend additional resources on a son’s education than a daughter’s because they consider it an investment in the future. For girls in upper or middle-class families in particular, there may be adequate resources to let them study up to and beyond high school. But this does not mean that they attend the same quality school as their brothers or that they receive the same amount of support for their education. Anecdotal evidence suggests that Indian parents have been known to spend a lot more money on getting competent tutors for their sons while allowing their daughters to fend for themselves.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

88 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

Additionally, there needs to be research that investigates more definitively the reasons behind some of the findings here. Among the findings from this study, it appears that one of the most important conclusions that can be drawn has to do with adult female education as a precursor to the lowering of gender bias in education against female children. It is clear that mothers with low or no educa- tion tend to invest less in their daughter’s than their son’s education, leading to a disadvantage that is perpetuated over generations. On the other hand, women with some education tend to invest more in their daughter’s education, indicating that they value their daughters more. One possible avenue for future research is to try to find out more about the motivations for this greater investment in daughters of educated mothers. For example, is it because she feels that a more educated daughter would be able to contribute more to the family economically, or is it because as an educated mother, she has undergone a cultural shift in perspective? It is possible that the effect is partially because an educated mother has a greater understanding of how education can make a woman better in her traditional roles as wife and mother by allowing her to keep better household accounts and super- vise her children’s education at home (rather than pay for a tutor as an uneducated mother would have to do). A study that focuses on parental aspirations could help in our understanding of this issue and could contribute greatly to enhancing opportunities for future generations of female children in India.

REFERENCES

Afridi, Farzana. 2010. “Women’s Empowerment and the Goal of Parity between the Sexes in Schooling in India.” Population Studies 64(2):131–45.

Amin, Sajeda. 1996. “Female Education and Fertility in Bangladesh: The Influence of Mar- riage and the Family.” Pp. 184–204 in Girl’s Schooling, Women’s Autonomy and Fertility Change in South Asia, edited by R. Jeffery and A. M. Basu. New Delhi: Sage Publications.

Arnold, Fred. 1997. Gender Preferences for Children, DHS Comparative Studies No. 23. Calver- ton, MD: Macro International Inc.

———. 2001. “Son Preference in South Asia.” Pp. 281–99 in Fertility Transition in South Asia, edited by Z. A. Sathar and J. F. Phillips. New York: Oxford University Press.

Banerjee, Nirmala. 1998. “Household Dynamics and Women in a Changing Economy.” Pp. 245–63 in Gender, Population and Development, edited by M. Krishnaraj, R. M. Sudar- shan, and A. Shariff. India: Oxford University Press.

Basu, Alaka M. 1992. Culture, the Status of Women, and Demographic Behaviour, Illustrated with the Case of India. Oxford, UK: Clarendon Press.

Beteille, Andre. 1974. Studies in Agrarian Social Structure. New York: Oxford University Press. Bhargava, P. K. 2002. “Changes in Disparity in Literacy Rate during 1991-2001.” Paper pre-

sented at the seminar on Progress of Literacy in India: What the Census 2001 Reveals, New Delhi, India.

Bose, Sunita. 2011. “The Effect of Women’s Status and Community on the Gender Differential in Children’s Nutrition in India.” Journal of Biosocial Science 43(5):513–33.

Census of India. 2001. Primary Census Abstract. Delhi: Census of India. Das, Raju J. 2001. “The Spatiality of Social Relations: An Indian Case-Study.” Journal of Rural

Studies 17(3):347–62. Das Gupta, Monica. 1987. “Selective Discrimination against Female Children in Rural Pun-

jab, India.” Population and Development Review 13(2):77–100.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 89

———. 1995. “Life Course Perspectives on Women’s Autonomy and Health Outcomes.” American Anthropology 97(3):481–91.

———. 1998. Missing Girls in China, South Korea and India: Causes and Policy Implications. Cambridge, MA: Harvard Center for Population and Development Studies.

Deolalikar, Anil B. 1993. “Gender Differences in Returns to Schooling and in School Enrol- ment Rates in Indonesia.” Journal of Human Resources 28(4):899–932.

Desai, Sonalde. 1998. “Engendering Population Policy.” Pp. 44–69 in Gender, Population and Development, edited by M. Krishnaraj, R. M. Sudarshan, and A. Shariff. India: Oxford University Press.

Dhruvarajan, Vanaja. 1989. Hindu Women & the Power of Ideology. Granby, MA: Bergin & Garvey Publishers, Inc.

Dreze, Jean and Amartya Sen. 1995. India: Economic Development and Social Opportunity. Oxford, UK: Oxford University Press.

Dube, Leela. 1997. Women and Kinship: Comparative Perspectives on Gender in South and South- East Asia. New York: United Nations University Press.

Dumont, Louis. 1970. Homo Hierarchicus. Chicago and London: The University of Chicago Press. ———. 1980. Religion/Politics and History in India. Paris/The Hague: Mouton Publishers. Dyson, Tim and Mick Moore. 1983. “On Kinship Structure, Female Autonomy, and Demo-

graphic Behaviour in India.” Population and Development Review 9(1):35–60. Etzioni, Amitai. 1996. “The Responsive Community: A Communitarian Perspective.” Ameri-

can Sociological Review 61(1):1–11. Ghosh, Shanti. 1995. “Integrated Health of the Girl Child.” Social Change 25(2&3):44–54. Hossain, Mohammad A. and Clement A. Tisdell. 2005. “Closing the Gender Gap in Ban-

gladesh: Inequality in Education, Employment and Earnings.” International Journal of Social Economics 32(5):439–53.

Husain, Zakir and Swagata Sarkar. 2011. “Gender Disparities in Educational Trajectories in India: Do Females Become More Robust at Higher Levels?” Social Indicators Research 101(1):37–56.

Jeffery, Patricia and Roger Jeffery. 1994. “Killing My Heart’s Desire: Education and Female Autonomy in Rural North India.” Pp. 125–71 in Women as Subjects South Asian Histories, edited by N. Kumar. Charlottesville and London: University Press of Virginia.

——— and Roger Jeffery.1996. “What’s the Benefit of Being Educated? Girls’ Schooling, Women’s Autonomy and Fertility Outcomes in Bijnor.” Pp. 150–83 in Girls’ Schooling, Women’s Autonomy and Fertility Change in South Asia, edited by R. Jeffery and A. M. Basu. New Delhi, Thousand Oaks, London: Sage Publications.

Joshi, Sharmila. 1998. “India’s ‘Nowhere’ Girls. Voices of Girls 1: India.” People and the Planet 7(3):26–28.

Kabeer, Naila. 2000. “Inter-Generational Contracts, Demographic Transitions, and the ‘Quan- tity-Quality’ Trade Off: Parents, Children, and Investing in the Future.” Journal of Inter- national Development 12(4):463–82.

Karve, Irawati. 1965. Kinship Organization in India, 2nd edition. Bombay, India: Asia Publish- ing House.

Kingdon, Geeta G. 2002. “The Gender Gap in Educational Attainment in India: How Much Can Be Explained?” Journal of Developmental Studies 39(2):25–53.

———. 2005. “Where Has All the Bias Gone? Detecting Gender Bias in the Intrahousehold Allocation of Educational Expenditure.” Economic Development and Cultural Change 53(2):409–51.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

90 SOCIOLOGICAL PERSPECTIVES Volume 55, Number 1, 2012

Kishor, Sunita. 1995. “Gender Differentials in Child Mortality: A Review of the Evidence.” Pp. 19–54 in Women’s Health in India Risk and Vulnerability, edited by M. Das Gupta, L. C. Chen, and T. N. Krishnan. India: Oxford University Press.

Kravdal, Oystein. 2004. “Child Mortality in India. The Community Level Effect of Educa- tion.” Population Studies 58(2):177–92.

Malhotra, Anju, Reeve Vanneman, and Sunita Kishor. 1995. “Fertility, Dimensions of Patri- archy, and Development in India.” Population and Development Review 21(2):281–305.

Miller, Barbara D. 1981. The Endangered Sex. Ithaca, NY/London: Cornell University Press. ———. 1997. “Social Class, Gender and Intrahousehold Food Allocations to Children in

South Asia.” Social Science and Medicine 44(11):1685–95. Mines, Mattison. 1992. “Individuality and Achievement in South Indian Social History.”

Modern Asian Studies 26(1):129–56. Mukhopadhyay, Swapna and Surekha Garimella. 1998. “The Contours of Reproductive

Choice for Poor Women: Findings from a Micro Survey.” Pp. 98–121 in Women’s Health, Public Policy and Community Action, edited by S. Mukhopadhyay. New Delhi: Manohar.

Nag, Moni. 1991. “Sex Preference in Bangladesh, India and Pakistan, and Its Effect on Fertil- ity.” Demography India 20(2):163–85.

Nambissan, Geetha B. 2005. “Integrating Gender Concerns.” Changing English 12(2):191–99. National Family Health Survey (NFHS-3) 2005-06. 2007. Mumbai, India: International Institute

for Population Sciences (IIPS) and ORC Macro. Obasi, Emma. 1997. “Structural Adjustment and Gender Access to Education in Nigeria.”

Gender and Education 9(2):161–77. Oster, Emily. 2009. “Proximate Sources of Population Imbalance in India.” Demography

46(2):325–39. Pande, Rohini P. 2003. “Selective Gender Differences in Childhood Nutrition and Immuniza-

tion in Rural India: The Role of Siblings.” Demography 40(3):395–418. Parashar, Sangeeta. 2005. “Moving Beyond the Mother-Child Dyad: Women’s Education,

Child Immunization, and the Importance of Context in Rural India.” Social Science and Medicine 61:989–1000.

Phillip, Mini and Kathakali S. Bagchi. 1995. The Endangered Half. New Delhi: Upalabdhi, Trust for Development Initiatives.

Raudenbusch, Stephen W. and Anthony S. Bryk. 2002. Hierarchical Linear Models. Thousand Oaks, London, Delhi: Sage Publications.

Satia, J. K. and Shireen J. Jejeebhoy, Eds. 1991. The Demographic Challenge: A Study of Four Large Indian States. Bombay, India: Oxford University Press.

Sen, Amartya K. 1990. “Gender and Cooperative Conflicts.” Pp. 123–49 in Persistent Inequali- ties, edited by I. Tinker. New York: Oxford University Press.

Somjee, Geeta. 1989. Narrowing the Gender Gap. Oxford, UK: Macmillan Press in association with the Centre for Cross-Cultural Research on Women, Queen Elizabeth House.

Sopher, David E. 1980. “The Geographical Patterning of Culture in India.” Pp. 289–326 in An Exploration of India: Geographical Perspectives on Society and Culture, edited by D. E. Sopher. Ithaca, NY: Cornell University Press.

Srinivas, Mysore N. 1984. Some Reflections on Dowry. Delhi, India: Oxford University Press. ———. 1998. Indian Society through Personal Writings. Delhi, Calcutta, Chennai, Mumbai:

Oxford University Press. Unni, Jeemol. 1998. “Gender Differentials in Schooling.” Pp. 141–58 in Gender, Population and

Development, edited by M. Krishnaraj, R. M. Sudarshan, and A. Shariff. India: Oxford University Press.

Vadnais, Daniel, Adrienne Kols, and Noureddine Abderrahim. 2006. Women’s Lives and Expe- riences: Changes in the Past 10 Years. Washington, DC: ORC Macro/USAID.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms

A Contextual Analysis of Gender Disparity in Education in India 91

Velkoff, Victoria A. 1998. Women of the World Women’s Education in India. Washington, DC: U.S. Bureau of the Census, WID/98-1.

——— and Arjun Adlakha. 1998. Women of the World Women’s Health in India. Washington, DC: U.S. Bureau of the Census, WID/98-3.

Vlassoff, Carol. 1996. “Against the Odds: The Changing Impact of Schooling on Female Autonomy and Fertility in an Indian Village.” Pp. 218–34 in Girls’ Schooling, Women’s Autonomy and Fertility Change in South Asia, edited by R. Jeffery and A. M. Basu. New Delhi, Thousand Oaks, London: Sage Publications.

Worku, Yelfign. 2001. “Ethiopia: From Bottom to Top in Higher Education-Gender Role Prob- lems.” International Journal of Sociology and Social Policy 21(1/2):98–104.

This content downloaded from ������������134.197.153.42 on Tue, 03 Mar 2020 21:46:03 UTC�������������

All use subject to https://about.jstor.org/terms