impact of the spread of mass education on domestic violence...
TRANSCRIPT
Impact of the Spread of Mass Education on Domestic Violence against Wives
Abstract
This paper investigates the association between mass education and domestic violence
against wives in rural Nepal. Previous research on domestic violence in South Asian societies
reveals the important influence of prevailing patriarchal ideology and the subordinate status of
women within the family. However, the recent spread of mass education is likely to have
important consequences in the opposite direction. Using data from 1,778 currently married
women and their husbands from the Chitwan Valley Family Study in Nepal, we test the
consequences of spread of mass education for domestic violence against wives. The results show
that women’s childhood access to school, their parents’ schooling, and women’s own and their
husbands' schooling each tend to reduce wives’ likelihood of experiencing domestic violence.
Indeed husbands’ education is a particularly powerful force reducing domestic violence against
wives. These associations explain the historical declines in the incidence of domestic violence
against wives in Nepal.
Key words: (Domestic violence, Education, Marriage, Social change)
1
Rarely do we have opportunities for empirical insights into the forces driving variation in
women’s exposure to domestic violence from representative studies of the general population in
rapidly changing Asian societies. Many recent studies indicate high levels of domestic violence
against women worldwide (Diop-Sidibe, Campbell, & Becker, 2006; Naved, Azim, Bhuiya, &
Persson, 2006; WHO, 2001), with the highest level and most severe forms of domestic violence
in the South Asian region (Carvalho, 2007; Naved et al., 2006; Panday, Dutt, & Banerjee, 2008).
Research has documented significant social, physical, and mental health consequences for
victims of domestic violence (Devries et al., 2011; Ellsberg, Jansen, Heise, Watts, & García-
Moreno, 2008; Naved & Akhtar, 2008; Stephenson, Koenig, & Ahmed, 2006). Despite high
levels of violence against women in South Asia, most empirical evidence on domestic violence
comes from Western, industrialized societies and little is known about the forces shaping
domestic violence in South Asia (Koenig, Stephenson, Ahmed, Jejeebhoy, & Campbell, 2006;
Naved et al., 2006; Panday et al., 2008). To address this gap we investigate the association
between Western-oriented, mass education and domestic violence against wives. Because
women experience a disproportionately high level of domestic violence compared to men
(Bagheri, Nabavi, & Hosseinzadeh, 2009; Pradhananga & Shrestha, 2009; Yount, Halim, Hynes,
& Hillman, 2011), our investigation focuses on physical violence against married women.
This study advances our knowledge of domestic violence against wives in two important
ways. First, we provide a new theoretical framework for the study of variation in the incidence of
interpersonal violence that focuses on education. This theoretical framework identifies
multidimensional connections between the spread of Western-oriented, mass education and the
incidence of domestic violence against wives. This new framework is fundamentally multilevel,
grounded in the idea that changes in the community educational context restructure individual-
2
level opportunities and constraints, which in turn affect individual behavior (Alexander, 1988;
Coleman, 1990; Durkheim, 1984 [1893]). The framework is also intergenerational, identifying
mechanisms across generations that may shape the incidence of domestic violence. We develop
hypotheses regarding variation in community educational context, individual and family
participation in schooling, and domestic violence against wives. Building on the life course
perspective, this framework acknowledges the importance of change over time at multiple levels
as well as the sequencing of these changes, allowing us to order various dimensions of
educational variation over the lifetimes of respondents (Axinn & Barber, 2001; Elder, 1985). We
consider women’s own, their husbands’, their parents’, and their in-laws’ educational
experiences as mechanisms linking educational opportunity to individual-level experiences of
domestic violence.
Second, we use unusually detailed educational measures from a contemporary, large-
scale South Asian panel study to conduct empirical tests of the framework. Because most studies
of the spread of mass education are historical, significant empirical limitations preclude the
exploration of important theoretical issues regarding consequences of mass education (Barber &
Axinn, 2004; Caldwell, Reddy, & Caldwell, 1988; Khaniya & Kiernan, 1994). In a setting where
the spread of mass education occurred recently, however, it is possible to gather direct
information about its spread, consequences, and the individual-level behaviors that produced
those consequences. By using measures of childhood exposure to education at the community
level, variations in parental, husbands’, and wives’ educational experiences, and the incidence of
domestic violence, we provide comprehensive empirical estimates of the multidimensional,
educational factors producing variations in domestic violence against wives. We use data from
the Chitwan Valley Family Study (CVFS) in Nepal. The CVFS data feature measures of multiple
3
dimensions of individual and parental experiences from identical but separate interviews with
husbands and wives, providing a unique opportunity to investigate the influence of both the
perpetrator’s and the victim’s characteristics.
Theoretical Framework
Scholars argue that the high levels and severe forms of domestic violence in South Asia
are linked to the prevailing patriarchal ideology, a major component of the South Asian, cultural
configuration (Ahmad, Riaz, Barata, & Stewart, 2004; Cain, Khanam, & Nahar, 1979; Johnson,
1995; Srinivas, 1952). This cultural configuration evolved through Hindu religious doctrines that
emphasize strong hierarchal relations based on gender (male supremacy over women), caste
(high caste Hindus over low caste Hindus), and seniority (seniors over juniors) (Carvalho, 2007;
Majumdar, 2003; Naved et al., 2006; Panday et al., 2008). For example, according to Manusmriti
(9.3), one of the most regarded Hindu religious scripts, a woman is protected by her father in
childhood, by her husband in youth, and by her sons in old age, and is never fit for independence.
Once a woman is married, she becomes the property and responsibility of her husband. Her life
is supposed to be devoted to her husband’s service and taking care of his children.
Although Hindu ideology is presumed to be originally associated with the Indo-Aryan
group, they persuaded others to follow their religion and customs (Adhikari, 1998; Regmi,
1999). In many parts of South Asia this ideology has become pervasive even among other non-
Indo-Aryan groups through the adoption of values, beliefs, and customs of the high caste Hindus
who are guided by Sanskrit scholarship—a process often referred as “sanskritization.” As a
result, both as an ideological and normative force, Hindu religious doctrine has played an
important role in both marriage and marital relationships, including domestic violence against
women (Acharya & Bennett, 1981).
4
Until the mid-1950s, Nepal was socially, politically, and economically isolated from
other countries (Berreman, 1972; Bista, 1972). The people of Nepal had little exposure to and
interaction with people who had vastly different views and beliefs about social relationships—
people who believed in independence and personal freedom, gender equality, marital
relationships based on individual choice, and did not believe in a caste system (Macfarlane,
1986; Smith, 1973). Since the mid-1970s, Nepal has experienced dramatic social and economic
change (English, 1985; Panday, 1999). Nepal also started receiving a large portion of its
development budget in foreign aid targeted at the improvement of living conditions of the rural
poor through improvement in education systems and rural infrastructure (Panday, 1999). The
result has been a dramatic spread of formal education, wage work, transportation and
communication infrastructure, and the mass media. The first signs of the consequences of these
social changes include increasing participation in education, exposure to mass media, access to
health services, transportation and communication technology (Beutel & Axinn, 2002; Jamison
& Lockheed, 1987). Among the other aspects of social change, exposure to community
educational context and educational experiences have had dramatic consequences for many
dimensions of family change, including marriage timing, marital arrangements, childbearing and
marital relationships (Allendorf & Ghimire, Forthcoming; Ghimire, Axinn, Yabiku, & Thornton,
2006; Hoelter, Axinn, & Ghimire, 2004).
Spread of Mass Education
The spread of mass education has been an important component of social change
worldwide and a powerful vehicle for the spread of Western ideas in South Asia (Caldwell,
1982; Caldwell et al., 1988; English, 1985; Thornton, 2005). Since the 1950s, schools have
spread quickly in Nepal with enrollments in school rising quickly, and female enrollment rapidly
5
catching up to male enrollment (Beutel & Axinn, 2002; Stash & Hannum, 2001). Because the
Nepalese educational system and course materials were adopted from the British educational
system (sometimes via India), the new school environment provided exposure to Western ideas
and values such as individual freedom, equality, independence, and consensual and late marriage
(Sharma, 1994). In Nepal, the spread of mass education also gave women new skills and
credentials, providing a completely new route to economic independence from families and
husbands (Axinn & Barber, 2001). This change gave women new opportunities to react to
violence by leaving the household. A powerful force for change, we expect that multiple
dimensions of this mass education may have independent consequences for domestic violence
against wives. However, in all dimensions we predict the direction of these consequences is
toward lower likelihood of domestic violence against wives.
Community Educational Context. The community educational context may be associated
with domestic violence against wives in two ways. First, at the very beginning of the spread of
mass education in a setting like Nepal, exposure to educational institutions during childhood
shapes individuals’ long-term personality characteristics independent of their educational
experiences (Axinn & Yabiku, 2001). In a Hindu context, young people who had exposure to
nearby schools in early childhood will be more likely to have exposure to male-female
interactions outside of family settings, building interaction skills based on consensual behavior
rather than patriarchal relationships. Children in such settings are also more likely to be exposed
to values that discourage domestic violence against wives even if they themselves do not go to
school. Together these factors are likely to then shape their social interactions later in life,
including less domestic violence against wives.
6
Second, the spread of schools is linked to the education of individuals across the life
course (Elder, 1985; Axinn & Barber, 2001). For parents or children to have the opportunity to
go to school, schools must be located nearby, or parents must send their children to live near
schools. In rural Nepal, families sometimes send children great distances so they can attend
school, but the majority of rural families do not have the resources necessary to take advantage
of distant schooling opportunities (Acharya & Bennett, 1981; Brauner-Otto, Forthcoming; Stash
& Hannum, 2001). As schools become available in a community, individuals will be more likely
to attend school, thereby increasing the educational attainment of both parents and children over
time (Beutel & Axinn, 2002; Jamison & Lockheed, 1987). As we describe below, by making
education more available to individuals, the spread of schools unleashes a powerful chain of
individual experiences that shape the likelihood of domestic violence against wives.
Hypothesis 1: Proximity to school during childhood should reduce domestic violence
against wives.
Parents’ Education. A large body of literature has documented important
intergenerational influences of education (Cleland & Rodriguez, 1988; Marini 1978; Thornton et
al., 2007). This literature identifies socialization, social control and similarity of social position
as mechanisms that combine to make parents an extraordinarily influential dimension of
children’s lives (Bengtson, 1975; Gecas & Seff, 1990; Smith, 1988; Axinn and Thornton, 1993).
Through these mechanisms it is likely parental education will influence children, including
children’s domestic violence against wives. In the Nepalese setting educated parents are more
likely to hold values and beliefs discouraging domestic violence against wives among their sons
and increasing intolerance of domestic violence among their daughters. Because behavior of
husbands is a crucial determinant of women’s experience of domestic violence (Majumdar,
7
2003; Panday et al., 2008), a woman’s parent-in-laws’ educational experiences may affect her
experience of domestic violence. But the education of women’s own parents may also be
relevant, either because it shapes women’s own behavior toward their husbands in ways that
discourage husbands’ violent acts or because it shapes the parents’ willingness to have their
daughter participate in the selection of her husband (Ghimire et al., 2006). Thus, both parent-in-
law and parental educational experiences may have consequences for domestic violence in this
setting. In both cases, more highly educated parents are expected to reduce the likelihood of
domestic violence against wives.
Hypothesis 2: Parental education (both wife’s own parents and her in-laws) should
discourage domestic violence against wives.
Wife’s Own and Her Husband’s Education. Finally, women’s and their husbands’
educational experiences are likely to be associated with the incidence of domestic violence
against wives in ways that are independent of both community context and parental experience.
At least three different mechanisms are relevant, beginning with the formation of the marriage
itself.
First, recent evidence indicates that education increases youth participation in spouse
choice (Ghimire et al., 2006). Changes toward more involvement of young people in the
selection of their own spouse are associated with change the nature of marital relationships,
leading to more communication and affection and less disagreement and conflict between
spouses (Rindfuss & Morgan, 1983; Thornton & Lin, 1994). The increase in communication and
affection and reduction in disagreement and conflict is likely to encourage mutual respect and
equality within marriages (Allendorf & Ghimire, Forthcoming; Ghimire et al., 2006; Hoelter et
al., 2004), which in turn is likely to reduce domestic violence against wives.
8
Second, scholars argue that increased education promotes young people’s independence,
increasing the adoption of new ideas and values. In the Nepal setting, this change is likely in the
direction away from historical Hindu ideas regarding the patriarchal basis for marital
relationships and toward more Western ideas regarding independence, equality, and interpersonal
relations based on mutual consent as the basis for marriage (Allendorf & Ghimire, Forthcoming;
Hoelter et al., 2004). Young peoples’ adoption of these new ideas and values will likely
discourage violence against wives.
Third, scholars also argue that education in the South Asian context increases a woman’s
autonomy relative to her husband, mother-in-law, or other key household members (Dyson &
Moore, 1983; Mason, 1987). If so, more highly educated women may have the means to mitigate
domestic violence and more highly educated husbands may expect their wives to have these
means. Economic independence and social autonomy produced by increasing education may be
an especially powerful force determining domestic violence in South Asia.
Hypothesis 3: Both women’s own and their husband’s educational attainment will
discourage domestic violence against wives.
Of course, by its nature, domestic violence toward wives is a behavior of a woman’s
husband that may be independent of any of her characteristics or experiences including her
parent’s characteristics or her in-law’s characteristics. For some, community context, parental
background, and women’s own ideas and options may be entirely irrelevant to the violent actions
of husbands. Because of this, it is possible that husbands’ own characteristics and experiences
are substantially more powerful determinant of the incidence of domestic violence than other
factors. If exposure to education in Western-oriented schools propagates the idea that domestic
violence is unacceptable, then husbands with more education may be significantly less likely to
9
commit acts of domestic violence than husbands with less education. Even if education merely
teaches men that in current circumstance women have options for reacting to domestic violence
they did not have before, such as leaving the household or contacting police, it is also possible
that independent of their ideas about the acceptability of domestic violence toward their wives
husbands with more education may be significantly less likely to commit acts of domestic
violence than husbands with less education. Finally, the links between educational opportunity
and schooling experience, parental education and their children’s schooling experience, and
women’s education and choices of husbands (educationally assortative mating practices
strengthened by arranged marriage) all mean that if husbands’ educations are a particularly
powerful force predicting domestic violence toward wives, husbands’ education may be the
mechanism that explains the observed association between every other dimension of educational
context and experience and domestic violence against wives. Schools nearby in childhood
increase husbands’ schooling, higher in-law education increases husbands’ schooling, higher
parental education increases the chance parents favor the choice of a husband with more
schooling, and higher women’s education increases the chance women favor the choice of a
husband with more schooling.
Hypothesis 4: The observed associations between community educational context, wives’
parents’ educations, husbands’ parents’ educations and wives’ own educations and domestic
violence against wives are partially explained by husbands’ educations.
A small number of factors are especially likely to affect both individuals’ educational
experiences and domestic violence, with the potential to render observed associations spurious.
These include the historical time period (birth cohort), ethnicity, and factors associated with the
nature and timing of the marriage itself. To focus our empirical tests on the four hypotheses
10
identified here, we will estimate multivariate models that include measures of these factors, as
described below.
Method
Data
The data used to test our hypotheses came from a study of 171 communities scattered
throughout the Western Chitwan Valley that closely resemble the characteristics of the entire
Chitwan Valley population (Barber et al., 1997).
Once a community was selected, a household census along with a household relationship
grid was administered in all households in the selected communities. All individuals aged 15 to
59 residing in the sampled households and their spouses, who may be living elsewhere and
outside of the age range of 15-59, were interviewed using a standardized questionnaire and a life
history calendar (LHC) with a 97 % response rate. This process yielded a sample of 1,913
currently married women. Because a small number of women (30) married more than once, we
included the couples married only once. Out of these 1,913 women, 1,783 women had their
husbands interviewed using the same survey instruments. Out of 1,783 women, five wives and
two husbands had missing information, which resulted in a sample of 1,778 wives.
In the standardized interviews, individuals were asked questions about their family
background, personal characteristics, experiences, and childhood community context. The LHC
portion of the survey collected information on residence, marriage, childbearing, schooling,
travel, and work experience (Axinn, Pearce, & Ghimire, 1999). Because the outcome—domestic
violence against wives—is measured retrospectively and the timing and frequency of violence is
not available, in order to maintain temporal order we limited the wives’ and husbands’
experiences to premarital experiences.
11
Measuring domestic violence
The outcome of interest is domestic violence against wives. Despite worldwide
engagement in understanding and addressing this issue, there is no single accepted definition.
The main point of contention is that violence against women is perceived differently in different
societies. Some urge a broad definition that includes any act of omission that causes harm to
women or keeps them in a subordinate position. For example, Article 2 of the United Nations
Declaration on the Elimination of Violence Against Women (1993) states:
“Violence against women shall be understood to encompass, but not be
limited to, the following: (a) Physical, sexual and psychological violence
occurring in the family, including battering, sexual abuse of female children in
the household, dowry-related violence, marital rape, female genital mutilation
and other traditional practices harmful to women, non-spousal violence and
violence related to exploitation; (b) Physical, sexual and psychological
violence occurring within the general community, including rape, sexual
abuse, sexual harassment and intimidation at work, in educational institutions
and elsewhere, trafficking in women and forced prostitution; (c) Physical,
sexual and psychological violence perpetrated or condoned by the State,
wherever it occurs” (Rauschning, Wiesbrock, & Lailach, 1997:355).
In spite of declarations against domestic violence and social actions designed to reduce it,
many cultures—including the one we study here—have beliefs, norms, and social institutions
that legitimize the subordination of women (Mathur, 2004; Yount et al., 2011). For example, an
occasional slap or blow performed by an outsider would be punishable, but the same action
would often go unchallenged or even be considered routine behavior when performed by the
12
husband (Heise, Ellsberg, & Gottemoeller, 1999). Thus, the concept of domestic violence not
only includes many potential dimensions but also varies by cultural contexts, making it difficult
to measure. Moreover, there is reason to suspect that it is under-reported to survey interviewers.
Because of these complexities, in this study we focus on physical violence—a common and less
ambiguous form of domestic violence. Many of our respondents admit being hit to our
interviewers. Our research includes months of direct observation and unstructured interviewing
to investigate the survey responses, and our interpretations of the results are informed by this
ethnographic investigation. Our approach is consistent with the interpretation that the individuals
who responded “yes” to having been hit are vastly more likely to have experienced domestic
violence than those who responded “no” to this question.
Domestic violence is measured by women’s responses to the question in the private
individual interview, “Has your husband ever beaten you?” If the respondent answered
affirmatively, this response was categorized as having experienced domestic violence and coded
as “1” (otherwise coded as “0”). As shown in Table 1, 17% of women in our sample reported
having experienced domestic violence at least once since they were married.
Measures of education
Our goal was to evaluate the association between spread of mass education and domestic
violence against wives. We operationalized education in three different ways: community
educational context during childhood, parents’ and in-laws educational experience, and the
wife’s own and husband’s educational experiences. Because the CVFS has collected information
separately from both the wife (the victim of the violence) and the husband (perpetrator of the
violence), we have separate measures of childhood community educational context, parents’
education, and individual schooling for the wife and the husband.
13
Childhood community educational context. Information about access to school during
childhood was collected through individual interviews. During these interviews, respondents
were asked, “Was there a school within a one-hour walk from your home at any time before you
were 12 years old?” If the response to this question is positive, it is coded as “1” and “0” if
otherwise. We then constructed dummy variables for whether there was a school within a one-
hour walk from the place of the respondent’s residence at any time before she or he was 12 years
old. As shown in Table 1, 82% of wives and 81% of husbands had a school within a one-hour
walk from their childhood residence. To create measures of couple's access to school during
childhood we summed up the wives’ and husbands’ responses and divided that by two, resulting
in a scale of 0 to 1.
(Table 1 About Here)
Parents’ education. We measured parents’ education in terms of whether or not each of
the parents ever went to school. The measure of parents’ education is derived from individual
interviews conducted separately with wives and husbands. In the individual interview, wives and
husbands were asked, “Did your mother ever go to school?” If the response to this question was
yes, the mother’s schooling was coded as “1” and “0” if otherwise. The same question was
repeated for father’s education. We then summed wife’s responses to those questions to create a
measure of wife’s parents’ education and husband’s responses to create measures of husband’s
parents’ education. This resulted in a scale of 0 to 2, where “0” means that none of the parents
(neither mother nor father) ever went to school, “1” means that either one of the parents went to
school, and “2” means that both parents went to school. A mean of .31 for wife’s parents’
education and .19 for husband’s parents’ education (Table 1) suggest that in the parental
generation less than one third were ever exposed to formal public schooling. Likewise, to create
14
a measure of couples’ parents’ education we summed up the wives’ parents’ and husbands’
parents’ education and divided by two, resulting in a scale of 0 to 2.
Schooling experiences. The CVFS collected a complete history of respondents’
educational experiences, including adult education (literacy programs), via the LHC. Schooling
was measured as the total number of years spent in school or adult education up to one year
before the respondents got married. Using this information we created a series of categorical
variables, 0=no education, 1-5 years=primary, 6-10 years=high school, and 12 and more years=
college and treated “no education” as the reference group. As shown in Table 1, men and women
in Nepal have significantly different educational experiences. For example, more than half (57%)
of wives have no education, whereas only one third (31%) of husbands have no education.
Other Factors associated with Domestic Violence against Wives
Recent evidence indicates that a number of wives’ and husbands’ background
characteristics and experiences other than education may have important independent influences
on domestic violence against wives (Koenig, Ahmed, Hossain & Mozumder, 2003; Yount, 2005;
Yount & Carrera, 2006). Our models of education and domestic violence against wives also
include measures of wives’ and husbands’ backgrounds (age and birth cohort, ethnicity, and
marital experiences) to help insure the associations we observe with education are not a product
of these other factors.
Age and birth cohort. Research suggests significant differences between birth cohorts
both in terms of exposure to education and spousal relationship quality (Hoelter et al., 2004).
Compared to older cohorts, more recent cohorts have a significantly higher level of exposure to
and experience with education and more egalitarian spousal relationships. We also use birth
cohort as an indication of social change in this population. We do so by examining both the
15
overall effect of birth cohort on trends in domestic violence against wives and the extent to
which the effect of this historical change can be accounted for by trends in the childhood
community educational context, parents’, husbands’ and wives’ own education. Respondent’s
birth cohort was coded in four categories: cohorts born between 1972-81, 1962-71, 1952-61, and
1936-51. We treated the youngest cohort born between 1972-81 as the reference group. Our
expectation is that the introduction of measures of spread of mass education into the analysis
explains much of the historical change in domestic violence against wives, thereby suggesting
that changes in the spread of mass education (childhood community context, parents’, husbands’
and wives’ own education) has played a substantial role in the historical change in domestic
violence against wives.
Ethnicity. Nepali society consists of many ethnic groups (Bista, 1972; Macfarlane, 1976;
Thapa, 1997) that are likely to have significantly different educational experiences and marital
relationships. Scholars have often categorized these ethnicities into five major groups for
analytical purposes: Brahmin/Chhetri (high caste Hindus), Dalit (low caste Hindus), Newar, Hill
Janajati (Hill indigenous), and TeraiJanjati (Terai indigenous) (Axinn & Yabiku, 2001). We
coded individuals as “1” if they are members of a specific category and “0” if not, and treated
Brahmin/Chhetrias a reference group. Given the historical background of these ethnic groups, we
expect Brahmin/Chhetri to have more education, and Janajati to have more egalitarian marital
relationships.
Marital experience. Many dimensions of marital experiences may be related to
educational attainment and affect the spousal relationship and lead to violence. Perhaps the
strongest and most obvious of these is marriage timing (age at marriage). In a setting like Nepal
where the vast majority of people marry while they are still young, older age at first marriage is
16
associated with higher women’s autonomy and better spousal relationships (Allendorf &
Ghimire, Forthcoming), likely reducing occurrences of domestic violence against wives. In an
arranged marriage society like Nepal, another marital experience likely to affect the likelihood of
violence against wives is participation in spouse selection. In general, higher participation in
spouse selection is found to be associated with a better spousal relationship and lower domestic
violence (Allendorf & Ghimire, Forthcoming; Hoelter et al., 2004; Link, 2011).
The CVFS collected a complete history of respondents’ life experiences, including adult
marital experiences, via the LHC. Using this information we created a series of categorical
variables including married at age 15 and younger, married between ages 16 and 20, and married
after age 20; we treated married at age 15 and younger as the reference group. As shown in Table
1, men and women in Nepal have significantly different marital experiences. Although 41% of
women got married at age 15 and younger, only 10% of men married at those ages. Likewise, the
mean age at first marriage is 16.25 years old for wives, but it is 20.65 years old for husbands
(4.40 years difference).
The CVFS measured spouse choice using responses to a question that asks, “People
marry in different ways. Sometimes our parents or relatives decide whom we should marry, and
sometimes we decide ourselves. In your case, who selected your (first) spouse? Your parents or
relatives, yourself, or both?” We coded our measure of participation in spouse choice as “1” if a
respondent answered “parents or relatives”, and “5” if the respondent answered “myself.” If the
respondent answered with “both,” he or she was asked the follow-up question, “Although both of
you may have decided, one of you may have had a little more influence than the other. Who had
more influence in choosing your spouse? You, or your parents and relatives?” If the respondent
answered with “parents or relatives,” the response was coded as “2,” if he or she responded with
17
“myself,” the response was coded as “4,” and if the respondent answered with “both equally,”
the response was coded as “3.” This established an ordinal scale of 1 to 5 in which “1” means the
respondent did not participate in the choice of his or her spouse (arranged marriage), and “5”
means the respondent solely chose his or her spouse (individual choice marriage). We then
recoded the measure into a dummy variable: “1” if the individual had any say in their spouse
selection and “0” if not.
Analytical Strategy
Our analytical strategy estimated the independent and interrelated relationships between
the spread of mass education and domestic violence against wives. Our models of mass
education were guided by the life course perspective, starting with models of childhood
community effects, followed by parental education, and finally the couple’s own educational
experiences. We estimated multivariate models using standard logistic regression procedures.
Logistic regression is an appropriate statistical technique for analyzing multivariate models of
dichotomous dependent variables (Kmenta, 1986).
Results
Childhood community educational context
Table 2 presents our estimates of the association between wives’ and husbands’
childhood community educational context and incidence of domestic violence against wives.
Coefficients in the tables are the odds ratio—exponentiated log-odds of raw coefficients, which
can be interpreted as the multiplicative effect of a one-unit change in the independent variable on
the odds of domestic violence against wives.
We found that wives’ childhood community educational context has a strong negative
and statistically significant relationship with the odds of domestic violence against wives. The
18
odds multiplier of 0.75 in Model 2indicates that having a school within a one-hour walk from
wives’ place of residence during childhood reduces the odds of domestic violence against wives
by 25%. Husbands’ childhood community educational context is also negatively associated with
the odds of domestic violence against wives, but this association is not statistically significant. In
Model 3, we combined husbands’ and wives’ childhood community educational context into a
single measure of the couple’s childhood community educational context. As expected, the
combined negative association became even stronger than the individual association, and is
statistically significant. The odds multiplier of 0.65 in Model 4 indicates that having a school
within a one hour-walk during the couple's childhood (both wife and husband) is associated with
a 35% reduction in the odds of domestic violence against wives.
(Table 2 About Here)
In terms of controls, both wives’ ethnicity and birth cohort are strongly associated with
the odds of domestic violence against wives. In general, compared to high caste Brahmin/Chhetri
women, women from other ethnic backgrounds have higher odds of reporting domestic violence
against wives. This difference, however, is not significant for women of Hill Janjati ethnicity.
Compared to women from the youngest cohort (born between 1972-81), women from older
cohorts have higher odds of experiencing domestic violence.
Parents’ education
Childhood community educational context is closely associated with the residential
choices that parents make. Because there is great cross-generational stability in place of
residence in a setting like rural Nepal, parental educational experience may both shape the
presence of schools within a one-hour walk and be shaped by the presence of schools within a
one-hour walk during childhood. We examine the independence of these two dimensions of mass
19
education directly by controlling for the childhood community educational context and
estimating the association of parents’ educations with the odds of domestic violence against
wives. First, we estimated the models of wife’s and husband’s parents separately and then we
estimated couple models.
(Table 3 About Here)
We found both wives’ and husbands’ parents’ education are each negatively associated
with the odds of domestic violence against wives. The odds multiplier of 0.69 in Model 1 of
Table 3 indicates that a wife having a parent ever attend school is associated with a 31%
reduction in the odds of experiencing domestic violence. This means that if both of the wife’s
parents ever went to school, she is more than 60% less likely to experience domestic violence—
an enormous association. Although husband’s parents’ education is also negatively associated
with the odds of domestic violence against wives, this association is not statistically significant.
When we estimated models of a couple's parents' education, the magnitude and statistical
significance of the association became even stronger. The odds multiplier of 0.61 in Model 3
indicates that a one-point increase in a couple’s parents’ education reduces the odds of domestic
violence against wives by 39%.
Although the effect of the couple’s childhood community educational context remained
negative and similar in magnitude to that of the final model in Table 2, it is no longer statistically
significant once parental education is included in the model. This finding is consistent with the
interpretation that part of the effect of the couple’s childhood community educational context
works through their parents’ education, particularly the wife’s parents’ education.
Wives’ and husbands’ educational experiences
20
Childhood access to schools and parental education are both key determinants of
children’s own educational experiences (Beutel & Axinn, 2002; Blau & Duncan, 1967; Marini,
1978). In a setting like Nepal, as is true in much of South Asia, parental education may also
shape the selection of a spouse, as the majority of marriages are arranged by parents (Ghimire et
al., 2006). Therefore, we have strong reasons to anticipate that the effects of childhood access to
schools and parental education may affect domestic violence against wives via women’s and
their husbands’ schooling. To investigate this, we use the same models discussed above, now
controlling for couple’s childhood educational context, parents’ educations, and couple’s marital
experiences, to estimate effects of wives’ and husbands’ educational experiences. In addition,
because both husband's and wife's age at marriage and education are intricately linked with each
other, we also control for difference in age at first marriage between the husband and wife.
Model 1 through Model 3 of Table 4 present estimates of wives’ and husbands’
educations. We found that a wife’s educational attainment is negatively associated with the odds
of domestic violence against wives. The odds multiplier of 0.33 for educational attainment in
Model 1 of Table 4 indicates that compared to women with no education before marriage,
women with a college education before marriage are more than 67% less likely to experience
domestic violence by husbands during their marriage.
(Table 4 About Here)
Similarly, we found that husband’s education is also strongly, negatively associated with
domestic violence against wives. The odds multiplier of 0.29 for those who had a college
education (Model 2,Table 4) indicates that compared to those women who have husbands with
no education before marriage, women who have husbands with a college education before
marriage are more than 71% less likely to experience domestic violence by their husbands during
21
marriage. In Model 3, Table 4, we include both the wife’s and husband’s education. As shown in
Model 3, the odds multiplier for wife with a college education increased from 0.33 in Model 1 to
0.62 and is no longer statistically significant. The odds multiplier for husband with a college
education also slightly increased from 0.29 in Model 2, to 0.34 in Model 3, however this
association remains statistically significant. Given educational assortative marriage in general
(Mare, 1991) and arranged marriage in this setting that may exacerbate that sorting (Ghimire et
al., 2006), it is not surprising to find that the consequences of husbands’ and wives’ educations
are not independent. Within this lack of independence, it is the potential perpetrator’s
educational attainment that directly affects the likelihood of domestic violence, not the potential
victim’s educational attainment. Childhood community educational context, parental
background, and women’s educational attainment all increase the likelihood that a woman will
marry a highly educated man (Ghimire et al., 2006; Thornton et al., 2007), but with all the other
dimensions of education controlled, it is the husband’s own education that most significantly
reduces the chances that he will perpetrate domestic violence against his wife.
These findings are also net of couples’ marital experiences. As hypothesized, the odds
multiplier of 0.97 for difference in husband’s and wife’s age at first marriage suggests that
husband age at first marriage is negatively associated with the odds of domestic violence against
his wife, with those who marry later being less likely to perpetrate domestic violence against
their wives. Though the odds multipliers for couples’ childhood community educational context
and wife’s parents’ education remained negative, none of these remain statistically significant.
These results reinforce the interpretation that part of the effects of the couple’s childhood
community access to schools and parents’ educations works through their own education,
particularly husbands’ education. Though wives’ parents’ educations had a substantial and
22
significant effect reducing the likelihood of experiencing domestic violence by her husband, the
mechanism for this intergenerational influence may be the choice of a husband with high
educational attainment, rather than a direct effect of parental education itself.
Understanding Historical Trends
There have been historical changes in the incidence of domestic violence against wives in
this setting. This can be seen in the results by cohort in Table 2. More recent cohorts of married
women are substantially less likely to experience domestic violence than the earliest cohorts.
As shown in Model 1 of Table 2, in an equation containing only birth cohort and
ethnicity, we found that women born between 1936-51 were 86% more likely to experience
domestic violence from their husbands than those born between 1972-81. With the introduction
of wife's childhood community educational context, the odds multiplier for the 1936-51 birth
cohort was reduced to 1.64. These cohort difference continue to decline as we add more
dimensions of the dramatic changes in mass education to our models. By the final model
presented in Table 4, including all the dimensions of mass education we measure, the difference
by cohort are virtually entirely gone. The difference is stunning—rarely does empirical evidence
provide such a clear explanation for cohort change over time. Virtually the total trend in
domestic violence against wives can be explained by the spread of mass education, as
conceptualized as access to community school during childhood, parental education, wife’s and
husband’s education. Thus our expectation that the spread of mass education played a significant
role in changing domestic violence against wives in Nepal is strongly supported by these data.
Discussion
Social theory predicts that the spread of mass education will significantly alter families,
family events, and family relationships (Caldwell, 1982; Goode, 1970; Thornton & Lin, 1994;
23
Thornton, 2005). Theory identifies mechanisms through which these macro-level changes work
to delay marriage, delay childbearing within marriage, and limit the total number of children
parents have (Caldwell, 1982; Thornton & Lin, 1994). Empirical investigation of these changes
in rural Nepal is entirely consistent with these overall expectations (Axinn & Yabiku, 2001;
Ghimire & Axinn, 2010). Moreover, research has repeatedly demonstrated that education in
schools dominated by Western European-style curricula is among the most powerful forces
affecting family formation events and relationships (Caldwell et al., 1988; Hoelter et al., 2004;
Thornton & Lin, 1994).
Theories of change and variation in societal levels of domestic violence do not have the
same decades of empirical testing as theories of marriage timing and fertility. Nevertheless, there
are strong reasons to expect that multiple dimensions of mass education would influence the
incidence of domestic violence. Studies of family change in Nepal have clearly demonstrated
that the spread of mass education produces delays in marriage (Yabiku, 2005), a transition away
from arranged marriage toward individual choice of spouse (Ghimire et al., 2006), and an
increase in positive relations between husbands and wives (Hoelter et al., 2004). These changes
in the nature of marital relationships have the potential to profoundly change the likelihood that
husbands strike their wives. The spread of school buildings through communities can affect the
chances individuals attend school and mere exposure to them can alter attitudes and expectations
about appropriate behavior. Parental education has the potential to increase children’s education,
change parents’ expectations for their children’s behavior, and in an arranged marriage society
increase parental commitment to finding educated spouses for their children. Both women’s and
men’s education may alter their expectations of the marital relationship and change the way
24
husbands and wives treat each other in marriage. Each of these mechanisms may connect the
spread of mass education to a reduced likelihood that husbands hit their wives.
Our results are consistent with this possibility that multiple dimensions of mass education
reduce the likelihood of domestic violence against wives. Having schools close by during
women’s childhoods is strongly associated with a lower likelihood of experiencing domestic
violence. This association is greatly reduced when parental education is added to the model –
much of the association between access to school in childhood and likelihood of domestic
violence against wives may be due to parental education. Parental education also reduces
women's likelihood of experiencing domestic violence. The association between women’s
parents’ educations and women's likelihood of experiencing domestic violence is strong, but this
association is greatly reduced by adding women’s own educational attainment to the model, so
that much of the effect of parental education may be due to the education of their children.
Finally, both women’s and their husbands’ educations are each strongly associated with
likelihood husbands hit their wives, but these different dimensions of education are not entirely
independent. When both women’s and their husbands’ educations are included in the same
model, only the association between husbands’ education and wives likelihood of experiencing
domestic violence continues to be strong and statistically significant. Thus, though women’s
education is important, it may be important for the educational assortative marriages that connect
highly educated women to more highly educated husbands. These findings support our
hypothesis that in a predominantly patriarchal society such as Nepal, the spread of mass
education is likely to be associated with less domestic violence against wives through multiple
mechanisms. Equally important, these findings provide a substantial advance in our insight into
the interrelationships among these multiple complex mechanisms. Finally, the evidence we
25
present here also shows that the reduction in domestic violence across cohorts is largely
explained by the spread of mass education.
Domestic violence is associated with poor mental and physical health, lower quality
marital relationships, higher marital dissolution, and poor economic performance (Allendorf,
2007; Kishor & Johnson, 2004; Rijal, 2003; UN, 1995; Weiss, 1996). Although our investigation
is limited to cross-sectional measures from one point in time and a single setting, our findings
constitute an important advance in family theory and research on domestic violence. Just as in
other domains of family change, our findings indicate that multiple dimensions of spread of mass
education simultaneously shape domestic violence against wives. Our findings for this setting
also demonstrate the strong effects of mass education explain significant change across birth
cohorts in the incidence of domestic violence against wives. We argue future research on the
relationship of education to family change should consider multiple dimensions of mass
education. Though we study a narrow element of domestic violence against wives – physical
violence – our results are consistent with our argument that studies of other dimensions of
domestic violence would also benefit from this multidimensional conceptualization of education.
Of course our data come from one region in Nepal and cannot be used for inference to
other Nepalese populations or populations of other countries. Diversity across populations may
limit the extent to which we should expect similar associations elsewhere. However, evidence
from multiple settings shows a powerful influence of spread of Western-oriented mass education
on other aspects of family and marital relationships (Allendorf and Ghimire Forthcoming;
Caldwell & Caldwell 1982; Holter 2004). Thus we have substantial reason to expect similar
consequences of mass education in other settings. Studies of these associations in a broader
range of populations remains a high scientific priority.
26
References
Acharya, M., & Bennett, L. (1981). Rural women of Nepal: An aggregate analysis and summary
of 8 village studies. Kathmandu: Centre for Economic Development and Administration.
Adhikari, D. P. (1998). The history of Nepalese nationalism. Nayabazar, Kathmandu: Jeewan
Printing Support Press.
Ahmad, F., Riaz, S., Barata, P., & Stewart, D. E. (2004). Patriarchal beliefs and perception of
abuse among South Asia immigrant women. Violence Against Women, 10, 262–282.
DOI:10.1177/1077801203256000
Alexander, J. C. (1988). Action and its environments: Toward a new synthesis. New
York: Columbia University Press.
Allendorf, K. (2007). Couples’ report of women’s autonomy and health-care use in
Nepal. Studies in Family Planning, 38, 35–46. DOI:10.1111/j.1728-4465.2007.00114.x
Allendorf, K., & Ghimire D. J. Forthcoming. Dimensions and Determinants of Marital
Quality in Nepal. Social Science Research.
Axinn, W. G., & Barber, J. S. (2001). Mass education and fertility transition. American
Sociological Review, 66, 481–505.
Axinn, W. G., Pearce, L. D., & Ghimire, D. J. (1999). Innovations in life history calendar
applications. Social Science Research, 28, 243–264. DOI:10.1006/ssre.1998.0641
Axinn, W. G., & Thornton, A. (1993). Mothers, children, and cohabitation: The intergenerational
effects of attitudes and behavior. American Sociological Review, 58, 233–246.
Axinn, W. G., & Yabiku, S. T. (2001). Social change, the social organization of families, and
fertility limitation. American Journal of Sociology, 106, 1219–1261.
DOI:10.1086/320818
27
Bagheri, M., Nabavi, A., & Hosseinzadeh, A. H. (2009). Prevalence of Marital Violence in the
Family. The Indian Journal of Social Work, 70, 117–132.
Barber, J. S., & Axinn, W. G. (2004). New ideas and fertility limitation: The role of mass media.
Journal of Marriage and Family, 66, 1180–1200. DOI:10.1111/j.0022-
2445.2004.00086.x
Barber, J. S., Shivakoti, G. P., Axinn, W. G., & Gajurel, K. (1997). Sampling strategies for rural
settings: A detailed example from the Chitwan Valley Family Study Nepal. Nepal
Population Journal, 6, 193–203.
Bengston, V. L. (1975). Generation and family effects in value socialization. American
Sociological Review, 40, 358–371.
Berreman, G. D. (1972). Hindus of the Himalayas. Berkeley: University of California Press.
Beutel, A. M., & Axinn, W. G. (2002). Gender, social change, and educational attainment.
Economic Development and Cultural Change, 51, 109–135.
Bista, D. B. (1972). People of Nepal. Kathmandu: Ratna Pustak Bhandar.
Blau, P. M., & Duncan, O. D. (1967). The American occupational structure. New York:Wiley.
Brauner-Otto, S. R. Forthcoming. Schools, their spatial distributions and characteristics, and
fertility limitations. Rural Sociology.
Cain, M., Khanam, S. R., & Nahar, S. (1979). Class, patriarchy, and women’s work in
Bangladesh. Population and Development Review, 5, 405–438.
Caldwell, J. C. (1982). Theory of fertility decline. London: Academic Press Inc.
Caldwell, J. C., Reddy, P. H., & Caldwell, P. (1988). The causes of demographic change:
Experimental research in South India. Madison: University of Wisconsin Press.
Carvalho, N. (2007, March 5). Rising domestic violence against women. AsiaNet.It. Retrieved
28
from http://www.asianews.it/news-en/Rising-domestic-violence-against-women-
8651.html
Cleland, J., & Rodriguez, G. (1988). The effect of parental education on marital fertility in
developing countries. Population Studies, 42, 419–442.
Coleman, J. S. (1990). Social capital in the creation of human capital. American Journal of
Sociology, 94, 95–120.
Diop-Sidibe, N., Campbell, J. C., & Becker, S. (2006). Domestic violence against women in
Egypt—wife beating and health outcomes. Social Science and Medicine, 62, 1260–1277.
Devries, K., Watts C., Yoshihama, M., Kiss, L., Schraiber, L. B., Deyessa, N., Heise, L., Durand,
J., Mbwambo, J., Jansen, H., Berhane, Y., Ellsberg, M., Garcia-Moreno, C., & WHO
Multi-Country Study Team. (2011). Violence against women is strongly associated with
suicide attempts: Evidence from the WHO multi-country study on women’s health and
domestic violence against women. Social Science and Medicine, 73, 79–86.
Durkheim, E. (1984). The division of labor in society. New York: Free Press. (Original work
published in 1893)
Dyson, T., & Moore, M. (1983). On kinship structure, female autonomy, and demographic
behavior in India. Population and Development Review, 9, 25–60.
Elder, G. H. (1985). Life course dynamics: Trajectories and transitions, 1968-1980. Ithaca, NY:
Cornell University Press.
Ellsberg, M., Jansen, H. A., Heise, L., Watts C. H., & García-Moreno, C. 2008. Intimate partner
violence and women’s physical and mental health in the WHO multi-country study on
women’s health and domestic violence: an observational study. Lancet, 371, 1165–1172.
English, R. (1985). Himalayan state formation and the impact of British rule in the
29
Nineteenth century. Mountain Research and Development, 5, 61–78.
Gecas, V., & Seff, M. A. (1990). Families and adolescents: A review of the 1980s. Journal of
Marriage and the Family, 52, 941–958.
Ghimire, D. J., & Axinn, W. G. (2010). Community context, land use and first birth. Rural
Sociology, 75, 478–513. DOI:10.1111/j.1549-0831.2010.00019.x
Ghimire, D. J., Axinn, W. G., Yabiku, S. T., & Thornton, A. (2006). Social change, premarital
non-family experiences and spouse choice in an arranged marriage society. American
Journal of Sociology, 111, 1181–1218. DOI:10.1086/498468
Goode, W. J. (1970). World revolution and family patterns. New York: The Free Press.
Heise, L. L., Ellsberg, M. C., & Gottemoeller, M. (1999). Ending violence against women.
(Population Reports Series L, No 11). Baltimore: Johns Hopkins University School of
Public Health, Population Information Program.
Hoelter, L. F., Axinn, W. G., & Ghimire, D. G. (2004). Social change, premarital non-family
experiences, and marital dynamics. Journal of Marriage and Family, 66, 1131–1151.
DOI:10.1111/j.0022-2445.2004.00083.x
Jamison, D. T., & Lockheed, M. E. (1987). Participation in schooling: Determinants and learning
outcomes in Nepal. Economic Development and Cultural Change, 35, 279–306.
Johnson, M. P. (1995). Patriarchal terrorism and common couple violence: Two forms of
violence against women. Journal of Marriage and the Family, 57, 283–294.
Khaniya, T. R., & Kiernan, M. A. (1994). Nepal: System of education. In T. Husen and T.N.
Postlethwaite (Eds.), The International Encyclopedia of Education (2nd ed.)
(pp.40604067). Oxford: Elsevier Science Ltd.
Kishor, S., & Johnson, K. (2004). Domestic violence in nine developing countries:
30
A comparative study. Calverton: ORC Macro International.
Kmenta, J. (1986). Elements of Econometrics (2nd ed.). New York: Macmillan Publishing
Company.
Koenig, M. A., Ahmed, S., Hossain, M. B., & Mozumder, A. B. M. K. A. (2003). Women's
status and domestic violence in rural Bangladesh: Individual and community-level
effects. Demography, 40, 269–288.
Koenig, M. A., Stephenson, R., Ahmed, S., Jejeebhoy, S. J., & Campbell, J. (2006). Individual
and contextual determinants of domestic violence in North India. American Journal of
Public Health, 40, 269–288. DOI:10.2105/AJPH.2004.050872
Link, C. F. (2011). Spousal communication and contraceptive use in rural Nepal: An event
history analysis. Studies in Family Planning, 42, 83–92. DOI:10.1111/j.1728-
4465.2011.00268.x
Macfarlane, A. (1976). Resources and population: A study of the Gurungs of Nepal.
Cambridge: Cambridge University Press.
Macfarlane, A. (1986). Marriage and love in England: Methods of reproduction 1300-1840.
Oxford: Blackwell Publishers.
Majumdar, S. (2003, November 6). In India domestic violence rises with education. Women’s
eNews. Retrieved fromhttp://www.womensenews.org/story/domestic-
violence/031106/india-domestic-violence-rises-education
Mare, R. D. (1991). Five decades of educational assortive mating. American Sociological
Review, 56,15–32.
Marini, M. M. (1978). The transition to adulthood: Sex differences in educational attainment and
age at marriage. American Sociological Review, 43, 483–507.
31
Mason, K. O. (1987). The impact of women’s social position on fertility in developing countries.
Sociological Forum, 1, 718–745. DOI:10.1007/BF01124382
Mathur, K. (2004). Countering gender violence: Initiatives towards collective action in
Rajasthan. New Delhi: Sage Publications India Pvt Ltd.
Naved, R. T., & Akhtar, N. (2008). Spousal violence against women and suicidal ideation in
Bangladesh. Women's Health Issues, 18, 442-452.
Naved, R. T., Azim, S.,Bhuiya, A., & Persson, L. A. (2006). Physical violence by husbands:
Magnitude, disclosure, and help seeking behavior of women in Bangladesh. Social
Science and Medicine, 62, 2917-2929. DOI:10.1016/j.socscimed.2005.12.001
Panday, D. R. (1999). Nepal's failed development: Reflections on the mission and the maladies.
Kathmandu: Nepal South Asia Centre.
Panday, G.K., Dutt, D., & Banerjee, B. (2008). Partner and relationship factors in
domestic violence: Perspective of women from a slum in Calcutta, India. Journal of
Interpersonal Violence, 30, 1–17. DOI:10.1177/0886260508322186
Pradhananga, R. B., & Shrestha, P. (2009). Domestic violence against women in Nepal: Concept,
history and existing laws. Working paper.
Rauschning, D., Wiesbrock, K., & Lailach, M. (Eds.) (1997). Key Resolutions of the United
Nations General Assembly, 1946–1996. Cambridge, UK: Cambridge University Press.
Regmi, R.R. (1999). Dimensions of Nepali society and culture. SAAN Research Institute,
Gairidhara, Kathmandu.
Rijal, Sanjita. (2003, April 1). Domestic violence: A major cause of increasing divorce cases.
The Kathmandu Post.
Rindfuss, R. R., & Morgan, S. P. (1983). Marriage, sex and first birth interval: The quiet
32
revolution in Asia. Population and Development Review, 9, 259–278.
Sharma, G. (1994). Development of educational administration in Nepal. Kathmandu: Indu
Chhapakhana Pvt. Ltd.
Smith, D. S. (1973). Parental power and marriage patterns: an analysis of historical trends in
Hingham, Massachusetts. Journal of Marriage and the Family, 25, 419–428.
Smith, T. E. (1988). Parental Control Techniques: Relative Frequencies and Relationships with
Situational Factors. Journal of Family Issues, 9, 155–176.
Srinivas, M. N. (1952). Religion and society among the Coors of South India. Oxford, UK:
Clarendon Press.
Stash, S., & Hannum, E. (2001). Who goes to school? Educational stratification by gender,
caste, and ethnicity in Nepal. Comparative Education Review, 45, 354–378.
DOI:10.1086/447676
Stephenson, R., Koenig, M. A., & Ahmed, S. (2006). Domestic violence and symptoms of
gynecologic morbidity among women in North India. International Family Planning
Perspectives, 32, 201–208.
Thapa, S. (1997). Timing of family formation in the ethnic mosaic Nepal: A district-level
analysis. Asia-Pacific Population Journal, 12, 75–87.
Thornton, A. (2005). Reading history sideways: The fallacy and enduring impact of the
developmental paradigm on family life. Chicago: University of Chicago Press.
Thornton, A., Axinn, W. G., & Xie, Y. (2007). Marriage and cohabitation. Chicago: University
of Chicago Press.
Thornton, A., & Lin, H.-S. (1994). Social change and the family in Taiwan. Chicago: University
of Chicago Press.
33
United Nations. (1993). Declaration on the Elimination of Violence Against Women. General
Assembly Resolution 48/104, 20 December 1993. United Nations, NY.
United Nations. (1995). Report of the Fourth World Conference on Women. Beijing, September
4-15, 1995. United Nations, NY.
Weiss, L. J. (1996). Women alone: The causes and consequences of non-marriage and marital
disruption among high caste Hindus in Nepal. (Unpublished doctoral dissertation)
Columbia University, New York.
World Health Organization. (2001). Putting women first: Ethical and safety
recommendations for research on domestic violence against women. Geneva
Switzerland: Department of Gender and Women’s Health.
Yabiku, S. T. (2005). The effect of non-family experiences on age of marriage in a setting of
rapid social change. Population Studies, 59, 339–354. DOI:10.1080/00324720500223393
Yount, K. M. (2005). Resources, family organization, and domestic violence against married
women in Minya, Egypt. Journal of Marriage and Family, 67, 579–596.
DOI:10.1111/j.1741-3737.2005.00155.x
Yount, K. M., & Carrera, J. S. (2006). Domestic violence against married women in Cambodia.
Social Forces, 85, 355–387. DOI:10.1353/sof.2006.0151
Yount, K. M., Halim, N., Hynes, M., & Hillman, E. R. (2011). Response effects to attitudinal
questions about domestic violence against women: A comparative perspective. Social
Science Research, 40, 873–884. DOI:10.1016/j.ssresearch.2010.12.009
34
Table 1: Descriptive Statistics of Measures used in the Analyses (N=1,778 Currently Married in 1996)
1Note: For husband this cohort include all age 45 and above.
Wives Husbands
Coding Mean S. Dev. Min. Max. Mean S. Dev. Min. Max.
Childhood community educational context
Had a school within one-hour walk 0=no, 1=yes 0.82 0.38 0 1 0.81 0.39 0 1
Parents’ education
Parents ever went to school 0=none, 1=either, 0.31 0.53 0 2 0.19 0.43 0 2
2=both
Premarital education
Schooling Years 3.11 4.50 0 20 5.94 5.41 0 22
No education 0=no, 1=yes 0.57 0.49 0 1 0.31 0.46 0 1
1-5 years (Primary)
0=no, 1=yes 0.15 0.36 0 1 0.18 0.38 0 1
6-11 year (High school) 0=no, 1=yes 0.21 0.40 0 1 0.34 0.47 0 1
12 and more year (College) 0=no, 1=yes 0.07 0.26 0 1 0.17 0.37 0 1
Marital experiences
Age at first marriage Years 16.25 3.23 5 36 20.65 4.33 7 45
Married at age 15 and younger 0=no, 1=yes 0.41 0.49 0 1 0.10 0.30 0 1
Married between 16 and 20 0=no, 1=yes 0.50 0.50 0 1 0.41 0.49 0 1
Married after 20
0=no, 1=yes 0.08 0.27 0 1 0.49 0.50 0 1
Participation in spouse choice
0=no, 1=yes 0.24 0.43 0 1 0.51 0.50 0 1
Controls
Age in years Years 33.60 10. 88 15 59 39.23 12.45 17 80
Birth Cohort
Born bet. 1972–81 (age 15–24, Cohort0) 0=no, 1=yes 0.25 0.43 0 1
Born bet. 1962–71 (age 25–34, Cohort1) 0=no, 1=yes 0.30 0.46 0 1
Born bet. 1952–61 (age 35–44, Cohort2) 0=no, 1=yes 0.25 0.43 0 1
Born bet. 1936–51 (age 45–59, Cohort3) 0=no, 1=yes 0.20 0.39 0 1
Ethnicity
Brahmin/Chhetri 0=no, 1=yes 0.47 0.50 0 1
Dalit 0=no, 1=yes 0.12 0.32 0 1
Hill Janjati 0=no, 1=yes 0.17 0.38 0 1
Newar 0=no, 1=yes 0.07 0.24 0 1
Terai Janajati 0=no, 1=yes 0.17 0.39 0 1
Outcome
Ever beaten by husband 0=no, 1=yes 0.17 0.37 0 1
35
Table 2: Logistic Regression Estimates of Association Between Childhood Community Educational Context and Domestic Violence Against Wives (N=1,778)
Note: * P<.05, ** P<.01; all probabilities are one-tailed. Odds ratios are reported on the first line, with Z statistics in the parentheses on second line. All models were estimated using multivariate logistic regression. † Born 1972-81 (age 15-24 Cohort 0) as reference group. †† Brahmin/Chhetri as reference group.
Model
1 2 3 4 Childhood community educational context Wife had a school within one-hour walk 0.75*
(-1.67) Husband had a school within one-hour walk 0.81 (-1.17) Couple had school within one-hour walk 0.65* (-1.86) Controls Birth cohort† Born between 1962 – 71 (Cohort1) 1.19 1.17 1.17 1.16
(0.92) (0.82) (0.85) (0.79) Born between 1952 – 61 (Cohort2) 1.54* 1.47* 1.48* 1.42*
(2.27) (2.00) (2.03) (1.81) Born between 1936 – 51 (Cohort3 ) 1.86** 1.64** 1.66** 1.51* (3.22) (2.38) (2.34) (1.83) Ethnicity†† Dalit 2.58** 2.51** 2.60** 2.55** (4.87) (4.72) (4.91) (4.81) Hill Janjati 1.29 1.25 1.27 1.24 (1.34) (1.16) (1.25) (1.12) Newar 1.74* 1.74* 1.77* 1.74* (2.15) (2.14) (2.20) (2.16) Terai Janjati 1.95** 1.89** 1.95** 1.90** (3.88) (3.66) (3.87) (3.70) Intercept 0.10 0.14 0.13 0.16 (-13.58) (-8.28) (-8.58) (-6.53) - 2 Log L
1559 1556 1558 1556 Df 7 8 8 8 N 1778 1778 1778 1778
36
Table 3: Logistic Regression Estimates of Association Between Parents’ Education on Domestic Violence Against Wives (N=1,778)
Note: * P<.05, ** P<.01; all probabilities are one-tailed. Odds ratios are reported on the first line, with Z statistics in the parentheses on second line. All models were estimated using multivariate logistic regression. † Born 1972-81 (age 15-24 Cohort 0) as reference group. †† Brahmin/Chhetri as reference group.
Model
1 2 3 Parents’ education Wife’s parents ever went to school 0.69** (-2.45) Husband’s parents ever went to school 0.87 (-0.82) Couple parents' ever went to school 0.61* (-2.24) Childhood community educational context Couple had School within one-hour walk 0.78 0.82 0.69 (-1.49) (-1.12) (-1.62) Controls Birth cohort† Born between 1962 – 71 (Cohort1) 1.09 1.15 1.07
(0.45) (0.74) (0.36) Born between 1952 – 61 (Cohort2) 1.31 1.43* 1.25
(1.36) (1.84) (1.10) Born between 1936 – 51 (Cohort3 ) 1.46* 1.60** 1.33 (1.80) (2.13) (1.22) Ethnicity†† Dalit 2.35** 2.55** 2.36** (4.32) (4.77) (4.34) Hill Janjati 1.19 1.26 1.19 (0.92) (1.22) (0.90) Newar 1.73* 1.82* 1.78* (2.11) (2.30) (2.22) Terai Janjati 1.75** 1.91** 1.75** (3.17) (3.74) (3.18) Intercept 0.17 0.13 0.19 (-7.22) (-8.19) (-5.69) - 2 Log L 1549 1556 1550 df 9 9 9 N 1778 1778 1778
37
Table 4: Logistic Regression Estimates of Association Between Wife and Husband’s Education and Domestic Violence Against Wives (N=1,778)
Model 1 2 3 Premarital education
Wife’s education ††† With 1- 5 years of education 1.04 1.15 (0.21) (0.65) With 6- 11 years of education 0.68* 0.94 (-1.72) (-0.26) With 12 and more years of education 0.33** 0.62 (-2.40) (-0.99) Husband’s education††† With 1- 5 years of education 1.03 1.07 (0.16) (0.35) With 6- 11 years of education 0.66* 0.70* (-2.20) (-1.83) With 12 and more years of education 0.29** 0.34** (-4.04) (-3.24) Parents’ education Wife’s parents ever went to school 0.76* (-1.76) Husband’s parents ever went to school 1.04 (0.20) Couple's parents ever went to school 0.81 (-0.88) Childhood community educational context Wife had a school within one-hour walk 0.80 (-1.33) Husband had a school within one-hour walk
0.90 (0.54) Couple had a school within one-hour walk 0.76 (-1.14) Individual experience Marital experience Wife’s age at first marriage †††† Married between age 16 and 20
1.15 (1.01) Married after age 20 0.90 (-0.37) Husband’s age at first marriage †††† Married between age 16 and 20
0.72 (-1.61) Married after age 20 0.54**
(-2.99) Wife’s age at first marriage (in years) 0.98 (-0.65) Difference in age at first marriage (husband-Wife) 0.97* (-1.86) Wife participation in husband selection
1.15 1.04
(0.89) (0.20) Husband participation in wife selection
0.94
(-0.44) Difference in participation in spouse choice (Husband-Wife)
0.92
38
Note: *P<.05, ** P<.01; all probabilities are one-tailed. Odds ratios are reported on the first line, with Z statistics in the parentheses on second line. All models were estimated using multivariate logistic regression. † Born 1972-81 (age 15-24 Cohort 0) as reference group. †† Brahmin/Chhetri as reference group. ††† No education as reference group. †††† Married at age 15 and before as reference group.
(0.58) Controls Birth cohort† Born between 1962 – 71 (Cohort1) 1.01 1.01 0.97 (0.02) (0.03) (0.17) Born between 1952 – 61 (Cohort2) 1.14 1.06 1.01 (0.62)
(0.26) (0.02) Born between 1936 – 51 (Cohort3 ) 1.29 1.06 1.01 (1.07)
(0.22) (0.02)
Ethnicity†† Dalit 2.04** 1.95** 1.80** (3.50) (3.28) (2.84) Hill Janjati 1.07 1.20 1.07 (0.34) (0.92) (0.31) Newar 1.70* 2.01** 1.83* (1.99) (2.63) (2.27) Terai Janajati 1.48* 1.41* 1.32 (2.04) (1.84) (1.41) Intercept 0.19 0.34 0.41 (-5.94) (-3.24) (-1.71) - 2 Log L 1538 1519 1521 Df 15 15 19 N 1777 1775 1776