The Relationship Between a Mother’s Attitude Toward Domestic Violence and Children’s Schooling Outcomes in Turkey

Abstract

This study explores the relationship between a mother’s attitude toward domestic violence and her children’s schooling outcomes in Turkey. The sample is drawn from the 2003 Turkish Demographic and Health Survey and consists of 7,951 children within the ages of legally mandated compulsory education. A probit regression model is used to analyze the data. The results suggest that the daughters of mothers who find domestic violence acceptable are 2.6 percentage point less likely to enroll in school, all else being equal, than the daughters of mothers who do not tolerate abuse. In comparison, the schooling outcomes of male children are not statistically sensitive to the mother’s attitude toward wife beating. The policy implications of the results are discussed.

Keywords

wife beating domestic violence education gender turkey

Introduction

Violence against women in their homes is a reflection of the exercise of power in a conjugal relationship and can encompass physical, emotional, and economic means of attaining and sustaining control by one spouse over the other. Most frequently exerted by the man, domestic violence not only affects the woman’s physical and emotional well-being but also hampers the cognitive and physical development of her children (Diop-Sidibé, Campbell, & Becker, 2006; Evans, Davies, & Dilillo, 2008; Holt, Buckley, & Whelan, 2008; Sousa et al., 2011). Recent empirical studies have identified important patterns in the incidence of domestic violence against women. Among those factors identified, previous exposure to violence, particularly in the natal families of both husband and wife, contributes to its exercise and acceptance (Abramsky et al., 2011; Dhaher, Mikolajczyk, Maxwell, & Krämer, 2010; Linos, Khawaja, & Al-Nsour, 2010; Ozcakir, Bayram, Ergin, Selimoglu, & Bilgel, 2008). Weak female bargaining power in the family, by which I mean a woman’s relative power over household decision making vis-à-vis her spouse, can intensify the risk of violence and internalization of the norms that sanction wife beating contributes to its persistence (Haj-Yahia, 2002; Taft, Bryant-Davis, Woodward, Tillman, & Torres, 2009; Yoshioka, DiNoia, & Ullah, 2001). In some contexts, direct state intervention to eliminate wife beating has been successful, but in other cases has failed (Altinay & Arat, 2009; UNWomen, 2012). Improvement in a woman’s fallback options outside of marriage can reinforce her bargaining power but in other contexts, improving women’s economic independence has led to increased domestic violence. For example, women’s employment has had an ambiguous relationship to domestic violence, while in some cases employment had led to the empowerment of women, in other contexts it is shown to be positively associated with violence (Agarwal & Panda, 2007; Krishnan et al., 2010; Simister & Mehta, 2010; Villarreal, 2007). Despite these variations, we know that addressing social norms that regard violence against women as permissible can reduce the incidence of wife beating and can help to improve the well-being of both women and children.

In this article, I explore the relationship between a mother’s acceptance of physical violence from her husband and investment in the next generation. I argue that a mother’s attitude toward wife beating is an indication of her perception of her own agency and is therefore linked to the well-being outcomes of other members of the household within a complex web of power relationships. Understanding the spillover effects of individual and social tolerance of wife beating is more than a scholarly puzzle; it is essential to the design of policy interventions to improve the well-being of mothers and children. Using data from Turkey, I seek to empirically understand whether the schooling outcomes are affected only by income and opportunity constraints or also by the transmission of social norms from a mother to her children. As we shall see, in the Turkish context, a mother’s attitude toward wife beating is closely associated with her daughter’s schooling outcome but not that of her sons.

Literature Review on Attitudes Toward Wife Beating and Children’s Schooling

There is a wide and diverse body of literature examining the attitudes toward domestic violence, as well as a wealth of studies examining the factors influencing children’s schooling outcomes in developing countries. None of these studies, however, connect the attitudes toward wife abuse and children’s educational opportunities within the locus of the household.

Studies exploring male attitudes have found patriarchal expectations, his partner’s contraceptive use without permission, and alcohol abuse to be markers associated with the justification of wife beating (Haj-Yahia, 1997; Johnson & Das, 2009; Lawoko, 2008; Marshall & Furr, 2010; Ogunjuyigbe, Akinlo, & Ebigbola, 2005; Speizer, 2010; Stuart et al., 2003; Zhu & Dalal, 2010). Couple-level factors may also contribute to spousal violence. In the Vietnamese context, the association between marital violence and husbands’ gender attitudes is found to be contingent on the level of women’s expectations of equity in the relationship (Luke, Schuler, Mai, & Vu Thien, 2007). Across cultures, the common predictors for women who tolerate physical abuse are their level of education, their employment status, whether they have a history of witnessing violence in their natal family, and family size (Dhaher et al., 2010; Krishnan et al., 2010; Rocca, Rathod, Falle, Pande, & Krishnan, 2009). A number of studies suggest that cultural norms play an important role in shaping tolerance to wife beating (Haj-Yahia, 2002; Hindin, 2003; Jewkes, 2002; Linos et al., 2010). Studies of efforts to understand spousal violence indicate that changes in behavior are especially difficult in cases in which the women are in agreement with the social norms and systems which support male privilege (Obeid, Chang, & Ginges, 2010; Rani, Bonu, & Diop-Sidibe, 2004; Yoshioka et al., 2001). A social or cultural tolerance to wife beating is reported to increase the likelihood of intimate partner violence (Abramsky et al., 2011; Flood & Pease, 2009). Studies also indicate that women who have been subject to violence themselves are more likely to justify wife beating (Boy & Kulczycki, 2008; Khawaja, Linos, & El-Roueiheb, 2007). In these cases, the argument follows that the primary target of efforts to prevent violence should be the social norms that sanction the use of violence against individuals who transgress traditional gender roles. This is a critical point, especially in situations in which younger people tend to support wife beating. The intergenerational transmission of social norms which shape the prevalence of gender-based violence make a compelling case for focusing on shifting perceptions of what is socially acceptable (Lunkenheimer, Kittler, Olson, & Kleinberg, 2006; Rani & Bonu, 2009).

The scholarly literature on children’s schooling has identified income and opportunity constraints as playing a decisive role in parental decisions on investment in the next generation. Previous research has focused on poverty as the main factor in low levels of schooling (Alcaraz, Chiquiar, & Salcedo, 2012; Lloyd, Mete, & Grant, 2009; Zhao & Glewwe, 2010). The distance to, price of, and quality differences within schools explain schooling decisions, with girls’ schooling likely to be more responsive to price incentives (Glick, 2008). Other factors, such as care obligations within the household and competition for economic resources among siblings, are also discussed in the literature (Dayioglu, Kirdar, & Tansel, 2009; Marazyan, 2011; Rose & Al Samarrai, 2001). However, an important trigger in increasing children’s level of education has been found in the mother’s level of empowerment and her role in household decision making (DeGraff & Levison, 2009; Schuler & Rottach, 2010).

It is possible to derive two sets of results from this literature: For women, the acceptance of domestic violence may be linked to a perception of women’s agency as weak. At the same time, improving child well-being is at least in part dependent on women’s empowerment. Examining the effect of a mother’s attitude toward wife beating to family investment in the next generation provides an opportunity to better understand the link between parental commitment to gender equality and children’s schooling outcomes. This study attempts to fill this gap using Turkey as a case study.

Conceptual Model and Context

Conceptual Model

Little work has been done on the relationship between a mother’s tolerance for wife beating and her children’s schooling outcomes. In what ways can we expect the attitude of a mother toward spousal violence to affect her children’s schooling outcomes? The extent to which the mother’s toleration of wife beating reflects her perceived or actual lack of bargaining power vis-à-vis her husband may be reflected in a failure on the part of her children to recognize schooling as a path to a fulfilling life in adulthood. However, it is also possible that children could react with the opposite conclusion and view education as a means of breaking free from the patterns of their natal families. Thus, the conceptual framework supports a multi-level examination: In the first stage, I examine empirically whether a mother’s attitude toward wife beating exerts any statistically significant influence on her children’s schooling outcomes, thus demonstrating a spillover effect of attitudes toward wife beating within the family. I then ask whether such influence differs according to the gender of the child, reflecting the complexity of gendered and intergenerational relationships within a household.

Understanding Turkish Context

Comparable and timely national statistics on the prevalence of marital violence are rare and Turkey is not an exception. Those studies on the incidence of marital violence which are available show a wide variation in its prevalence, ranging from 19.5% to 72% (Akar, Aksakal, Demirel, Durukan, & Özkan, 2010; Marshall & Furr, 2010; Mayda & Akkuş, 2005; UN Department of Economic and Social Affairs, 2010). The factors that are associated with marital violence run from the employment opportunities available to women to childhood exposure to violence in the home (Ergin, Bayram, Alper, Selimoglu, & Bilgel, 2005; Kocacik, 2007; Ozcakir et al., 2008). A recent comprehensive and detailed study that examined violence against women in Turkey found that the overwhelming majority of the Turkish women interviewed in the study do not justify wife beating or conceptualize spousal violence as a family affair (Altinay & Arat, 2009). The authors report no statistical differences in the attitudes of the participants toward violence after controlling for regional differences in income and education.

Turkey has shown steady improvement in the gender balance in basic schooling, but a noticeable gap remains. The latest official statistics indicate the female-to-male ratios in school enrollment are 0.98 and 0.88 for primary and secondary schooling, respectively. A decade earlier, the female-to-male ratios in school enrollment were measured as 0.88 and 0.75 for primary and secondary schooling, respectively. Part of this persistent gender inequity is due to culture; children’s schooling is closely correlated with local practices of socialization (Bakirci, 2002). Household factors also contribute to the persistence of this pattern. Following a shock to household income, parents prefer to keep boys in school and send girls to paid market-based work if a child’s contribution to the household budget is needed (Dayioglu, 2006). Even when parents are willing to send their children to school, a lack of geographically accessible schools in the area may prevent them from doing so (Erturk & Dayioglu, 2004). A cross-cutting ethnic dimension has also been noted. Ethnic differences in levels of school enrollment are significant and that the gender gap is wider for Kurdish children than for Turkish children, after accounting for income constraints (Kirdar, 2009). Rural to urban migration is associated with increased school drop-outs among the migrating households (Goksen & Cemalcilar, 2010). Sibling composition, household wealth, and mother’s employment are also found among the factors that play a role in children’s educational attainment, although their effects vary with the gender of the child (Dayioglu et al., 2009).

Data

The data used in this study are drawn from the 2003 Turkish Demographic and Health Survey (DHS), Phase 4. The survey covers approximately 10,800 households with 48,000 individuals and collected information primarily on women’s fertility choices, attitudes toward wife beating and knowledge of HIV/AIDS, and women’s agency in household decision making. The participants are women of childbearing age, 15 to 49. Questions are also included on household demographic and educational characteristics, labor market attachment, and household assets. To understand the children’s educational outcomes, the sample of interest is drawn from children between 7 and 14 years of age, that is, those children within the ages of legally mandated compulsory education. There are 8,623 observations in this age range, including children without a father present in the household. In light of the possibility that a mother’s attitude toward wife beating may differ in households with and without a father, I excluded single-mother-headed households, 340 households, as well as additional observations with incomplete information, 332 households, from the sample. As a result, the final working sample consists of 7,951 children.

While the data provide rich information, for the purposes of this inquiry, the survey design has a few major drawbacks. First, because the primary concern of the survey is women’s fertility choices, the sample neither includes community characteristics, such as the location of schools, nor information on the costs of schooling to families. Second, only women between 15 and 49 years of age, women who are in their childbearing years, are included in the domestic violence module. The resulting sample is thus restricted to mothers in this age group and their children’s schooling outcomes. Second, household income is categorized by the primary income earner. No individual-level data on earned and unearned income or expenditures could be extracted from the survey, preventing the capture of full income effects on educational outcomes.

Descriptive Results

Perceptions About Wife Beating

The module on domestic violence includes a set of questions asking whether wife beating is justified when a wife (a) burns the food, (b) neglects her children, (c) argues with her husband, (d) wastes money, and finally (e) refuses intercourse. Thus, agreement with some or all of the listed reasons measures the acceptance of wife beating within the norms of social behavior and the belief in the culpability of the victim (e.g., she burns food) by failing to be a good housewife. Used extensively in other similar surveys, each attitude question measures the respondent’s level of tolerance for the “transgression of gender norms,” within a given gendered division of labor and behavioral norms (Rani et al., 2004). The distribution of responses agreeing with each of the reasons offered for wife beating is provided in Table 1. The results show that the survey participants believe that both submission to male authority and frugality play important roles in preventing wife abuse. Arguing with the husband is frequently reported as an acceptable reason for wife beating, suggesting that challenging the male dominant/female submissive social norm is considered to be a serious violation. Also, more than a third of the respondents agree that if a wife wastes money, she deserves to be beaten by the husband, which conforms to a male breadwinner/female dependent normative pattern. In comparison, the least accepted option was burning food: approximately 1 out of 10 respondents perceived burning food as sufficient cause to beat one’s wife. The relatively high level of tolerance of wife abuse for wasting money is also remarkable considering that less than one third of the participants within the sample are gainfully employed. The differences in acceptance levels suggest that the majority of women seem to be keenly aware of the husband’s contribution to household income, yet, in their own spheres as housewives, they are less likely to agree that the husband holds power to punish minor failings, such as burning food. Table 1 also summarizes the stark differences between mothers in rural and urban areas, and between literate and illiterate mothers, and indicates that mothers who live in rural areas or are illiterate are more tolerant toward wife beating.

Table 1.

Agreement for Reasons of Wife Beating.

	All Sample	Rural	Urban	Illiterate	Literate
	% Agree	% Agree	% Agree	% Agree	% Agree
1. Burns food	8.21	14.98	5.06	16.90	4.10
2. Neglects children	27.91	36.32	23.98	41.52	21.47
3. Argues with husband	35.24	50.57	28.10	54.05	26.35
4. Wastes money	34.05	50.17	26.55	54.10	24.58
5. Refuses intercourse	22.95	34.04	17.79	40.54	14.65
Average (1-5)	27.67	37.22	20.30	41.42	18.23

Note. A respondent may agree with more than 1 reason, so the total does not add to 100%.

Characteristics of Households

The sample statistics, summarized in Table 2, show that the age of the representative child is 10 and that the working sample includes slightly more number of boys, at 51%. On the average, the fathers are 5 years older than the mothers and the mean years of mothers’ schooling of is 2 years less than that of fathers, 4 and 6 years, respectively. In the majority of the households, the father is employed, although fathers working in the formal sector constitute only 47%, indicating the prevalence of informal labor sector in Turkey. The low rate of mothers’ labor force participation reflects the trend observed in Turkey, where female labor force participation dropped from 33% to 23% between 1988 and 2007 (Buğra & Yakut-Cakar, 2010). The average score of agreement with at least one of the reasons for wife beating is recorded as 1.28. In the sample, more than half of the parental marriages are arranged, 64%, and approximately one third of the mothers reported Turkish as their native language. Urban households are somewhat overrepresented in the sample, as only 30% of the households lived in rural areas. In addition, more than half of the households, 63%, own their own dwelling. The distribution of households across asset groups indicates that the representative household is placed within the third quintile. The regional distribution indicates that over half of the survey participants are located in the highly populated western and eastern regions of the country, 23% and 38%, respectively. The remaining sample is distributed across south (14%), central (15%), and northern (10%) regions. Table 2 also summarizes the primary variable of interest, the schooling status of children. Of the boys included in the sample, 95% attended school, while the figure for girls is 87%.

Table 2.

Sample Statistics (N = 7,951).

Variable	M	SD	Minimum	Maximum
Attending school (=1)	0.91	0.27	0	1
Attending school—Boys (= 1)	0.95	0.21	0	1
Attending school—Girls (= 1)	0.87	0.32	0	1
Gender (= 1 female)	0.49	0.50	0	1
Age	10.22	2.31	7	14
Mother’s age	34.92	5.79	22	49
Father’s age	39.24	6.22	24	66
No. of children 0-2	0.33	0.61	0	5
No. of children 3-6	0.64	0.88	0	6
No. of children 7-14	1.57	1.21	0	8
Mother tolerates wife beating	1.28	1.56	0	5
Father works (= 1)	0.86	0.35	0	1
Father’s work offers pension plan (= 1)	0.47	0.50	0	1
Mother works (= 1)	0.27	0.44	0	1
Arranged marriage (= 1)	0.64	0.47	0	1
Mother’s native language (= 1)	0.67	0.48	0	1
Mother’s years of schooling	4.03	3.62	0	19
Father’s years of schooling	6.63	3.81	0	19
Urban (= 1)	0.67	0.47	0	1
Household owns the dwelling (= 1)	0.63	0.48	0	1
Household asset wealth index quintile	2.89	1.42	1	5
Region	3.26	1.61	1	5
% Households (west = 1)	22.8	—	—	—
% Households (south = 1)	14.2	—	—	—
% Households (central = 1)	15.2	—	—	—
% Households (north = 1)	9.6	—	—	—
% Households (east = 1)	38.2	—	—	—

Although these statistics are informative, the survey data present additional information about the bargaining power of the mothers in the working sample. The respondents were asked about the specific features of their marriages and selected indicators drawn from the survey are summarized in Table 3. The average duration of marriage is 16.7 years. In only one third of the marriages did a civil ceremony take place before the religious ceremony, meaning that in two thirds of the marriages, the women had to wait to attain their legal rights and entitlements until the civil ceremony, since Civil Law does not recognize religious ceremony. The majority of wives lived with their in-laws at some point during the marriage, and more than half of the marriages were arranged. The native language of approximately two thirds of the mothers is Turkish, with Kurdish as the second most reported native language. The generational inequities in educational attainment are reported in Table 3 and the average literacy rates of the parents of the respondents are 26% versus 68%, for mothers and fathers respectively.

Table 3.

Marriage Characteristics of Mothers.

Variable	M	SD
Duration of marriage	16.66	5.85
Arranged marriage (= 1)	0.64	0.48
Married first with civil ceremony (= 1)	0.33	0.47
Months between religious and civil ceremony	5.48	5.13
Mother’s native language (= 1 Turkish)	0.65	0.48
Mother’s mother literate (= 1)	0.26	0.44
Mother’s father literate (= 1)	0.68	0.47
Mother lived with in-laws (= 1)	0.70	0.46
Mother lived with own parents (= 1)	0.00	0.04

While survey participants were not asked their individual level of earnings, a question was included on the main source of household income. Table 4 shows that for approximately 79% of the households, the husband’s earnings are the main source of household income. In less than 10% of the households, the joint earnings of husband and wife are classified as the household’s primary source of income. For a non-negligible part of the sampled households, 5.9%, the husband’s family appears as the main source of income. When compared with the very low percentage of households in which the wife’s earnings are the primary source of household income, 1.6%, the sample statistics suggest that Turkish households are more likely to receive help from the husband’s family than for wives to seek employment, should there be a financial need.

Table 4.

Primary Source of Household Income.

Source of Income	Percent
Husband’s earnings	78.85
Joint earnings	9.19
Wife’s earnings	1.63
Husband’s family	5.92
Wife’s family	1.02
Transfers (social and remittances)	1.11
Combined household earnings	2.28
Total	100

Empirical Specification and Results

Empirical Specification

For the empirical work, I assume a canonical probit model expressing the schooling outcomes. The model is hypothesized as follows:

Pr (Y = 1 | X) = φ (β_{0} + β_{1} X_{c} + β_{2} T_{m} + β_{3} X_{p} + β_{4} X_{h}),

where Pr(Y = 1|X) is the probability of school enrollment, if realized, is equal to 1 and conditioned on the independent variables, X, with a standard normal distribution function of ϕ (.), also known as the z-index value. The indicator T_m measures the mother’s tolerance for wife abuse; the vector X_c is of the child’s characteristics; X_p is the vector of father’s and mother’s characteristics such as age, schooling, marriage type, having a formal job; and X_h is a vector of household characteristics.

The probability of children’s school enrollment, the outcome variable in this study, is represented by a binary categorical variable which consists of two mutually exclusive states. A value of 1 represents age-appropriate grade attendance and repeat grade, and a value of 0 represents drop-out and never attended. While this approach combines the variety of children’s schooling experiences, given the level of schooling selected in this study, it captures the rule of law in Turkey: For children between ages of 7 and 14, schooling is compulsory.

The empirical model tests the association between a mother’s tolerance level for wife beating and the school outcomes of her children, after controlling for other factors. I measure a mother’s tolerance to wife beating by adding up the categories (burns food, neglects children, argues with husband, wastes money, and refuses intercourse) that she agrees with. A score of 0 is assigned to mothers who reject all of these statements. A score of 5 is assigned to mothers who agree with all of them. The schooling status of parents is taken as a measure of the value the parents place on education, therefore both mothers’ and fathers’ years of schooling are included in the model. To control for household income, an asset index is incorporated into the empirical model. Using data on household’s ownership of identifiable assets, such as owning a TV, materials used in house construction, and access to sanitation and safe water, the asset index is generated by principal components analysis, and is divided into quintiles. The bottom wealth quintile represents the poorest 20% of the population, whereas the top quintile corresponds to the most affluent 20% in the population. The asset index can be considered as the best approximation to household living standards in the absence of information on expenditure or income (Filmer & Pritchett, 2001). Another concern is the prevalence of the informal sector in Turkey, which not only affects recorded and unrecorded incomes but also separates decent jobs from precarious ones (Buğra & Yakut-Cakar, 2010). A variable measuring whether father’s work offers a pension plan is included and assigned the value of 1 for households that are attached to the formal labor markets and therefore receive a steady income in addition to social transfers. The household demographic composition is included as another set of control variables. Given that children between the ages of 0 and 2 require more care work than pre-school children between the ages of 3 to 6 and may possibly influence schooling decisions for girls, the siblings between the ages of 0 to 2 and siblings between the ages of 3 to 6 are incorporated separately. To test competition for household resources in schooling, the number of siblings in the compulsory school age bracket is added as a control variable, in addition to the reference child’s age. The last controls reflect some dimensions of female bargaining power: difference in parental schooling and marriage type, an indicator variable that takes the value of 1 for arranged marriages, both of which capture the status of a mother vis-à-vis the father. Finally, I include the mother’s native language to capture the ethnic dimension in schooling outcomes. A value of 1 represents that the mother’s native language is Turkish, and a value of 0 represents other languages, such as Kurdish. Table 2 above provides the coding details of the variables.

Results on Children’s Schooling Outcomes

The reduced form models are presented in Tables 5 and 6, separated by the gender of the child. The non-linearity in the estimates of the probit model is adjusted and the corresponding marginal effects are reported in the tables. I construct the empirical model in three steps. First, only the variables measuring household composition and resources, regional controls and the main variable of interest, the mother’s level of tolerance for wife beating, are included. While not a perfect measure, regional controls capture the wide variation in school accessibility in the country, as noted in Erturk and Dayioglu (2004). In the second step, measures of parental educational differences, marriage type, and income security are added. The last model incorporates the cross-cutting ethnic dimension using a variable indicating whether Turkish is the mother’s native language. In this way, I test whether the mother’s attitude toward wife beating exerts any influence beyond other factors which may influence her conceptualization of her agency. This strategy helps to identify the extent to which the association between schooling outcomes and a mother’s tolerance to wife beating is robust to the inclusion of other agency measures. To avoid redundancy, I will discuss primarily the results of the augmented model, Model 3, reported in the last two columns of Tables 5 and 6.

Table 5.

Probability of School Attendance—Boys.

	Model 1	Marginal Effects	Model 2	Marginal Effects	Model 3	Marginal Effects
Age	0.023	0.001	0.021	0.001	0.026	0.001
Age	(0.028)	(0.002)	(0.028)	(0.002)	(0.028)	(0.001)
Father’s schooling	0.036*	0.002*			0.035*	0.002*
Father’s schooling	(0.020)	(0.001)			(0.020)	(0.001)
Mother’s schooling	0.012	0.001			0.022	0.001
Mother’s schooling	(0.024)	(0.001)			(0.026)	(0.001)
No. of siblings aged 0 to 2	−0.177**	−0.011**	−0.183**	−0.010**	−0.180**	−0.010**
No. of siblings aged 0 to 2	(0.085)	(0.005)	(0.090)	(0.005)	(0.091)	(0.005)
No. of siblings aged 3 to 6	−0.006	−0.000	−0.001	−0.000	−0.017	−0.001
No. of siblings aged 3 to 6	(0.076)	(0.005)	(0.077)	(0.004)	(0.076)	(0.004)
No. of siblings aged 7 to 14	−0.059	−0.004	−0.060	−0.003	−0.060	−0.003
No. of siblings aged 7 to 14	(0.044)	(0.003)	(0.039)	(0.002)	(0.039)	(0.002)
Mother tolerates wife beating	−0.027	−0.002	−0.025	−0.001	−0.012	−0.001
Mother tolerates wife beating	(0.165)	(0.010)	(0.161)	(0.009)	(0.173)	(0.009)
Father’s work offers pension plan			0.329*	0.018*	0.330*	0.018*
Father’s work offers pension plan			(0.188)	(0.010)	(0.198)	(0.010)
Marriage (= 1 arranged)			−0.410***	−0.021***	−0.408***	−0.020***
Marriage (= 1 arranged)			(0.131)	(0.006)	(0.132)	(0.006)
Mother’s native language (= 1 Turkish)					0.313*	0.016**
Mother’s native language (= 1 Turkish)					(0.175)	(0.007)
Parental schooling difference			0.017	0.001
Parental schooling difference			(0.017)	(0.001)
Household asset wealth index (2nd quintile)	0.157	0.009	0.138	0.007	0.101	0.005
Household asset wealth index (2nd quintile)	(0.170)	(0.009)	(0.165)	(0.008)	(0.167)	(0.008)
Household asset wealth index (3rd quintile)	0.441**	0.021***	0.397**	0.018**	0.348*	0.015**
Household asset wealth index (3rd quintile)	(0.176)	(0.007)	(0.200)	(0.007)	(0.184)	(0.007)
Household asset wealth index (4th quintile)	0.546**	0.025***	0.513*	0.022**	0.472*	0.020**
Household asset wealth index (4th quintile)	(0.224)	(0.008)	(0.267)	(0.009)	(0.253)	(0.008)
Household asset wealth index (5th quintile)	0.509***	0.023***	0.434**	0.019***	0.277	0.012
Household asset wealth index (5th quintile)	(0.183)	(0.006)	(0.204)	(0.007)	(0.224)	(0.009)
Region (south = 1)	−0.159	−0.011	−0.152	−0.011	−0.109	−0.008
Region (south = 1)	(0.162)	(0.013)	(0.163)	(0.013)	(0.161)	(0.013)
Region (central = 1)	−0.112	−0.008	−0.097	−0.006	−0.054	−0.003
Region (central = 1)	(0.170)	(0.013)	(0.170)	(0.012)	(0.169)	(0.012)
Region (north = 1)	0.163	0.009	0.156	0.009	0.200	0.121
Region (north = 1)	(0.226)	(0.011)	(0.226)	(0.011)	(0.224)	(0.011)
Region (east = 1)	−0.292*	−0.020*	−0.286*	−0.019*	−0.266*	−0.019*
Region (east = 1)	(0.152)	(0.011)	(0.152)	(0.011)	(0.149)	(0.011)
Constant	1.225		1.634		1.648
Constant	(0.360)***		(0.315)***		(0.332)***
Observations	4,023		4,023		4,023
Model χ²	2,231.86		2,943.37		2,941.08
Log likelihood	−654.33		−642.91		−637.85
Pseudo R²	.14		.16		.16

Note. Reference category for wealth index is bottom 20% quintile, reference category for regions is West. Robust standard errors in parentheses.

p < .1. **p < .05. ***p < .01.

Table 6.

Probability of School Attendance—Girls.

	Model 1	Marginal Effects	Model 2	Marginal Effects	Model 3	Marginal Effects
Age	−0.169***	−0.015***	−0.163***	−0.019***	−0.173***	−0.015***
Age	(0.022)	(0.003)	(0.022)	(0.003)	(0.022)	(0.003)
Father’s schooling	0.093***	0.008***			0.080***	0.008***
Father’s schooling	(0.018)	(0.002)			(0.018)	(0.002)
Mother’s schooling	0.090***	0.008***			0.087***	0.007***
Mother’s schooling	(0.018)	(0.002)			(0.022)	(0.002)
No. of siblings aged 0 to 2	−0.351***	−0.031***	−0.345***	−0.039***	−0.344***	−0.031***
No. of siblings aged 0 to 2	(0.076)	(0.009)	(0.082)	(0.011)	(0.079)	(0.009)
No. of siblings aged 3 to 6	0.015	0.001	−0.013	−0.001	0.023	0.002
No. of siblings aged 3 to 6	(0.073)	(0.007)	(0.077)	(0.009)	(0.076)	(0.007)
No. of siblings aged 7 to 14	−0.107**	−0.010**	−0.155***	−0.018***	−0.096**	−0.009**
No. of siblings aged 7 to 14	(0.046)	(0.004)	(0.044)	(0.005)	(0.046)	(0.004)
Mother tolerates wife beating	−0.301*	−0.027*	−0.435***	−0.049**	−0.297*	−0.026*
Mother tolerates wife beating	(0.161)	(0.015)	(0.165)	(0.020)	(0.160)	(0.014)
Father work offers pension plan			0.187* (0.104)	0.021* (0.012)	0.173* (0.102)	0.010* (0.006)
Marriage (= 1 arranged)			−0.106	−0.012	−0.032	−0.003
Marriage (= 1 arranged)			(0.116)	(0.013)	(0.119)	(0.011)
Mother’s native language (= 1 Turkish)					0.205	0.019
Mother’s native language (= 1 Turkish)					(0.167)	(0.017)
Parental schooling difference			0.016	0.002
Parental schooling difference			(0.013)	(0.002)
Household asset wealth index (2nd quintile)	0.121	0.010	0.215*	0.022**	0.114	0.010
Household asset wealth index (2nd quintile)	(0.114)	(0.009)	(0.116)	(0.011)	(0.111)	(0.009)
Household asset wealth index (3rd quintile)	0.562***	0.038***	0.709***	0.058***	0.541***	0.037***
Household asset wealth index (3rd quintile)	(0.192)	(0.009)	(0.172)	(0.010)	(0.189)	(0.009)
Household asset wealth index (4th quintile)	0.671***	0.045***	0.904***	0.072***	0.653***	0.044***
Household asset wealth index (4th quintile)	(0.230)	(0.010)	(0.227)	(0.011)	(0.234)	(0.010)
Household asset wealth index (5th quintile)	0.383	0.028**	0.836***	0.064***	0.343	0.025
Household asset wealth index (5th quintile)	(0.257)	(0.014)	(0.295)	(0.014)	(0.273)	(0.016)
Region (south = 1)	0.161	0.015	0.143	0.014	0.152	0.177
Region (south = 1)	(0.138)	(0.012)	(0.138)	(0.012)	(0.143)	(0.014)
Region (central = 1)	0.130	0.013	0.110	0.011	0.182	0.021
Region (central = 1)	(0.148)	(0.014)	(0.149)	(0.014)	(0.143)	(0.014)
Region (north = 1)	−0.024	−0.002	−0.038	−0.004	−0.055	−0.006
Region (north = 1)	(0.167)	(0.18)	(0.167)	(0.018)	(0.161)	(0.019)
Region (east = 1)	−0.104	−0.011	−0.117	−0.012	−0.099	−0.012
Region (east = 1)	(0.113)	(0.012)	(0.112)	(0.012)	(0.106)	(0.014)
Constant	2.296***		3.007***		2.225***
Constant	(0.328)		(0.314)		(0.359)
Observations	3,928		3,928		3,928
Model χ²	968.44		1,552.01		1,403.96
Log likelihood	−965.78		−1,014.88		−962.85
Pseudo R²	.34		.31		.35

Note. Reference category for wealth index is bottom 20% quintile, reference category for regions is West. Robust standard errors in parentheses.

p < .1. **p < .05. ***p < .01.

While only the schooling level of the father is statistically significant for the reference boy in the sample, for the reference girl, the schooling levels of both parents are associated with her school outcomes. A year increase in mother’s schooling increases the probability of school enrollment on average by 0.7 percentage points, at a level only slightly less than that of the father, 0.8 percentage points. The age of the boy child is not statistically associated with his school enrollment, contrary to the case observed for girls. The model predicts a 1.5 percentage point decrease in the probability of school enrollment for the reference girl, from age 7 to 14, annually for each year. An increase in the number of siblings aged 0 to 2 decreases the probability of school enrollment both for boys and girls, and this association is statistically significant. The model does not predict a similar association between siblings aged 3 to 6 and school enrollment of the reference child. The statistical association between same-age siblings and the schooling outcome of the reference child, however, presents an interesting outcome. For boys, no statistically significant association exists but for girls, an increase in the size of similar-age siblings decreases the probability of school enrollment and this link is statistically significant. The results suggest that, while the reference boy does not compete for household resources, school enrollment for the reference girl, partly depends on her siblings.

Household asset index quintile level is correlated with the observed probability of school enrollment in the direction expected. The children of better-off families are more likely to enroll in school and the reported marginal effects show that it is slightly stronger for girls than for the boys. The variable measuring household income security, whether father’s job is in the formal labor market, shows that all else being equal, children living in households which receive steady income and social transfers, such as old-age, widowhood, disability, heating assistance transfers, are more likely to enroll in school, compared with children whose fathers face insecure employment.

I next consider the parameters which incorporate a mother’s bargaining power into the model: schooling differences between parents and whether the parental marriage was arranged. Tables 5 and 6 show that parental schooling differences fail to exert any statistically significant influence on children’s schooling status. This is likely to be a result of the prevalence of arranged marriages, which often lead to matching of similar persons in the marriage market. Contrary to differences in parental educational attainment, the results suggest that having a mother whose marriage was arranged may lead to different schooling outcomes for her children. To the extent that arranged marriage captures the low agency power of these mothers, the results suggest that having a mother with relatively high agency increases the probability of school enrollment for the reference boy, but not for the reference girl. Why would such a difference exist? One possible explanation is related to the structure of the Turkish labor market. Given the low rate of female participation in the labor market, mothers holding a relatively better bargaining position may opt to educate sons rather than daughters.

The estimates of the main variable of concern, labeled “Mother’s Tolerance” in Tables 5 and 6, show two interesting results. First, they suggest that a mother’s attitude toward wife beating does exert an influence on child schooling outcomes beyond that which the material resources of the household and the mother’s bargaining power may hold. Second, this association is statistically significant only for girls. The more a mother agrees with the reasons given as meriting beating, the less likely is her daughter to be enrolled in school, a mother agreeing with four rather than three of the five abuse indicators is associated with a 2.6 percentage point decrease in the probability of her daughter being enrolled in school. A change in the level of tolerance of wife abuse does not exert a statistically significant influence on the schooling outcome of the reference boy. The empirical models presented in Tables 5 and 6 provide some support to the conceptual model. The differential results of a mother’s bargaining power on the schooling outcomes of boys and girls suggest that, at least in Turkey, the spillover effects of norms sanctioning wife beating to well-being outcomes for children exist, but only for the girls.

Discussion and Policy Implications

The literature on children’s schooling in developing countries revolves primarily around the material condition of the household. This study is undertaken with the intention of both adding to that existing understanding and of building on it by moving beyond the conceptualization of the household as a monotonic unit. After reaffirming the correlation between household economic status and children’s schooling outcomes, I looked at one aspect of the complex power relationships within the household, which I hypothesized to also influence children’s schooling. The results suggest that further study of intra-household dynamics is needed to better understand the differences in educational outcomes among households of similar economic standing. The conceptual model framed the relationship between a mother’s tolerance of wife abuse with her children’s schooling outcomes to highlight the spillover effects from gender norms to educational outcomes. The empirical results suggest that in Turkey within the compulsory schooling age group, the daughters of mothers who tolerate being beaten by the husbands are less likely to enroll in school, all else being equal, than the daughters of mothers who do not tolerate abuse. In comparison, the schooling outcomes of male children are not sensitive to the mother’s attitude toward wife beating, and the greater the mother’s bargaining power, the more likely is her son to be enrolled in school. The results suggest that mothers may rely on different strategies available to them for the schooling of boys and girls. I should also note that the data set used in this study is not without drawbacks. The data set contains no information on the costs of schooling or the characteristics of communities. The mothers included in the study are between 15 and 49 years of age. Finally, the empirical results fall short of understanding the full income effects on schooling outcomes, because no individual-level data on income could be extracted from the survey. The results nevertheless suggest new research directions by drawing attention to the importance of social norms of gender equity to children’s schooling outcomes beyond that of the material circumstances of the household. Future studies could explore the channels through which the transmission of such norms takes place. A mother who tolerates beating in marriage may impose her values on her daughter, or the daughter may internalize norms that fail to evaluate the full value of education. We may learn more by incorporating community-level variables and understand the extent to which the community structure imposes active and passive barriers to schooling. A mother who lives in an area without schools may approve or disapprove not only marital violence, but also the value of education differently from a mother who lives in a community within which schools are available. The fact that a mother’s tolerance of marital violence exerts a statistically significant effect on her daughter’s schooling, but not on that of her sons, suggests that focusing only on price and income incentives to households may fail to improve girls’ schooling outcomes. In contrast, integrated strategies which include interventions targeting the social norms that sanction gender inequalities can be expected to be more likely to affect parental commitment to their children’s schooling, at least within the Turkish context.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biography

Sevinc Rende is an assistant professor of economics at the Department of Economics, Isik University, Istanbul. Her research concentrates on the intersections of social policy and social networks.

References

Abramsky

Watts

C. H.

Garcia-Moreno

Devries

Kiss

Ellsberg

Heise

(2011). What factors are associated with recent intimate partner violence? Findings from the WHO multi-country study on women’s health and domestic violence. BMC Public Health, 11, Article 109. doi:10.1186/1471-2458-11-109

Agarwal

Panda

(2007). Toward freedom from domestic violence: The neglected obvious. Journal of Human Development, 8, 359-388. doi:10.1080/14649880701462171

Akar

Aksakal

F. N.

Demirel

Durukan

Özkan

(2010). The prevalence of domestic violence against women among a group woman: Ankara, Turkey. Journal of Family Violence, 25, 449-460. doi:10.1007/s10896-010-9306-8

Alcaraz

Chiquiar

Salcedo

(2012). Remittances, schooling, and child labor in Mexico. Journal of Development Economics, 97, 156-165. doi:10.1016/j.jdeveco.2010.11.004

Altinay

Arat

(2009). Violence against women in Turkey: A nationwide survey (Knowledge Creation Diffusion Utilization). Istanbul, Turkey: Punto.

Bakirci

(2002). Child labour and legislation in Turkey. International Journal of Children’s Rights, 10, 55-72.

Boy

Kulczycki

(2008). What we know about intimate partner violence in the Middle East and North Africa. Violence Against Women, 14, 53-70. doi:10.1177/1077801207311860

Buğra

Yakut-Cakar

(2010). Structural change, the social policy environment and female employment in Turkey. Development and Change, 41, 517-538. doi:10.1111/j.1467-7660.2010.01643.x

Dayioglu

(2006). The impact of household income on child labour in urban Turkey. Journal of Development Studies, 42, 939-956. doi:10.1080/00220380600774723

10.

Dayioglu

Kirdar

M. G.

Tansel

(2009). Impact of sibship size, birth order and sex composition on school enrolment in urban Turkey. Oxford Bulletin of Economics and Statistics, 71, 399-426. doi:10.1111/j.1468-0084.2008.00540.x

11.

DeGraff

D. S.

Levison

(2009). Children’s work and mothers’ work—What is the Connection? World Development, 37, 1569-1587. doi:10.1016/j.worlddev.2009.01.006

12.

Dhaher

E. A.

Mikolajczyk

R. T.

Maxwell

A. E.

Krämer

(2010). Attitudes toward wife beating among Palestinian women of reproductive age from three cities in West Bank. Journal of Interpersonal Violence, 25, 518-537. doi:10.1177/0886260509334409

13.

Diop-Sidibé

Campbell

J. C.

Becker

(2006). Domestic violence against women in Egypt—Wife beating and health outcomes. Social Science & Medicine, 62, 1260-1277. doi:16/j.socscimed.2005.07.022

14.

Ergin

Bayram

Alper

Selimoglu

Bilgel

(2005). Domestic violence: A tragedy behind the doors. Women & Health, 42, 35-51.

15.

Erturk

Dayıoglu

(2004). Gender, education and child labour in Turkey ( Mottaz

, Ed.). (1st ed.). Geneva, Switzerland: International Labor Organization.

16.

Evans

Davies

Dilillo

(2008). Exposure to domestic violence: A meta-analysis of child and adolescent outcomes. Aggression and Violent Behavior, 13, 131-140. doi:10.1016/j.avb.2008.02.005

17.

Filmer

Pritchett

(2001). Estimating wealth effects without expenditure data—Or tears: An application to educational enrollments in states of India. Demography, 38, 115-132. doi:10.1353/dem.2001.0003

18.

Flood

Pease

(2009). Factors influencing attitudes to violence against women. Trauma, Violence, & Abuse, 10, 125-142. doi:10.1177/1524838009334131

19.

Glick

(2008). What policies will reduce gender schooling gaps in developing countries: Evidence and interpretation. World Development, 36, 1623-1646. doi:16/j.worlddev.2007.09.014

20.

Goksen

Cemalcilar

(2010). Social capital and cultural distance as predictors of early school dropout: Implications for community action for Turkish internal migrants. International Journal of Intercultural Relations, 34, 163-175. doi:10.1016/j.ijintrel.2009.11.006

21.

Haj-Yahia

M. M.

(1997). Predicting beliefs about wife beating among engaged Arab men in Israel. Journal of Interpersonal Violence, 12, 530-545.

22.

Haj-Yahia

M. M.

(2002). Beliefs of Jordanian women about wife-beating. Psychology of Women Quarterly, 26, 282-291. doi:10.1111/1471-6402.t01-1-00067

23.

Hindin

M. J.

(2003). Understanding women’s attitudes towards wife beating in Zimbabwe. Bulletin of the World Health Organization, 81, 501-508.

24.

Holt

Buckley

Whelan

(2008). The impact of exposure to domestic violence on children and young people: A review of the literature. Child Abuse & Neglect, 32, 797-810. doi:10.1016/j.chiabu.2008.02.004

25.

Jewkes

(2002). Intimate partner violence: Causes and prevention. Lancet, 359, 1423-1429. doi:10.1016/S0140-6736(02)08357-5

26.

Johnson

K. B.

Das

M. B.

(2009). Spousal violence in Bangladesh as reported by men: Prevalence and risk factors. Journal of Interpersonal Violence, 24, 977-995. doi:10.1177/0886260508319368

27.

Khawaja

Linos

El-Roueiheb

(2007). Attitudes of men and women towards wife beating: Findings from Palestinian refugee camps in Jordan. Journal of Family Violence, 23, 211-218. doi:10.1007/s10896-007-9146-3

28.

Kirdar

M. G.

(2009). Explaining ethic disparities in school enrollment in Turkey. Economic Development and Cultural Change, 57, 297-333.

29.

Kocacik

(2007). Domestic violence against women: A field study in Turkey. The Social Science Journal, 44, 698-720. doi:10.1016/j.soscij.2007.10.016

30.

Krishnan

Rocca

C. H.

Hubbard

A. E.

Subbiah

Edmeades

Padian

N. S.

(2010). Do changes in spousal employment status lead to domestic violence? Insights from a prospective study in Bangalore, India. Social Science & Medicine, 70, 136-143. doi:10.1016/j.socscimed.2009.09.026

31.

Lawoko

(2008). Predictors of attitudes toward intimate partner violence: A comparative study of men in Zambia and Kenya. Journal of Interpersonal Violence, 23, 1056-1074. doi:10.1177/0886260507313972

32.

Linos

Khawaja

Al-Nsour

(2010). Women’s autonomy and support for wife beating: Findings from a population-based survey in Jordan. Violence and Victims, 25, 409-419. doi:10.1891/0886-6708.25.3.409

33.

Lloyd

Mete

Grant

(2009). The implications of changing educational and family circumstances for children’s grade progression in rural Pakistan: 1997-2004. Economics of Education Review, 28, 152-160. doi:10.1016/j.econedurev.2007.04.005

34.

Luke

Schuler

S. R.

Mai

B. T. T.

Vu Thien

(2007). Exploring couple attributes and attitudes and marital violence in Vietnam. Violence Against Women, 13, 5-27.

35.

Lunkenheimer

E. S.

Kittler

J. E.

Olson

S. L.

Kleinberg

(2006). The intergenerational transmission of physical punishment: Differing mechanisms in mothers’ and fathers’ endorsement? Journal of Family Violence, 21, 509-519. doi:10.1007/s10896-006-9050-2

36.

Marazyan

(2011). Effects of a sibship extension to foster children on children’s school enrolment: A sibling rivalry analysis for Indonesia. Journal of Development Studies, 47, 497-518.

37.

Marshall

G. A.

Furr

L. A.

(2010). Factors that affect women’s attitudes toward domestic violence in Turkey. Violence and Victims, 25, 265-277. doi:10.1891/0886-6708.25.2.265

38.

Mayda

A. S.

Akkuş

(2005). Domestic violence against 116 Turkish housewives: A field study. Women & Health, 40, 95-108. doi:10.1300/J013v40n03_07

39.

Obeid

Chang

D. F.

Ginges

(2010). Beliefs about wife beating: An exploratory study with Lebanese students. Violence Against Women, 16, 691-712. doi:10.1177/1077801210370465

40.

Ogunjuyigbe

P. O.

Akinlo

Ebigbola

J. A.

(2005). Violence against women: An examination of men’s attitudes and perceptions about wife beating and contraceptive use. Journal of Asian and African Studies, 40, 219-229. doi:10.1177/0021909605055070

41.

Ozcakir

Bayram

Ergin

Selimoglu

Bilgel

(2008). Attitudes of Turkish men toward wife beating: A study from Bursa, Turkey. Journal of Family Violence, 23, 631-638. doi:10.1007/s10896-008-9185-4

42.

Rani

Bonu

(2009). Attitudes toward wife beating a cross-country study in Asia. Journal of Interpersonal Violence, 24, 1371-1397.

43.

Rani

Bonu

Diop-Sidibe

(2004). An empirical investigation of attitudes towards wife-beating among men and women in seven sub-Saharan African countries. African Journal of Reproductive Health, 8, 116-136.

44.

Rocca

C. H.

Rathod

Falle

Pande

R. P.

Krishnan

(2009). Challenging assumptions about women’s empowerment: Social and economic resources and domestic violence among young married women in urban south India. International Journal of Epidemiology, 38, 577-585. doi:10.1093/ije/dyn226

45.

Rose

Al Samarrai

(2001). Household constraints on schooling by gender: Empirical evidence from Ethiopia. Comparative Education Review, 45, 36-63.

46.

Schuler

S. R.

Rottach

(2010). Women’s empowerment across generations in Bangladesh. Journal of Development Studies, 46, 379-396.

47.

Simister

Mehta

P. S.

(2010). Gender-based violence in India: Long-term trends. Journal of Interpersonal Violence, 25, 1594-1611. doi:10.1177/0886260509354577

48.

Sousa

Herrenkohl

T. I.

Moylan

C. A.

Tajima

E. A.

Klika

J. B.

Herrenkohl

R. C.

Russo

M. J.

(2011). Longitudinal study on the effects of child abuse and children’s exposure to domestic violence, parent-child attachments, and antisocial behavior in adolescence. Journal of Interpersonal Violence, 26, 111-136. doi:10.1177/0886260510362883

49.

Speizer

I. S.

(2010). Intimate partner violence attitudes and experience among women and men in Uganda. Journal of Interpersonal Violence, 25, 1224-1241. doi:10.1177/0886260509340550

50.

Stuart

G. L.

Ramsey

S. E.

Moore

T. M.

Kahler

C. W.

Farrell

L. E.

Recupero

P. R.

Brown

R. A.

(2003). Reductions in marital violence following treatment for alcohol dependence. Journal of Interpersonal Violence, 18, 1113-1131. doi:10.1177/0886260503255550

51.

Taft

C. T.

Bryant-Davis

Woodward

H. E.

Tillman

Torres

S. E.

(2009). Intimate partner violence against African American women: An examination of the socio-cultural context. Aggression and Violent Behavior, 14, 50-58. doi:10.1016/j.avb.2008.10.001

52.

UN Department of Economic and Social Affairs. (2010). The world’s women 2010: Trends and statistics. New York, NY: United Nations.

53.

UNWomen. (2012). Progress of the world’s women: In pursuit of justice. New York, NY: Author.

54.

Villarreal

(2007). Women’s employment status, coercive control, and intimate partner violence in Mexico. Journal of Marriage and Family, 69, 418-434. doi:10.1111/j.1741-3737.2007.00374.x

55.

Yoshioka

M. R.

DiNoia

Ullah

(2001). Attitudes toward marital violence: An examination of four Asian communities. Violence Against Women, 7, 900-926.

56.

Zhao

Glewwe

(2010). What determines basic school attainment in developing countries? Evidence from rural China. Economics of Education Review, 29, 451-460. doi:10.1016/j.econedurev.2009.10.008

57.

Zhu

Dalal

(2010). Childhood exposure to domestic violence and attitude towards wife beating in adult life: A study of men in India. Journal of Biosocial Science, 42, 255-269.