State gender inequality, socioeconomic status and intimate partner violence (IPV) in India: a multilevel analysis.
While a growing body of literature has investigated the health
impact of intimate partner violence (IPV), less has been written on the
social determinants of IPV. The authors use multilevel modeling methods
to analyze data from a sample of 83,627 women in India to examine the
socioeconomic and demographic patterning of the state- and
neighborhood-level variation in, and the state- and/or
neighborhood-level characteristics related to, IPV. This study finds
social gradients in IPV in which women who are uneducated, from
marginalized castes, and living in poor households have higher
likelihood of reporting IPV than those living in advantaged
circumstances. The results also show differences in IPV between
neighborhoods and between states that are partially explained by state
levels of gender inequality. The results suggest that changing cultural
norms to promote the status of women and increasing the educational and
economic opportunities for all people could decrease the prevalence of
Key words: intimate partner violence, social determinants of health, India
Gender equality (Analysis)
Interpersonal relations (Analysis)
Ackerson, Leland K.
|Publication:||Name: Australian Journal of Social Issues Publisher: Australian Council of Social Service Audience: Academic Format: Magazine/Journal Subject: Sociology and social work Copyright: COPYRIGHT 2008 Australian Council of Social Service ISSN: 0157-6321|
|Issue:||Date: Autumn, 2008 Source Volume: 43 Source Issue: 1|
|Geographic:||Geographic Scope: India Geographic Code: 9INDI India|
Intimate partner violence (IPV) is an important public health concern that has recently received much attention, deservedly so (Garcia-Moreno et al. 2006; Heise & Garcia-Moreno 2002; Watts & Zimmerman 2002). While the negative health consequences of IPV include obvious outcomes such as trauma and murder, the impact is much broader influencing mental illness, psychosomatic illness, poor health-related behaviors, poor birth outcomes, suicide, and diseases such as asthma and gynecological morbidities (Ackerson et al. 2007; Campbell 2002; Heise & Garcia-Moreno 2002; Kumar et al. 2005; Stephenson et al. 2006; Subramanian et al. 2007; Sudha et al. 2007; Vizcarra et al. 2004). A recent WHO-sponsored multi-national study found that in each of 15 sites in Bangladesh, Brazil, Ethiopia, Japan, Namibia, Peru, Samoa, Serbia and Montenegro, Thailand, and Tanzania, at least 15% of women experience physical or sexual IPV and in one site (Ethiopia) the estimate was over 70% (Heise & Garcia-Moreno 2002). Prevalence of physical IPV among women in India has been reported to be approximately 40% (Kumar et al. 2005) which would place it among the countries in the study with higher rates of violence.
By its nature, IPV is a difficult issue to study as measurement of IPV relies on abusers, the abused, or some close witness to report what is often conceived as a highly personal and stigmatized condition. Research in societies marked by relative gender equality may conceptualize IPV in terms of sexual violence, such as unwanted sexual activity (Vest et al. 2002), or psychological violence, such as controlling a woman's daily activities (Hathaway et al. 2000). In societies marked by gender inequality, including India (Krishnan, 2005a, 2005b), such measures may be seen of less value in IPV research. Additionally, in India, where a substantial portion of men believe they are justified in physically hurting their wives in certain circumstances (Martin et al. 2002), research using men's reports of their own abuse against women may not be accurate. This notion is further supported by research showing that Indian men are consistently more likely to report that their wives have high levels of autonomy, a characteristic closely related to reduced levels of domestic violence, than are the wives themselves (Jejeebhoy 2002).
Research over the last decade has emphasized the importance of socioeconomic circumstances and contexts for individual health (Diez Roux 2001; Kawachi & Subramanian 2007; Subramanian et al. 2002). There is currently a large body of empirical research that shows an association between socioeconomic status and several health outcomes (Braveman et al. 2000; Braveman et al. 2005; Epel et al. 2004; Marmot 2002, 2007). Similarly, impoverished neighborhood conditions have also been shown to lead to poorer health outcomes (Berkman & Kawachi 2000; Diez Roux 2001; Kawachi & Berkman 2003; O'Campo 2003; Pickett & Pearl 2001; Sampson 2003). Even though the importance of socioeconomic status has been documented in qualitative investigations and community-specific studies of IPV in India (Koenig et al. 2006; Krishnan 2005b; Martin et al. 2002), few systematic analyses have investigated the potential impact of socioeconomic status on IPV in India.
In addition to socioeconomic status, demographic characteristics such as caste and age at marriage are theoretically motivated predictors of IPV. While public discrimination against individuals according to caste is prohibited by the Indian Constitution, discriminatory social practices continue to be practiced (Krishnan 2005a). Although programs to alleviate the historical discrimination experienced by legally-defined scheduled castes, scheduled tribes, and other backward classes have had some impact, the majority of individuals in these groups have lower living standards, less access to education and employment, and in the case of tribes, tend to live geographically and socially separated from members of the general classes that control the majority of India's economic and governmental institutions (Bhengra et al. 1999; Chimis 1997; Subramanian et al. 2006). For an Indian woman, age at marriage can be seen as a combination of how much power she wields within a family (Nath et al. 1999), as well as how prepared she is to manage a household (Martin, Tsui et al. 1999).
Little evidence exists on the question of whether geographic variations in IPV are mere reflections of the distribution of individual correlates of IPV (i.e. a compositional based explanation of the geographic variation in IPV) or whether the geographic differences exist independent of the distribution of individual determinants (i.e. a contextual interpretation of the geographic variation in IPV) (Kawachi & Subramanian 2007; Moon et al. 2005; Subramanian et al. 2002; Subramanian et al. 2007). Given the influence that context has been shown to exert on gender roles in India (Dyson & Moore 1983), and that previous research has found that community measures of autonomy and wife-beating norms have been associated with IPV risk in South Asia (Koenig et al. 2003; Koenig et al. 2006), it is reasonable to believe that other state and community characteristics may be associated with IPV prevalence as well. Since gender inequality has been found to predict health outcomes in other national surveys (Chen et al. 2005; Jun et al. 2004; Kawachi et al. 1999; Koenen et al. 2006; Pallitto & O'Campo 2005), we hypothesize that state-level gender inequality in India will be related to IPV. In this study, we use a large, nationally-representative, multilevel dataset to investigate the role of socioeconomic status and multiple residential contexts (operationalized at the neighborhood and state levels) on women's likelihood of experiencing IPV in India.
The 1998-99 Indian National Family Health Survey (INFHS) which contains information from 92,447 households (IIPS & ORC-Macro 2000) was utilized to understand the patterns and distribution of IPV in India. An adult member in each selected household reported basic demographic information about the household and its family members. From these households, 90,303 ever-married women aged 15-49 were selected who provided information about maternal and child health characteristics in face-to-face interviews. The response rates were 98% for the household survey and 96% for the women's survey. The current analysis was restricted to currently married women with complete data on IPV and socioeconomic and demographic characteristics yielding a final analytic sample of 83,627 women. These women resided in 3,215 primary sampling units (neighborhoods) in the 26 Indian states.
Two binary outcome measures were created for this analysis. Lifetime-IPV and recent-IPV measured whether a woman had been physically abused by her husband since age 15 or in the previous 12 months, respectively.
We included several socioeconomic and demographic variables in our analyses that have previously been shown to be related to IPV. These include age, age at marriage, religion, social caste, woman's education, husband's education, spousal education differential, household standard of living, employment status, and urban-rural status (Table 1). Age was specified in five year categories, and age at marriage was grouped as <15 years, 15-17 years, 18-20 years, and >20 years. Religion was grouped as Hindu, Muslim, Christian, Sikh, or Other. Caste was based on the identification of the head of the household as belonging to any of three socially marginalized groups, scheduled caste, scheduled tribe, and other backward class, or the non-marginalized general class. Educational attainment for both the women and their husbands were specified in the following categories: 0 years, 1-5 years, 6-8 years, 9-10 years, 11-12 years, or 13 or more years. Spousal education differential was classified as one of three categories where either the husband had a higher educational milestone, the wife had a higher educational milestone, or the couple had educational parity. Standard of living, conceptualized in terms of living environment and material possessions (Filmer & Pritchett 2001), was based on a linear combination of the scores for 19 different household characteristics that were weighted according to a factor analysis procedure (Rutstein & Johnson 2004) and then divided into quintiles. Employment was classified according to whether the woman was not working, or working for pay in a manual, non-manual, or agricultural profession. Finally, information from the 1991 Indian National Census was used to create categories defining whether each neighborhood was in a large city (an urban area of over one million people), a small city (an urban area of between 100,000 and one million people), a town (an urban area of less than 100,000 people), or a village (a rural area).
Defining residential contexts
We defined geographic contexts at two levels: neighborhood and states. The Primary Sampling Unit (PSU) was used as a proxy for people's immediate residential/ neighborhood context. PSUs in the INFHS were primarily villages or clusters of villages in the rural areas, and wards in the context of urban areas. We also considered the additional macro context of state capturing the broader socioeconomic, political and cultural context within which individuals and neighborhoods operate.
State- and neighborhood-level predictors
We evaluated the relationship between an individual's likelihood of experiencing IPV and four contextual exposures: neighborhood wealth, per capita state domestic product (PCSGDP), state gender equality (SGE), and state human development (SHD). Given the cultural differences (Dyson & Moore 1983) and the differential implementation of economic and social policy (Baddeley et at. 2006; Soo 2007) evidenced between Indian states, we found it important to conceptualize our area exposures at the state level. In addition, the majority of neighborhoods in this dataset are villages, a level of governance in rural areas tasked with formulating and implementing local policies, besides being natural units of social organization (Besley et at. 2005; Pur 2007). To account for this as well as prior empirical evidence regarding the influence of local economic conditions on IPV risk (Benson et at. 2003; Cunradi et at. 2000; Miles-Doan & Kelly 1997; Pearlman et at. 2003) we included a neighborhood-level measure of socioeconomic status as well. The substantive interest was mainly in the association between state gender equality and likelihood of experiencing IPV, after adjusting for individual as well as other state- and neighborhood-level characteristics. Neighborhood wealth was calculated from the INHFS survey data by averaging the mean weighted linear household standard of living score for each neighborhood and then classifying neighborhoods into quartiles. PCSDP, SGD, and SHD were obtained from the 2001 National Human Development Report published by the Government of India (India 2002). PCSDP is an accepted marker of overall economic development analogous to figures for Gross National Product used in international studies. Figures for 1997-1998 PCSDP were used consistent with previous research (Subramanian & Davey Smith 2006). SHD was measured through a weighted combination of the following characteristics: life expectancy at age one, infant mortality rate, literacy rate for people aged seven and older, intensity of formal education, per capita consumption and worker population ratio. SGE was created by calculating the SHD separately for men and women and then dividing the women's SHD by the men's SHD to determine the ratio of women's human development to men's human development for each state. SGE and SHD were standardized to provide scores between 0 and 1 for each state. For the purposes of this study, analyses were conducted to determine the impact of a one standard deviation increase in each of these two measures.
Multilevel statistical modeling techniques were used to partition the variation in IPV to different levels. The substantive as well as technical relevance of multilevel statistical procedures are well-known (Blakely & Subramanian 2006; Goldstein 2003; Subramanian 2004; Subramanian et al. 2007; Subramanian et al. 2003). The study had a three level hierarchic structure with 83,627 individuals (level-1), within 3,215 neighborhoods (level-2), within 26 states (level-3). Given the hierarchical structure of the sample and the binary outcome, a logistic multilevel modeling approach was adopted (Goldstein 2003). We estimated three types of models of the following specification with a binary response (y, reported to have experienced IPV or not) for individual i living in neighborhood j in state k of the form: [[pi].sub.ijk] : [y.sub.ijk] ~ Bernoulli (1, [[pi].sub.ijk]. For model 1, the probability [[pi].sub.ijk] was related to an overall mean and a random effect for neighborhood and state level, by a logit link function as logit([[pi].sub.ijk]) = log([[pi].sub.ijk]/ (1 -[[pi].sub.ijk])) = [[beta].sub.0] + [u.sub.0jk] + [v.sub.0k] (Model 1). The parameter [[beta].sub.0] estimates the mean log odds of reporting IPV across the sample. Meanwhile, the parameters [u.sub.0jk] and [v.sub.0k] represent the random differentials (from the overall mean) at the neighborhood and state levels, respectively. These random differentials are assumed to have an independent and identical distribution with variance estimated at the neighborhood ([[sigma].sup.2.sub.u0]) and state ([[sigma].sup.2.sub.v0]) level. The variance parameters at Model 1 estimate the unconditional or unadjusted variation that is attributable to neighborhoods and states. To Model 1 we then added individual level covariates to the fixed part of the model as: logit([[pi].sub.ijk]) = log([[pi].sub.ijk]/(1 - [[pi].sub.ijk])) = [[beta].sub.0] + [beta]X + [u.sub.0jk] + [v.sub.0k] (Model 2), where [beta][X.sub.ij] represents the regression coefficients associated with a vector of individual level independent variables (X). Model 2 re-estimates the variance at the neighborhood ([[sigma].sup.2.sub.u0]) and states ([[sigma].sup.2.sub.u0]) level after adjusting for the compositional make-up of the neighborhoods and states. These estimates provide evidence for the presence of geographic variation in IPV that is over and above which that can be attributed to the distribution of observed individual factors. Additionally, we also estimated posterior residuals at the state level using Models 1 and 2 to explore the state-level variation of IPV, before and after accounting for the geographic distribution of individual covariates (Subramanian, Glymour et al. 2007). To model (2) we then added variables measured at the neighborhood level (Model 3), and PCSDP (Model 4), and then SGE and SHD (Model 5) such that: logit([[pi].sub.ijk]) = log([[pi].sub.ijk]/(1 - [[pi].sub.ijk])) = [[beta].sub.0] + [beta]X + [u.sub.0jk] + [v.sub.0k], with the parameter a giving an estimate of the change in the response for a unit change in me neighborhood and state variables. Penalized quasi-likelihood approximation with a second order Taylor linearization procedure was used to estimate all models (Rasbash et al. 2005).
Lifetime- and recent-IPV was reported by 16.1% and 9.3% of women respectively (Table 1). The majority of the sample resided in rural areas, identified as Hindu, and did not work outside of the home. Half of the women had never attended school and over half were in one of the three socially marginalized caste groupings. The sample was relatively evenly distributed in terms of age, age at wedding, living standard, and husband's education. Half of the women were less educated than their husbands, while two fifths experienced educational parity.
Socioeconomic status and IPV
The likelihood of experiencing IPV followed a distinct socioeconomic gradient, regardless of how socioeconomic status was measured (Table 2, Figure 1). Women from the poorest quintile of household standard of living were more likely to report lifetime-IPV (OR 2.54, 95% CI 2.25-2.86) and recent-IPV (OR 3.05, 95% CI 2.60-3.58) than those from the wealthiest quintile of standard of living. Importantly, likelihood of IPV decreased with increases in household standard of living. Similarly, those who had no formal schooling were substantially more likely to report lifetime-IPV (OR 4.51, 95% CI 3.31-6.15), and recent-IPV (OR 5.51, 95% CI 3.47-8.74) than women with 13 years or more of education with a clear dose-response relationship. Additionally, women married to men with no formal schooling were more likely to report lifetime-IPV (OR 1.78, 95% CI 1.47-2.15) and recent-IPV (OR 1.82, 95% CI 1.42-2.32) than those married to men with 13 or more years of education. There was also patterning of IPV reports by social characteristics related to family power dynamics. Specifically, women with more education than their husbands were more likely to report lifetime-IPV (OR 1.17, 95% CI 1.07-1.29) and recent-IPV (OR 1.18, 95% CI 1.05-1.33) than those with spousal education parity. Women who worked outside of the home were more likely to report IPV than those who did not, with manual workers reporting the highest risk of both lifetime-IPV (OR 1.63, 95% CI 1.52-1.75) and recent-IPV (OR 1.55, 95% CI 1.42-1.69).
[FIGURE 1 OMITTED]
Caste and IPV
While women from scheduled tribes and other backward classes did not report elevated rates of IPV, women from scheduled castes were more likely to report lifetime-IPV (OR 1.33, 95% CI 1.24-1.42) and recent-IPV (OR 1.37, 95% CI 1.26-1.48) than those who were not from marginalized castes or classes.
Demographic correlates and IPV
Women married before the age of 15 were more likely to report lifetime-IPV (OR 1.79, 95% CI 1.63-1.96) and recent-IPV (OR 1.41, 95% CI 1.25-1.59) than those who married at age 21 or older. Likelihood of reporting lifetime-IPV (OR 0.79, 95% CI 0.70-0.90) and recent-IPV (OR 0.69, 95% CI 0.59-0.80) was lowest among women who lived in rural areas compared to those living in large cities. Compared to Hindu women, Sikhs (OR 1.35, 95% CI 1.02-1.79) and Muslims (OR 1.19, 95% CI 1.08-1.31) were more likely to report recent-IPV. Compared to women in the oldest group (45-49) all other groups were more likely to report recent-IPV with women aged between 25-29 years reporting the highest prevalence (OR 2.12, 95% CI 1.87-2.39). While women in the middle age groups of the sample (25-29 years, 30-34 years, 35-39 years, 40-44 years) were more likely to report lifetime-IPV than those of the oldest group (45-49 years) with the highest prevalence among women aged 30-34 years (OR 1.30, 95% CI 188.8.131.52), those in the youngest group (15-19 years) were less likely to report lifetime-IPV (OR 0.54, 95% CI 0.49-0.61).
Neighborhood- and State-variation in IPV
In unadjusted models, a substantial variation in IPV (lifetime and recent) was observed at the levels of states ([[sigma].sup.2.sub.v0] = 0.426, [[sigma].sup.2.sub.v0] = 0.427, respectively) and neighborhoods ([[sigma].sup.2.sub.u0] = 0.455, [[sigma].sup.2.sub.u0] = 0.498, respectively) (Figure 2). The corresponding variation in lifetime-IPV and recent-IPV after adjusting the distribution of individual covariates in these places was 0.326, and 0.291 for the state level and 0.326, and 0.359 for the neighborhood level. Thus, individual covariates only accounted for 23% and 32% of the unadjusted state variation in lifetime-IPV and recent-IPV, and 28% of the unadjusted neighborhood variation in both lifetime-IPV and recent-IPV. Applying the "latent variable" approach to calculating the relative amount of variation attributable to neighborhoods and states (Goldstein et al. 2002), neighborhoods and states accounted for 10.2% and 10.9% of the total variation in lifetime-IPV and 10.1% and 11.8% of the variation in recent-IPV in unadjusted models. The corresponding proportions for states and neighborhoods in adjusted models were 8.3% and 8.3% in lifetime-IPV and 7.4% and 9.1% in recent-IPV.
[FIGURE 2 OMITTED]
State geography of IPV
In adjusted models, women living in the southern state of Tamil Nadu had the highest likelihood of recent-IPV followed by Bihar, Uttar Pradesh, Arunachal Pradesh, New Delhi, Orissa, Andhra Pradesh, Madhya Pradesh all of which had estimates above the national mean (Figure 3). Meghalaya in the northeast was found to have the lowest adjusted rate of women reporting recent-IPV followed by Himachal Pradesh, Manipur, Rajasthan, Kerala and Gujarat, and in the northwest, which all had rates of reported recent-IPV significantly lower than that of the national average.
[FIGURE 3 OMITTED]
State Gender Equality and IPV
Level of state gender equality was inversely associated with the individual likelihood of reporting recent-IPV, even after accounting for individual covariates, neighborhood wealth, and state economic and human development (Table 3). For one standard deviation increase in state gender equality, the odds ratio of reporting recent-IPV was 0.75 (95% CI 0.58-0.97). There was no difference in the relationship betwe enstate gender equality and recent-IPV by household living standard. Other neighborhood--and state-level variables (i.e., neighborhood wealth, state per capita income, and state human development) were not associated in any substantial manner with the likelihood of reporting either recent- or lifetime-IPV. Although significant variation remained, adding state- and neighborhood-level covariates to the models reduced state variation in lifetime-IPV by 10.4% and reduced recent-IPV by 20.6%; neighborhood variation was unaffected.
This study has four salient findings. First, individual socioeconomic status is strongly associated with the likelihood of a women reporting IPV, with women in high SES categories less likely to report IPV as compared to those in low SES groupings. This pattern was true regardless of how SES was measured, such as household wealth, women's education or husband's education. Second, demographic characteristics related to family power dynamics, such as spousal education differential and woman's occupation were associated with the reporting of IPV. Third, the geographic variation in the likelihood of reporting IPV at the neighborhood- and state-level are not a reflection of the spatial distribution of the individual demographic and socioeconomic determinants of IPV. Put differently, substantial neighborhood and state variation remain even after accounting for individual covariates, suggesting an independent multilevel effect of residential and macro contexts on individual IPV. Finally, higher levels of state gender equality were associated with lower likelihood of reporting IPV. We now discuss each of these findings.
The finding that increased socioeconomic status leading to lower likelihood of IPV is largely consistent with prior evidence published on the socioeconomic patterning of IPV (Koenig et al. 2006; Martin et al. 2002; Martin, Tsui et al. 1999). These studies all came from the Male Reproductive Health Survey, a component of the Program Evaluation Review for Organizational Resource Management. By collecting information about socioeconomic status and IPV perpetration from over 8,000 men in Uttar Pradesh in 1995, this study constituted the largest population-based survey regarding domestic violence in India prior to the INFHS. Although they measured perpetration rather than victimization, the effect sizes and social gradients are similar to those observed in our study.
There are several reasons why IPV may be related to socioeconomic status. Families under stress due to limited financial resources may be more likely to resort to violence in order to settle disputes about how to allocate these resources or to ascribe blame for the reason behind a lack of resources (Martin, Tsui et al. 1999). The educational gradient observed in this study could also be a reflection of the lack of psychosocial resources available to a couple, such as a feeling of self-worth that would tend to keep a man from abusing his wife (Krishnan 2005a). In addition, a formal education may impart some knowledge to spouses allowing them to settle disputes without resorting to physical violence, or could influence their social context in such a way as to make IPV an unacceptable form of conflict resolution.
Power dynamics between spouses may account for some of the results observed in this analysis. A man who has less education than his wife may feel threatened by the social status that the education confers and respond by physically abusing her (Krishnan 2005b). In the same way, a man may be threatened by independence that a woman shows or by the monetary contribution that she makes to the household while working outside of the home. This may help explain why women who work outside of the home report higher rates of IPV. Alternatively, it may be that many women in India work outside the home only if forced to by household economic strain and that the stress of enduring this difficult economic situation may account for the relationship between employment and IPV.
Higher rates of IPV were also found among those women whose characteristics were indicative of stressful social circumstances. That those in scheduled castes were more likely to report abuse than those in the general classes may be indicative of the absence of opportunities available among this group (Krishnan 2005a). Men in the scheduled castes may react to their inability to improve their lot within the rigid caste system by venting their frustration on their wives. Despite experiencing economic and social marginalization comparable to that of the scheduled castes, however, women in the scheduled tribes did not report higher rates of IPV than those in the general classes. This finding suggests that aspects of tribal life are more egalitarian and communitarian (Bhengra et al. 1999) and may thereby protect women from experiencing IPV.
One possible alternative explanation for the demographic patterning of IPV report is that of stigma. Stigma has been identified as one reason that South Asian women do not disclose IPV victimization to their social network or seek help (Naved et al. 2006). It may be, therefore, that abused women in advantaged households are more sensitive to public opinion and are more likely to provide socially desirable responses to survey researchers to hide their abuse victimization than women living in disadvantaged homes. There is a fundamental difference, however, in disclosing abuse to a researcher who is outside of the social context rather than a close personal associate to the extent where researchers have found that many women actually appreciate the opportunity to discuss their experiences of violence with someone they view to be impartial (Krishnan 2005a). Additionally, although violence can be stigmatizing, recent findings that Indian men are socially expected to use force to impose their will on family members (Go et al. 2003; Krishnan 2005a), that many Indians profess a high level of tolerance towards IPV (Martin et al. 2002), and that over 40% of women experience abuse in their lifetimes (Kumar et al. 2005) indicate that IPV is normalized within a range of acceptability in India and may not have the same stigma associated with it that it has in countries with more egalitarian gender norms (Nayak et al. 2003). Finally, even if women from advantaged households are more socially sensitive and exhibit higher social desirability, evidence from Western countries indicates that there is little or no association between social desirability and report of IPV (Bell & Naugle 2007; Dutton & Hemphill 1992) making it unlikely that such a factor could account for the very strong socioeconomic gradients found in this study.
Our second finding of independent geographic differences in IPV by neighborhoods and states, over and above those that can be ascribed to individual factors, emphasizes the need to consider determinants measured at the macro level. To our knowledge, only one other study of India has attempted to quantify the potential importance of context regarding IPV. That study assessed the responses of 4520 men in Uttar Pradesh and did not find any geographic differences among the 92 primary sampling units included after adjusting for individual characteristics (Koenig et al. 2006). Several factors could account for the discrepancy of this null finding compared with our finding significant variation at both the state and neighborhood levels. First, our survey asked women about their experience with violence victimization rather than asking men about their perpetration. Second, it may be that there is less variation in violence across the communities of Uttar Pradesh than there is between the communities in other states. Finally, it could be that the fixed effects that Koenig and his associates used in their analyses, specifically a measure of intergenerational exposure to violence, accounted for much of the variation that would otherwise have been seen as neighborhood variation. That we should find variation in IPV reports across states and neighborhoods should not be surprising given the growing evidence of area effects on health (Diez Roux 2001; Kawachi & Subramanian 2007; Subramanian et al. 2002). This is especially true given the evidence regarding geographic variations in female autonomy (Dyson & Moore 1983) and social inequality (Bhengra et al. 1999) found in India.
Finally, the differential rates of IPV experienced by women between states can be partly explained by the difference in social power between men and women as shown by our third major finding that gender equality is related to lower levels of IPV. While, prior studies have found that community measures of increased women's autonomy in Bangladesh (Koenig et al. 2003) and of decreased wife beating norms in Uttar Pradesh (Koenig et al. 2006) predict lower rates of IPV, ours is the first study to find an association between gender equality and IPV. Dyson and Moore described regional patterns of female autonomy and gender inequality 25 years ago in their seminal work on India's demographic behavior (Dyson & Moore 1983). Although their description of regional patterns of social norms is still relevant today, their description of a North-South divide in demographic behavior appears to be more of an East-West differential in terms of IPV, with women in western India reporting less IPV than those in eastern India. Despite our findings, substantial variation remains to be explained at the neighborhood level which clearly needs further research attention.
We should note that the global measure of domestic violence used in the INFHS has been found to be less likely to elicit a report of violence victimization than measures which ask multiple behaviorally specific questions about what types of abuse the respondent has experienced (Ellsberg et al. 2001). This partially accounts for why the prevalence of abuse reported in the current study is smaller than that reported in previous research of Indian women (Kumar et al. 2005). If understanding of the question is patterned by demographic characteristics, for example if higher SES women are less likely to interpret "minor" abuse such as slapping as IPV, this could threaten the validity of our findings. In recent years, more progressive reproductive health policies in India and significant changes in beliefs regarding gender differences have begun to change women's attitudes towards violence and women from advantaged households seem to be those who are changing their attitudes most quickly indicating that they are decreasing their tolerance of IPV (Datta & Misra 2000; Nayak et al. 2003). We would thus expect that these advantaged women are more likely to report any act of aggression from their husbands as IPV, indicating that the social gradients uncovered in our investigation are, if anything, an underestimate.
In summary, IPV remains a problem of critical importance in India resulting in poor reproductive outcomes (Jejeebhoy 1998; Martin, Kilgallen et al. 1999), asthma (Subramanian, Ackerson et al. 2007), injury (Panchanadeswaran & Koverola 2005), psychological dysfunction (Kumar et al. 2005), suicide (Vizcarra et al. 2004), health behaviors (Ackerson et al. 2007), and murder (Shaha & Mohanthy 2006). Importantly, IPV represents a fundamental adverse endpoint by itself thereby necessitating urgent research and policy actions. The results from this study affirm the important independent effects of a woman's social environment on IPV. They also indicate the role that regional culture may play in dictating the social norms regarding violence against women. This suggests that initiatives directed at promoting gender equality and improving the socioeconomic environment for all people, especially women, may work to replace social norms that encourage violence against women with those that discourage violence.
We acknowledge the support of Macro International (www.measuredhs.com) for providing us access to the 1998-99 Indian National Family Health Survey data. We also acknowledge Amy Cohen for invaluable assistance with programming and data management. SVS is supported by the National Institutes of Health Career Development Award (NHLBI 1 K25 HL081275). No direct financial support was available for this study.
Ackerson, L. K., Kawachi, I., Barbeau, E. M., & Subramanian, S. V. (2007) 'Exposure to domestic violence associated with adult smoking in India: a population based study', Tobacco Control, 16(6), 378-383.
Baddeley, M., McNay, K., & Cassen, R. (2006) 'Divergence in India: Income differentials at the state level, 1970-97', Journal of Development Studies, 42(6), 1000-1022.
Bell, K. M., & Naugle, A. E. (2007) 'Effects of social desirability on students' self-reporting of partner abuse perpetration and victimization', Violence and Victims, 22(2), 243-256.
Benson, M. L., Fox, G. L., DeMaris, A., & Van Wyk, J. (2003) 'Neighborhood disadvantage, individual economic distress and violence against women in intimate relationships', Journal of Quantitative Criminology, 19(3), 207-235.
Berkman, L. F., & Kawachi, I. (2000) Social Epidemiology, Oxford, New York, Oxford University Press.
Besley, T., Rao, V., & Pande, R. (2005) 'Participatory democracy in action: Survey evidence from South India', Journal of the European Economic Association, 3(2-3), 648-657.
Bhengra, R., Bijoy, C. R., & Luithui, S. (1999) The Adivasis of India, London, Minority Rights Group International.
Blakely, T., & Subramanian, S. V. (2006) 'Multilevel studies'. In J. M. Oakes & J. S. Kaufman (Eds.) Methods in Social Epidemiology, San Francisco, Jossey-Bass.
Braveman, P. A., Cubbin, C., Egerter, S., Chideya, S., Marchi, K. S., Metzler, M., & Posner, S. (2005) 'Socioeconomic status in health research: one size does not fit all', JAMA (The Journal of the American Medical Association), 294(22), 2879-2888.
Braveman, P., Krieger, N., & Lynch, J. (2000) 'Health inequalities and social inequalities in health', Bulletin of the World Health Organisation, 78(2), 232-234; discussion 234-235.
Campbell, J. C. (2002) 'Health consequences of intimate partner violence', Lancet, 359(9314), 1331-1336.
Chen, Y. Y., Subramanian, S. V., Acevedo-Garcia, D., & Kawachi, I. (2005) 'Women's status and depressive symptoms: a multilevel analysis', Social Science & Medicine, 60(1), 49-60.
Chitnis, S. (1997) 'Definition of the terms scheduled castes and scheduled tribes: a crisis of ambivalence'. In V. A. Pai Panandiker (Ed.) The Politics of Backwardness: Reservation Policy in India/edited by V.A. Pai Panandiker, New Delhi, Konark.
Cunradi, C. B., Caetano, R., Clark, C., & Schafer, J. (2000) 'Neighborhood poverty as a predictor of intimate partner violence among White, Black, and Hispanic couples in the United States: a multilevel analysis', Annals of Epidemiology, 10(5), 297-308.
Datta, B., & Misra, G. (2000) 'Advocacy for sexual and reproductive health: the challenge in India', Reproductive Health Matters, 8(16), 24-34.
Diez Roux, A. V. (2001). 'Investigating neighborhood and area effects on health', American Journal of Public Health, 91(11), 1783-1789.
Dutton, D. G., & Hemphill, K.J. (1992) 'Patterns of socially desirable responding among perpetrators and victims of wife assault', Violence and Victims, 7(1), 29-39.
Dyson, T., & Moore, M. (1983) 'On kinship structure, female autonomy, and demographic behavior in India', Population and Development Review, 9(1), 35-60.
Ellsberg, M., Heise, L., Pena, R., Agurto, S., & Winkvist, A. (2001) 'Researching domestic violence against women: methodological and ethical considerations', Studies in Family Planning, 32(1), 1-16.
Epel, E. S., Blackburn, E. H., Lin, J., Dhabhar, E S., Adler, N. E., Morrow, J. D., & Cawthon, R. M. (2004) 'Accelerated telomere shortening in response to life stress', Proceedings of the National Academy of Sciences of the United States of America, 101(49), 17312-17315.
Filmer, D., & Pritchett, L. H. (2001) 'Estimating wealth effects without expenditure dataor tears: an application to educational enrollments in states of India', Demography, 38(1), 115-132.
Garcia-Moreno, C., Jansen, H. A., Ellsberg, M., Heise, L., & Watts, C. H. (2006) 'Prevalence of intimate partner violence: findings from the WHO multi-country study on women's health and domestic violence', Lancet, 368(9543), 1260-1269.
Go, V. F., Johnson, S. C., Bentley, M. E., Sivaram, S., Srikrishnan, A. K., Celentano, D. D., & Solomon, S. (2003) 'Crossing the threshold: engendered definitions of socially acceptable domestic violence in Chennai, India', Culture Health & Sexuality, 5(5), 393-408.
Goldstein, H. (2003) Multilevel Statistical Models, London, Arnold.
Goldstein, H., Browne, W., & Rasbash, J. (2002) 'Partitioning variation in generalised linear multi-level models', Understanding Statistics, 1(4), 223-232.
Hathaway, J. E., Mucci, L. A., Silverman, J. G., Brooks, D. R., Mathews, R., & Pavlos, C. A. (2000) 'Health status and health care use of Massachusetts women reporting partner abuse', American Journal of Preventive Medicine, 19(4), 302-307.
Heise, L.; & Garcia-Moreno, C. (2002) 'Violence by intimate partners'. In E. G. Krug & L. L. Dahlberg & J. A. Mercy & A. B. Zwi & R. Lozano (Eds.) World Report on Violence and Health, Geneva, World Health Organization.
IIPS, & ORC-Macro. (2000) National Family Health Survey, 1998-1999: India, Mumbai, International Institute for Population Sciences.
India. (2002) National Human Development Report--2001, New Delhi, Planning Commission-Government of India.
Jejeebhoy, S.J. (2002) 'Convergence and divergence in spouses' perspectives on women's autonomy in rural India', Studies in Family Planning, 33(4), 299-308.
Jejeebhoy, S.J. (1998) 'Associations between wife-beating and fetal and infant death: impressions from a survey in rural India', Studies in Family Planning, 29(3), 300-308.
Jun, H. J., Subramanian, S. V., Gortmaker, S., & Kawachi, I. (2004) 'A multilevel analysis of women's status and self-rated health in the United States', Journal of the American Medical Women's Association, 59(3), 172-180.
Kawachi, I., & Subramanian, S. V. (2007) 'Neighbourhood influences on health', Journal of Epidemiology and Community Health, 61(1), 34.
Kawachi, I., & Berkman, L. F. (2003). Neighborhoods and health. Oxford, New York, Oxford University Press.
Kawachi, I., Kennedy, B. P., Gupta, V., & Prothrow-Stith, D. (1999) 'Women's status and the health of women and men: a view from the States', Social Science & Medicine, 48(1), 21-32.
Koenen, K. C., Lincoln, A., & Appleton, A. (2006) 'Women's status and child well-being: a state-level analysis', Social Science & Medicine, 63(12), 2999-3012.
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, 96(1), 132-138.
Koenig, M. A., Ahmed, S., Hossain, M. B., & Khorshed Alam Mozumder, A. B. (2003) 'Women's status and domestic violence in rural Bangladesh: individual- and community-level effects', Demography, 40(2), 269-288.
Krishnan, S. (2005a) 'Do structural inequalities contribute to marital violence? Ethnographic evidence from rural South India', Violence Against Women, 11(6), 759-775.
Krishnan, S. (2005b) 'Gender, caste, and economic inequalities and marital violence in rural South India', Health Care for Women International, 26(1), 87-99.
Kumar, S., Jeyaseelan, L., Suresh, S., & Ahuja, R. C. (2005) 'Domestic violence and its mental health correlates in Indian women', The British Journal of Psychiatry, 187, 62-67.
Marmot, M. (2007) 'Achieving health equity: from root causes to fair outcomes', Lancet, 370(9593), 1153-1163.
Marmot, M. (2002) 'The influence of income on health: views of an epidemiologist', Health Affairs (Millwood), 21(2), 31-46.
Martin, S. L., Moracco, K. E., Garro, J., Tsui, A. O., Kupper, L. L., Chase, J. L., & Campbell, J. C. (2002) 'Domestic violence across generations: findings from northern India', International Journal of Epidemiology, 31(3), 560-572.
Martin, S. L., Kilgallen, B., Tsui, A. O., Maitra, K., Singh, K. K., & Kupper, L. L. (1999) 'Sexual behaviors and reproductive health outcomes: associations with wife abuse in India', JAMA (The Journal of the American Medical Association), 282(20), 1967-1972.
Martin, S. L., Tsui, A. O., Maitra, K., & Marinshaw, R. (1999) 'Domestic violence in northern India', American Journal of Epidemiology, 150(4), 417-426.
Miles-Doan, R., & Kelly, S. (1997) 'Geographic concentration of violence between intimate partners', Public Health Reports, 112(2), 135-141.
Moon, G., Subramanian, S. V., Jones, K., Duncan, C., & Twigg, L. (2005) 'Area-based studies and the evaluation of multilevel influences on health outcomes'. In A. Bowling & S. Ebrahim (Eds.) Handbook of Health Research Methods: Investigation, Measurement and Analysis, Berkshire, England, Open University Press.
Nath, D. C., Land, K. C., & Goswami, G. (1999) 'Effects of the status of women on the first-birth interval in Indian urban society', Journal of Biosocial Science, 31(1), 55-69.
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 & Medicine, 62(12), 2917-2929.
Nayak, M. B., Byrne, C. A., Martin, M. K., & Abraham, A. G. (2003) 'Attitudes toward violence against women: a cross-nation study', Sex Roles, 49(7-8), 333-342.
O'Campo, P. (2003) 'Invited commentary: Advancing theory and methods for multilevel models of residential neighborhoods and health', American Journal of Epidemiology, 157(1), 9-13.
Pallitto, C. C., & O'Campo, P. (2005) 'Community level effects of gender inequality on intimate partner violence and unintended pregnancy in Colombia: testing the feminist perspective', Social Science & Medicine, 60(10), 2205-2216.
Panchanadeswaran, S., & Koverola, C. (2005) 'The voices of battered women in India', Violence Against Women, 11(6), 736-758.
Pearlman, D. N., Zierler, S., Gjelsvik, A., & Verhoek-Oftedahl, W. (2003) 'Neighborhood environment, racial position, and risk of police-reported domestic violence: a contextual analysis', Public Health Reports, 118(1), 44-58.
Pickett, K. E., & Pearl, M. (2001) 'Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review', Journal of Epidemiology and Community Health, 55(2), 111-122.
Pur, K. A. (2007) 'Rivalry or synergy? Formal and informal local governance in rural India', Development and Change, 38(3), 401-421.
Rasbash, J., Steele, F., Browne, W., & Prosser, B. (2005) A User's Guide to MLwiN, Version 2.0., Bristol, UK, Centre for Multilevel Modelling.
Rutstein, S. O., & Johnson, K. (2004) The DHS Wealth Index: DHS Comparative Reports 6, Calverton, MD, ORC Macro.
Sampson, R. J. (2003) 'The neighborhood context of well-being', Perspectives in Biology and Medicine, 46(3 Suppl), S53-64.
Shaha, K. K., & Mohanthy, S. (2006) 'Alleged dowry death: a study of homicidal burns', Medicine Science & the Law, 46(2), 105-110.
Soo, K. T. (2007) 'Endogenous economic policy and the structure of production: Theory and evidence', Scottish Journal of Political Economy, 54(2), 220-253.
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(4), 201-208.
Subramanian, S. V., Ackerson, L. K., Subramanyam, M. A., & Wright, R.J. (2007) 'Domestic violence is associated with adult and childhood asthma prevalence in India', International Journal of Epidemiology, 36(3), 569-79.
Subramanian, S. V., Glymour, M., & Kawachi, I. (2007) 'Identifying causal ecologic effects on health: potentials and challenges'. In S. Galea (Ed.) Macrosocial determinants of population health, New York, Springer Media.
Subramanian, S. V., & Davey Smith, G. (2006) 'Patterns, distribution, and determinants of under- and overnutrition: a population-based study of women in India', The American Journal of Clinical Nutrition, 84(3), 633-640.
Subramanian, S. V., Nandy, S., Irving, M., Gordon, D., Lambert, H., & Davey Smith, G. (2006) 'The mortality divide in India: the differential contributions of gender, caste, and standard of living across the life course', American Journal of Public Health, 96(5), 818-825.
Subramanian, S. V. (2004) 'The relevance of multilevel statistical methods for identifying causal neighborhood effects', Social Science & Medicine, 58(10), 1961-1967.
Subramanian, S. V., Jones, K., & Duncan, C. (2003) 'Multilevel methods for public health research'. In I. Kawachi & L. F. Berkman (Eds.) Neighborhoods and Health, New York,: Oxford University Press.
Subramanian, S. V., Belli, P., & Kawachi, I. (2002) 'The macroeconomic determinants of health', Annual Review of Public Health, 23, 287-302.
Sudha, S., Morrison, S., & Zhu, L. (2007) 'Violence against women, symptom reporting, and treatment for reproductive tract infections in Kerala state, Southern India', Health Care for Women International, 28(3), 268-284.
Vest, J. R., Catlin, T. K., Chen, J. J., & Brownson, R. C. (2002) 'Multistate analysis of factors associated with intimate partner violence', American Journal of Preventive Medicine, 22(3), 156-164.
Vizcarra, B., Hassan, F., Hunter, W. M., Munoz, S. R., Ramiro, L., & De Paula, C. S. (2004) 'Partner violence as a risk factor for mental health among women from communities in the Philippines, Egypt, Chile, and India', Injury Control and Safety Promotion, 11(2), 125-129.
Watts, C., & Zimmerman, C. (2002) 'Violence against women: global scope and magnitude', Lancet, 359(9313), 1232-1237.
Appendix 1. State per capita gross domestic product, human development, and gender equality State Per Capita State Human Development Domestic Product Index (in SD from (in rupees) the mean) Andhra Pradesh 2550 -0.65 Arunachal Pradesh 3571 -1.24 Assam 1675 -1.00 Bihar 1126 -1.48 Goa 5640 1.75 Gujarat 3918 0.01 Haryana 4025 0.15 Himachal Pradesh 2556 0.47 Jammu 1932 -0.34 Karnataka 2866 -0.22 Kerala 2490 1.95 Madhya Pradesh 1922 -1.24 Maharashtra 5032 0.26 Manipur 1948 1.28 Meghalaya 1804 -0.79 Mizoram 2840 1.43 Nagaland 2164 0.68 New Delhi 6478 2.35 Orissa 1666 -1.03 Punjab 4389 0.54 Rajasthan 2226 -1.01 Sikkim 3461 -0.06 Tamil Nadu 3141 0.43 Tripura 2117 -0.50 Uttar Pradesh 1725 -1.41 West Bengal 2977 -0.32 State Gender Equality Index (in SD from the mean) Andhra Pradesh 0.82 Arunachal Pradesh 0.60 Assam -1.21 Bihar -2.17 Goa 0.59 Gujarat 0.04 Haryana 0.04 Himachal Pradesh 1.34 Jammu 0.27 Karnataka 0.39 Kerala 1.04 Madhya Pradesh -0.43 Maharashtra 0.75 Manipur 0.95 Meghalaya 0.88 Mizoram 0.54 Nagaland 0.17 New Delhi -0.18 Orissa -0.64 Punjab 0.00 Rajasthan -0.16 Sikkim -0.56 Tamil Nadu 0.93 Tripura -1.61 Uttar Pradesh -1.71 West Bengal -0.71
Table 1. Descriptive information for the 1998-1999 Indian National Family Health Survey Variable Subjects % (1) Lifetime-IPV % (2) Total 83627 100.0 13460 16.1 Location Large city 9213 11.0 993 10.8 Small city 5319 6.4 689 13.0 Town 11465 13.7 1464 12.8 Village 57630 68.9 10314 17.9 Age 45-49 7205 8.6 1047 14.5 40-44 9742 11.7 1584 16.3 35-39 12640 15.1 2193 17.4 30-34 14865 17.8 2675 18.0 25-29 17272 20.7 2964 17.2 20-24 15316 18.3 2249 14.7 15-19 6587 7.9 748 11.4 Age at Wedding 21 or older 13335 16.0 1029 7.7 18-20 23688 28.3 2992 12.6 15-17 31554 37.7 5721 18.1 less than 15 15050 18.0 3718 24.7 Religion Hindu 65167 77.9 10721 16.5 Muslim 9921 11.9 1769 17.8 Christian 4506 5.4 497 11.0 Sikh 1988 2.4 217 10.9 Other/missing religion 2045 2.5 256 12.5 Caste General 35159 42.0 3992 11.4 Scheduled caste 14174 17.0 3213 22.7 Scheduled tribe 9967 11.9 1669 16.8 Other backward class 24327 29.1 4586 18.9 Employment Not working 54159 64.8 6919 12.8 Non-manual 3978 4.8 423 10.6 Agricultural 18601 22.2 4405 23.7 Manual 6889 8.2 1713 24.9 Living Standard 5th (highest) quintile 17124 20.5 987 5.8 4th quintile 17264 20.6 2017 11.7 3rd quintile 16896 20.2 2843 16.8 2nd quintile 16676 20.0 3517 21.1 1st (lowest) quintile 15667 18.7 4096 26.1 Wife's Education 13 or more years 4374 5.2 74 1.7 11-12 years 3678 4.4 203 5.5 9-10 years 9838 11.8 711 7.2 6-8 years 10520 12.6 1273 12.1 1-5 years 13666 16.3 2330 17.1 No formal schooling 41551 49.7 8869 21.3 Husband's Education 13 or more years 9313 11.1 425 4.6 11-12 years 7002 8.4 627 9.0 9-10 years 17263 20.6 2030 11.8 6-8 years 13745 16.4 2196 16.0 1-5 years 14936 17.9 3095 20.7 No formal schooling 21368 25.6 5087 23.8 Spouse Education Differential Same level 33897 40.5 6023 17.8 Wife is more educated 7722 9.2 1084 14.0 Husband is more educated 42008 50.2 6353 15.1 Variable Retell-IPV % (3) Total 7749 9.3 Location Large city 556 6.0 Small city 361 6.8 Town 776 6.8 Village 6056 10.5 Age 45-49 394 5.5 40-44 706 7.3 35-39 1139 9.0 30-34 1525 10.3 25-29 1844 10.7 20-24 1539 10.1 15-19 602 9.1 Age at Wedding 21 or older 562 4.2 18-20 1721 7.3 15-17 3369 10.7 less than 15 2097 13.9 Religion Hindu 6153 9.4 Muslim 1018 10.3 Christian 283 6.3 Sikh 127 6.4 Other/missing religion 168 8.2 Caste General 2199 6.3 Scheduled caste 1972 13.9 Scheduled tribe 1009 10.1 Other backward class 2569 10.6 Employment Not working 4106 7.6 Non-manual 198 5.0 Agricultural 2476 13.3 Manual 969 14.1 Living Standard 5th (highest) quintile 455 2.7 4th quintile 1065 62.0 3rd quintile 1530 9.1 2nd quintile 2101 12.6 1st (lowest) quintile 2598 16.6 Wife's Education 13 or more years 29 0.7 11-12 years 96 2.6 9-10 years 402 4.1 6-8 years 718 6.8 1-5 years 1239 9.1 No formal schooling 5265 12.7 Husband's Education 13 or more years 226 2.4 11-12 years 334 4.8 9-10 years 1154 6.7 6-8 years 1258 9.2 1-5 years 1696 11.4 No formal schooling 3081 14.4 Spouse Education Differential Same level 3538 10.4 Wife is more educated 609 7.9 Husband is more educated 3602 8.6 (1)-Indicates the percentage of women with that descriptive characteristic (2)-Indicates the percentage of women with the characteristic who were abused since age 15 (3)-Indicates the percentage of women with the characteristic who were abused in the past year Table 2. Adjusted logistic models of lifetime- and recent-IPV by socioeconomic and demographic characteristics in the 1998-1999 Indian National Family Health Survey Lifetime Abuse Variable Odds Ratio 95% CI Location Large city (ref) 1.00 Small city 1.09 0.92-1.29 Town 1.00 0.87-1.15 Village 0.79 0.70-0.90 Age 45-49 (ref) 1.00 40-44 1.14 1.04-1.25 35-39 1.25 1.14-1.36 30-34 1.30 1.20-1.42 25-29 1.24 1.14-1.35 20-24 0.97 0.89-1.06 15-19 0.54 0.49-0.61 Age at Wedding 21 or older (ref) 1.00 18-20 1.18 1.09-1.29 15-17 1.47 1.35-1.60 less than 15 1.79 1.63-1.96 Religion Hindu (ref) 1.00 Muslim 1.22 1.13-1.32 Christian 0.94 0.81-1.08 Sikh 1.18 0.94-1.47 Other/missing religion 0.98 0.82-1.17 Caste General (ref) 1.00 Scheduled caste 1.33 1.24-1.42 Scheduled tribe 1.07 0.98-1.17 Other backward class 1.02 0.96-1.08 Employment Not working (ref) 1.00 Non-manual 1.47 1.30-1.66 Agricultural 1.47 1.39-1.56 Manual 1.63 1.52-1.75 Living Standard 5th (highest) quintile (ref) 1.00 4th quintile 1.47 1.33-1.61 3rd quintile 1.95 1.76-2.16 2nd quintile 2.25 2.01-2.52 1st (lowest) quintile 2.54 2.25-2.86 Wife's Education 13 or more years (ref) 1.00 11-12 years 2.59 1.93-3.48 9-10 years 2.94 2.23-3.88 6-8 years 3.88 2.91-5.17 1-5 years 4.55 3.39-6.11 No formal schooling 4.51 3.31-6.15 Husband's Education 13 or more years (ref) 1.00 11-12 years 1.27 1.10-1.46 9-10 years 1.46 1.28-1.66 6-8 years 1.61 1.40-1.85 1-5 years 1.80 1.55-2.09 No formal schooling 1.78 1.47-2.15 Spouse Education Differential Same level (ref) 1.00 Wife is more educated 1.17 1.07-1.29 Husband is more educated 1.00 0.92-1.10 Recent Abuse Variable Odds Ratio 95% CI Location Large city (ref) 1.00 Small city 0.96 0.79-1.18 Town 0.86 0.73-1.03 Village 0.69 0.59-0.80 Age 45-49 (ref) 1.00 40-44 1.35 1.18-1.54 35-39 1.73 1.52-1.96 30-34 2.01 1.77-2.27 25-29 2.12 1.87-2.39 20-24 1.90 1.68-2.15 15-19 1.32 1.14-1.52 Age at Wedding 21 or older (ref) 1.00 18-20 1.11 0.99-1.24 15-17 1.29 1.16-1.44 less than 15 1.41 1.25-1.59 Religion Hindu (ref) 1.00 Muslim 1.19 1.08-1.31 Christian 0.97 0.81-1.16 Sikh 1.35 1.02-1.79 Other/missing religion 1.09 0.88-1.35 Caste General (ref) 1.00 Scheduled caste 1.37 1.26-1.48 Scheduled tribe 1.04 0.93-1.17 Other backward class 1.01 0.94-1.09 Employment Not working (ref) 1.00 Non-manual 1.25 1.06-1.48 Agricultural 1.36 1.27-1.46 Manual 1.55 1.42-1.69 Living Standard 5th (highest) quintile (ref) 1.00 4th quintile 1.63 1.43-1.86 3rd quintile 2.15 1.86-2.48 2nd quintile 2.67 2.30-3.11 1st (lowest) quintile 3.05 2.60-3.58 Wife's Education 13 or more years (ref) 1.00 11-12 years 2.84 1.80-4.48 9-10 years 3.73 2.43-5.72 6-8 years 4.72 3.05-7.30 1-5 years 5.24 3.36-8.19 No formal schooling 5.51 3.47-8.74 Husband's Education 13 or more years (ref) 1.00 11-12 years 1.17 0.97-1.42 9-10 years 1.42 1.20-1.69 6-8 years 1.57 1.31-1.88 1-5 years 1.70 1.40-2.07 No formal schooling 1.82 1.42-2.32 Spouse Education Differential Same level (ref) 1.00 Wife is more educated 1.18 1.05-1.33 Husband is more educated 1.04 0.93-1.16 Table 3. Adjusted logistic models of lifetime- and recent-IPV by state- and neighborhood-level variables in the 1998-1999 Indian National Family Health Survey Lifetime-IPV Model 3 Model 4 Variable Odds 95% CI Odds 95% CI Ratio Ratio Neighborhood wealth 4th (richest) 1.00 1.00 quartile (ref) 3rd quartile 1.03 0.92-1.15 1.03 0.92-1.15 2nd quartile 0.97 0.84-1.12 0.97 0.84-1.12 1st (lowest) quartile 1.04 0.89-1.22 1.05 0.90-1.22 Per capita state GDP 1.08 0.91-1.28 (1000 rupees) State gender equality (1 SD) State human development (1 SD) Recent-IPV Model 3 Model 4 Variable Odds 95% CI Odds 95% CI Ratio Ratio Neighborhood wealth 4th (richest) 1.00 1.00 quartile (ref) 3rd quartile 1.10 0.96-1.27 1.10 0.96-1.27 2nd quartile 0.96 0.81-1.14 0.96 0.81-1.14 1st (lowest) quartile 1.05 0.87-1.26 1.05 0.88-1.27 Per capita state 1.06 0.90-1.25 GDP (1000 rupees) State gender equality (1 SD) State human development (1 SD) Lifetime-IPV Model 5 Variable Odds 95% CI Ratio Neighborhood wealth 4th (richest) 1.00 quartile (ref) 3rd quartile 1.03 0.92-1.15 2nd quartile 0.97 0.84-1.12 1st (lowest) quartile 1.04 0.90-1.22 Per capita state GDP 1.15 0.94-1.42 (1000 rupees) State gender 0.83 0.63-1.10 equality (1 SD) State human 0.95 0.71-1.26 development (1 SD) Recent-IPV Model 5 Variable Odds 95% CI Ratio Neighborhood wealth 4th (richest) 1.00 quartile (ref) 3rd quartile 1.10 0.96-1.27 2nd quartile 0.96 0.81-1.14 1st (lowest) quartile 1.05 0.87-1.26 Per capita state 1.15 0.95-1.39 GDP (1000 rupees) State gender 0.75 0.58-0.97 equality (1 SD) State human 0.96 0.74-1.24 development (1 SD) Note: All models are adjusted for location of residence, age, age at wedding, religion, caste, occupation, living standard, education, husband's education, and spousal education differential.
|Gale Copyright:||Copyright 2008 Gale, Cengage Learning. All rights reserved.|