Men’s Experiences of Family, Domestic and Honour-Related Violence in Gujarat and Uttar Pradesh, India

Background: Violence is a major problem in India with family, domestic and honour-related violence having significant impacts on the wellbeing of Indian families and communities. There has been little attention paid to men’s experiences, particularly in the Indian State’s of Gujarat and Uttar Pradesh. The objective of this study was to provide baseline data on Indian men’s experiences of violence and the key predictors such as age, income, education and religion. Methods: A cross-sectional survey of men from Uttar Pradesh and Gujarat. Data were obtained from a non-probability purposive sample of 1000 adult men over a two-month period in late 2009. The respondents were selected from public spaces in 10 cities or towns within these two provinces. Results: The results found a lifetime prevalence of experiencing family violence of 99.2% (15.6% often and 64.1% sometimes), domestic violence 92.2% (21.1% often and 44.9% sometimes), and honour-related violence 99.7% (45.1% often and 41.7% sometimes). Moreover, when we analysed the men’s responses across socio-demographic factors, age, income, religion, education and region were each shown to be important predictors of regular exposure to family, domestic and honour-related violence. Conclusions: The results of this survey provide much needed insight into the factors that predict experiences of violence. Such data can inform targeting by community development programs and Government policy in order to provide additional support to individuals and communities most likely to be exposed to these types of violence.


Introduction
India is a developing country with a vast and diverse population.Despite significant and sustained economic development over the last few years, much like other developing countries, different forms of community violence including domestic, family and honour-related, persist at very high levels (Ahmed et al. 2007;Martin et al. 2002;Pandey et al, 2009).While there is an increasing public health and social welfare sector working against violence in India, violence remains a constant and inevitable part of many people's everyday lives (Nilan et al. 2008).Violence stems from a broad range of social and economic issues, including property, caste structures, patriarchal structures, religious dynamics and ethnic unrest (Krishnan, 2005;Panda & Agarwal, 2005;Rao, 1997).Despite the prominence of violence, there has been relatively little research on the patterning of exposure to violence amongst Indian men (Nilan et al. 2008).This has largely been due to difficulties recording accurate prevalence figures (Watts & Zimmerman, 2002) including a lack of willingness of men to participate in violence research.In this study we attempted to circumvent this stigma by recording men's broader experiences of, and exposure to violence, moving away from an explicit focus on men as perpetrators.

Violence in India
Globally there exists a strong research and development agenda focused on addressing the issues of gender-based and community violence (Koenig et al. 2006).The health consequences of family, domestic and other forms of community violence have been well documented (Ackerson & Subramanian, 2008), and the tendency of violence to filter down generationally is a significant and ongoing social problem (Martin et al. 2002).Yet, at a grassroots level there remains an entrenched acceptability of different forms of violence in India.Studies have previously illustrated the prominence and impacts of interpersonal violence in India (Ahmed et al. 2007;Wilson-Williams et al. 2008), including the impacts on families, individual wellbeing, community life and economic progress (Pandey et al. 2009;Garcia-Moreno et al. 2006;Shaha et al. 2006).Estimates of the levels of violence differ significantly according to the region examined and the character of questionaries, with domestic and family violence affecting 20-70% of families (Martin et al. 2002;Pillai et al. 2009).Moreover, honour-related violence, including dowry-related violence and killings, persists in many parts of India (Kozel & Parker, 2003).The impacts of domestic, family and honour-related violence are vast, with serious and long term social impacts including depression, substance abuse and suicide (Krishnan, 2005).While there has been valuable research completed on Indian women's experiences of violence (Pandey et al. 2009;Garcia-Moreno et al. 2006;Shaha & Mohanthy, 2006) little work has been conducted with Indian men (Watts & Zimmerman, 2002).

Violence and Social Inequality in India
Studies of interpersonal violence in India should be situated within an understanding that violence is often intertwined with social inequalities and forms of social deprivation (Krisnan, 2005).Indeed, previous work has illustrated that exposure to different forms of violence is intimately intertwined in socioeconomic conditions (Jejeebhoy, 1998).In saying this, violence can also transcend class, caste and educational divides, illustrating the importance of research examining how and in what ways socio-demographics mediate exposure (Nilan et al. 2008).It is also important to consider that the various histories of particular regions shape the acceptability and prevalence of different forms of violence (Mehta & Metha, 2010).As such, the reports of exposure to violence reported here should be placed in historical, social and cultural context; moreover, as base-line data that can facilitate a wider mapping of men's experiences and exposure to violence in India (Nilan et al. 2008).

Settings
The study was undertaken in Uttar Pradesh (UP) and Gujarat.UP is the most populated state in India with over 190 million people.Poverty and low literacy levels have seen UP consistently described as one of the 'most backward' states of India, with domestic violence reported as a major problem, particularly in rural areas (Kumar, 2005;Martin et al. 2002).Poverty interplays with low caste and religious status, with many of the State's 22% Dalit and 18% Muslim population highly marginalised, thus experiencing severe material and social deprivation.
Gujarat is located in the North Western India, has a population of over 50 million people, and its economic achievements have seen it hailed as the 'model State' in India.Yet, Gujarat's economic development is at odds with its simmering religious and ethnic tensions, peaking with widespread riots in 2002 resulting in the deaths of 57 Hindus and close to 2000 Gujarati Muslims.The Gujarati riots involved significant and sustained attacks on Muslim communities, with mobs of men targeting women and children as well as Muslim men.This conflict reflects ongoing ethnic, economic and religious tensions in Gujarat tensions which in turn shape, and feed into, ongoing sites of interpersonal violence.These two States provide a potentially interesting comparison in terms of how violence may be situated within levels of economic development, gender relations, levels of education, and recent political tensions.

Definitions of Violence
We provided the following broad definition of violence to the participants: any act -physical, verbal or emotional -that is intended to, or results in, harm to another person or group.Specifically, family violence was defined as parent-to-child violence, domestic violence was defined as partner-to-partner violence, and honour-related violence was defined as violence perpetrated upon a woman when an honour code was perceived to have been broken (e.g.dowry-related, pre premarital relationships, refusing an arranged marriage and so on).The men surveyed were told verbally of these definitions of each form of violence to ensure consistency and we asked to them to respond in relation to their own personal experiences of and exposure to violence.

Study Sample
A non-probability purposive sample of 1000 men was drawn from Uttar Pradesh and Gujarat.This included a representative sample drawn from 5 cities/towns in each state.The respondents were selected randomly with a minimum of 30 respondents from each city/town.The interviewers approached men in public spaces rather than in their homes, including 191 different locations within the 10 towns to ensure representativeness.The response rate for the survey was over 99% with three men in Gujarat and seven in Uttar Pradesh refusing participation.The characteristics of the final sample are closely aligned to the demographics of the broader male population in Gujarat and Uttar Pradesh.The survey lasted around 20 minutes and the survey was pre-tested with 30 participants before the actual survey was conducted.The survey was done in Gujarati and Hindi, depending on the language of the participant and the English-Gujarati and English-Hindi translations were checked by an independent specialist in each language.The survey lasted around 20 minutes and the survey was pre-tested with 30 participants before the actual survey was conducted.

Outcome Measures
Participants were asked how often they had experienced different forms of violence (i.e.never, rarely, sometimes, often).We deliberately did not ask whether they were the perpetrators or not as this has been shown to limit reporting of exposure to violence. 19This results in data that does not illustrate severity or establish whether the male was a victim or perpetrator.This was a deliberate trade-off on the part of the authors.While some men may reveal their role in violence, the majority of men (and Indian men are not exception) will not talking honestly about if (and how often) they hit their wives, hit their children or are involved in honour-related violence.As such, below we provide predictors of degrees/extent of exposure which can be augmented by further research recording such things as severity, roles and so on.
A participant was considered to have experienced family violence, domestic or honour-related violence if they indicated that they had experienced this form of violence either sometimes or often.

Statistical Analyses
For the different types of violence, the associations with independent variables were examined using simple and multiple logistic regression models.The multiple logistic regression models were determined by a backwards stepwise elimination procedure.The fit of the final multiple logistic regression models was tested by Hosmer-Lemeshow goodness of fit tests.Here the outcome for each model was categorized as binary variable, defined as never/rarely and sometimes/often.

Lifetime Prevalence of Experiences of Violence
As a baseline measure we were interested in recording whether these men had personally experienced each form of violence.The lifetime prevalence of experiencing family violence for the men in this sample was 99.2% (15.6% often and 64.1% sometimes).The lifetime prevalence of experiencing domestic violence for the men in this sample was 92.2% (21.1% often and 44.9% sometimes).The lifetime prevalence of experiencing honour-related violence for the men in this sample was 99.7% (45.1% often and 41.7% sometimes).

Predictors of Experiences of Family Violence (Parent-to-child Violence)
Table 1 presents the distribution of participants by various characteristics (independent variables), and the levels of significance (p-values) of simple logistic regression models to examine the association between these independent variables and the three forms of violence.Of the 9 independent variables, 6 were significantly associated with family violence.2).Specifically, the odds of a male 40 years or older experiencing family violence is 2.34 (95% CI: 1.44-3.80)times greater than a male aged less than 30 years.In comparison to a male with only a primary school education, the odds of experiencing family violence is 0.28 (95% CI: 0.10-0.80)and 0.20 (95% CI: 0.07-0.58)times less for males with secondary and tertiary educations respectively.
Economic constraint has often been shown to be a key factor in interpersonal violence (Jejeebhoy, 1998;Krishnan, 2005) and certainly the results of this survey reinforce this.Males with a monthly income of 5000-9999 INR are 0.33 (95% CI: 0.21-0.50)times less likely to experience family violence compared to males with less than 5000 INR per month and males with a monthly income of 10000 INR or more are 0.14 (95% CI: 0.08-0.25)times less likely to experience violence compared to males with less than 5000 INR per month.Essentially, the lower the income bracket, the more likely to be exposed to family violence.
The religious factors mediating experiences of violence emerged strongly in the results, particularly in relation to family violence.Compared to Muslim males, Hindu males are 3.33 (95% CI: 2.06-5.39)times more likely to experience family violence.This sits in stark contrast to a lack of any association in the context of domestic violence and a weaker influence for honour-related violence (see below).The Hosmer-Lemeshow test (p=0.39)indicated that the multiple logistic regression model for family violence was a good fit.

Predictors of Experiencing Domestic Violence (Partner-to-partner Violence)
The simple logistic regressions used to examine the association between the various characteristics and domestic violence shown in Table 1 revealed that, of the 9 independent variables, 8 were significantly associated with domestic violence.However, after controlling for other factors within the multiple logistic regression model, only 5 variables remained statistically significant predictors of domestic violence (see Table 2).Specifically, the odds of males aged 30-39 year experiencing domestic violence are 2.69 (95% CI: 1.77-4.08)times that of males aged under 30 years, while males aged 40 years or over are 2.79 (95% CI:1.72-4.50)times more likely to experience domestic violence than males aged under 30 years.
Once again, domestic violence was linked to income with males with a monthly income of 5000-9999 INR 0.33 (95% CI: 0.23-0.47)times less likely to experience domestic violence compared to males with less than 5000 INR per month and males with a monthly income of 10000 INR or more are 0.17 (95% CI: 0.11-0.27)times less likely to experience domestic violence compared to males with less than 5000 INR per month.Pressure on families with children also emerged as important, with the survey showing that, in comparison to males with no children, males with one or two children and males with 3 or more children are 3.46 (95% CI: 2.06-5.81)and 5.27 (95% CI: 2.88-9.65)times more likely to experience domestic violence, respectively.
The survey found significant differences between the two states in terms of exposure to violence, but only in the context of domestic violence.The odds of males from Uttar Pradesh experiencing domestic violence is 2.21 (95% CI: 1.61-3.04)times that of males from Gujarat.Furthermore, political affiliations and membership also showed an effect with males who were not a member of a political party 1.51 (95% CI: 1.01-2.24)times more likely to experience domestic violence than males who are a member of a political party.The Hosmer-Lemeshow test (p=0.19)indicated that the multiple logistic regression model for domestic violence was a good fit.

Predictors of Experiencing Honour-related Violence
The simple logistic regressions used to examine the association between the various characteristics and honour-related violence shown in Table 1 revealed that, of the 9 independent variables, 5 were significantly associated with honour-related violence.After controlling for other factors within the multiple logistic regression model, the same 5 variables remained statistically significant predictors of honour-related violence (see Table 2).Specifically, the odds of males with a tertiary education experiencing honour-related violence are 0.07 (95% CI: 0.01-0.50)times less than males with a primary school education.
As was the case for exposure to domestic and family violence, income was shown to be influential in experiences of, and exposure to, honour-related violence.Males with a monthly income of 5000-9999 INR are 0.44 (95% CI: 0.27-0.64)times less likely to experience honour-related violence compared to males with less than 5000 INR per month and males with a monthly income of 10000 INR or more are 0.41 (95% CI: 0.22-0.75)times less likely to experience honour-related violence compared to males with less than 5000 INR per month.
Hindu males are 0.63 (95% CI: 0.41-0.97)times less likely to experience honour-related violence than Muslim males.The odds of males who are not members of a religious organisation experiencing honour-related violence are 1.83 (95% CI: 1.14-2.96)times that of members of a religious organisation.Males that are not members of a political party are 1.51 (95% CI: 1.01-2.24)times more likely to experience honour-related violence than males that are members of a political party.The Hosmer-Lemeshow test (p=0.33)indicated that the multiple logistic regression model for honour-related violence was a good fit.

Discussion
This paper reports the first survey of men's experiences of domestic, family and honour-related violence in India.The results indicate that direct experience of interpersonal violence is widespread amongst the male population in these two Indian states and that certain socio-demographic factors are strong predictors of exposure.
Perhaps unsurprisingly, income emerged as a key predictor and in this study, represents the only variable that was a predictor for all three forms of violence.That is, the higher the income the less likely to experience violence.Education was a predictor for family and honour-related violence, indicating that higher levels of education potentially lower the acceptability and/or prominence of certain forms of interpersonal violence (Garcia-Moreno et al. 2006).Yet, education was not a predictor of domestic violence, indicating that domestic violence may be a trans-socioeconomic phenomenon, rather than a practice largely of the 'uneducated' (Nilan et al. 2008;Rao, 1997).It also indicates that higher levels of education may not influence behaviours and attitudes to the degree we may expect, suggesting a need to further explore wider belief systems and gender structures that operate within all echelons of Indian society.Ultimately income provided the only across-the-board influence within the group of men surveyed.
This study also indicates a life course dimension to exposure to violence with age emerging as a key predictor for family and domestic violence -the older men surveyed were much more likely to report experiencing these forms of violence.Although further qualitative research is needed to explore why this patterning is occurring, potential factors may be the pressures of increasing family size, income needed to support the extended family, ageing and workforce involvement, and so on (Martin et al. 2002;Pandey et al. 2009;Watts & Zimmerman, 2002).What is certainly clear from these results is that further work is needed to unpack how and in what ways domestic and family violence may shift in significant or prominence over the life course (Nilan et al. 2008).
The potential interplay of religion and violence is a complex issue (Doron & Broom, 2011), particularly in the South Asian context, where economic conditions and social mobility are heavily influence by religious affiliations (Nilan et al. 2008;Shani, 2007).Taking into account these complexities, we found here that religion was indeed a predictor of family and honour-related violence, with Hindu males much more likely to have experienced family violence as compared to Muslim males.Moreover, Muslim males are more likely to have experienced honour-related violence.We were also interested in whether membership of religious organisations would shape exposure to violence.It would seem, from these results, that membership of a religious organisations does reduce the likelihood of experiencing honour-related violence.Further work is needed to explore the different types of religious organisations men are involved with and to explore what impacts participation has on individual, community and cultural beliefs about violence (see also Nilan et al. 2008).
Given the political, ethnic and economic differences outlined above regarding Gujarat and Uttar Pradesh (Martin et al. 2002;Kumar et al. 2005), it is perhaps unsurprising that the results showed quite considerable differences between the two States.It was, however, surprising that 'state of residence' predicted only exposure to domestic violence, not family or honour-related violence.This does however fit with previously identified patterns in attitudes towards women and socio-economic circumstances in the two States (Martin et al. 2002;Kumar et al. 2005).Specifically, levels of deprivation in Uttar Pradesh are extreme leading to high levels of household stress and potentially perpetuating domestic violence.In Gujarat, however, economic conditions are relatively good and the State has a (proud) history of tolerance towards females.
Given the various associations identified here it is vital that we now embark on further qualitative research to explore in more depth the reasons and meanings underpinning such statistical patterns.Such work could inform, in conjunction with the results presented here, more targeted development and community initiatives seeking to reduce violence within Indian communities (Nilan et al. 2008).
We acknowledge that the results are self-reported, so there is potential for recall bias.Furthermore, we deliberately did not ask these men about their role (as perpetrators or victims) in these three forms of violence.Previous studies have shown significant underreporting in surveys of men (Kumar et al. 2005) and we wanted to gain a broader snapshot of these forms of violence in family and community life.As suggested earlier, the movement away from a focus on roles (victim/perpetrators) gives us a novel perspective in terms of broader patterns of exposure and should be augmented with other studies that focus on severity, women's experiences, qualitative accounts and so on.
The findings reported on here illustrate the ubiquity of interpersonal violence in the lives of men (and indirectly women) in Gujarat and Uttar Pradesh and the vast importance of developing further research and community intervention programs to attempt to reduce its social impacts (Hajjar, 2006).It provides an impetus to provide additional support to those cohorts of Indian community most at risk of domestic, family and honour-based violence.And finally, it reinforces the importance of including and documenting men's experiences of violence as an important part of revealing the extent and impacts of violence in contemporary India (Nilan et al. 2008).

F
statistically significant association with family violence (p<0.05)D statistically significant association with domestic violence (p<0.05)H statistically significant association with honour-related violence (p<0.05)After controlling for other factors within the multiple logistic regression model, only 4 variables remained statistically significant predictors of family violence (see Table

Table 1 .
Distribution of study sample by their characteristics and univariate logistic regression analysis for the association between these characteristics and the different forms of violence (n=1,000)

Table 2 .
Significant predictors of the different forms of violence experienced assessed through multiple logistic regression (n=1,000)