Challenging assumptions about women's empowerment: social and economic resources and domestic violence among young married women in urban South India

Abstract

Background Although considerable research has documented the widespread prevalence of spousal violence in India, little is known about specific risk or protective factors. This study examines the relationships between factors that are often considered to be social and economic resources for women and recent occurrence of domestic violence.

Methods Data were collected from 744 young married women in slum areas of Bangalore, India. Unadjusted and adjusted multivariable logistic regression models were used to determine factors associated with having been hit, kicked or beaten by one's husband in the past 6 months.

Results Over half (56%) of the study participants reported having ever experienced physical domestic violence; about a quarter (27%) reported violence in the past 6 months. In a full multivariable model, women in ‘love’ marriages (OR = 1.7, 95% CI 1.1–2.5) and those whose families were asked for additional dowry after marriage (OR = 2.3, 95% CI 1.5–3.4) were more likely to report domestic violence. Women who participated in social groups (OR = 1.6, 95% CI 1.0–2.4) and vocational training (OR = 3.1, 95% CI 1.7–5.8) were also at higher risk.

Conclusions Efforts to help women empower themselves through vocational training, employment opportunities and social groups need to consider the potential unintended consequences for these women, such as an increased risk of domestic violence. The study findings suggest that the effectiveness of anti-dowry laws may be limited without additional strategies that mobilize women, families and communities to challenge the widespread acceptance of dowry and to promote gender equity. Longitudinal studies are needed to elucidate the complex causal relationships between ‘love’ marriages and domestic violence.

Introduction

Over a decade ago, the United Nations General Assembly called for increased research into the ‘causes, nature, seriousness and consequences’ of violence against women, reflecting a recognition that such violence represents a fundamental violation of women's health and human rights.¹ Since then, studies across India have documented the widespread prevalence of domestic violence, encompassing an array of physical, sexual and/or psychological acts inflicted by intimate male partners. However, little remains known about specific risk or protective factors. For example, a number of South Asian studies have examined the role of stressors, such as lower household socioeconomic status, level of education, low earning potential and young age at marriage and have found these factors to be associated with higher reported physical violence.^2–6 On the other hand, other studies have found no relationship between household poverty and violence⁷ or a positive relationship between women's level of education and violence.⁸

The links between gender-based power and domestic violence are widely recognized, with violence being viewed both as a manifestation of deeply entrenched gender power inequities as well as a mechanism by which such inequities are enforced.⁹ In spite of the broad consensus on the underlying role of inequities in gender-based power in the incidence of domestic violence, few studies have comprehensively examined the roles of specific aspects of power inequities that may pose a risk for domestic violence.

For instance, studies in South Asia have examined women's power in marital relationships in terms of power arising from conformity to social norms and expectations, such as how big a dowry she has brought to the marital household.^7,8,10 At the time of marriage, a bride's family typically gifts the new couple and the groom's family with money and goods to start a new home, a practice that has become customary across caste and class groups in India.¹¹ Women with relatively smaller dowries, those whose in-laws have expressed dissatisfaction with their dowries, and those who have faced post-marriage dowry requests have been repeatedly found to be more likely to report domestic violence.^2–4,8,10 Research in Tamil Nadu indicates that items given as dowry may provide a woman with financial resources that allow her to negotiate her status in the marital household.⁸ In addition, the amount of dowry given may influence the social standing of the husband's family, with a larger dowry reflecting greater social status.⁸ However, young people are increasingly choosing their marriage partner, known in local parlance as ‘love’ marriages (as opposed to a marriage arranged at least in part by family members). While the rise of ‘love’ marriages and associated factors has been studied in a number of Asian societies, the impact of these social trends on dowry practices and subsequent relationship dynamics, including domestic violence, remains unexplored.^12–14

Studies have yielded conflicting information on the association between violence and women's empowerment, particularly in terms of economic opportunity, control of assets and social group participation. While women with greater economic resources, such as ownership of land, jewelry and other valuables, were less likely to report violence in one study,³ employed women have been found to report violence more frequently than unemployed women in a number of other studies.^3,4,15 The health implications for women who participate in community social groups have not been examined in India. However, studies in Bangladesh have documented both elevated as well as lowered odds of reported violence among women participating in microcredit programmes.^7,16

Programme strategies and policy recommendations often assume that helping women to empower themselves, particularly through female education, vocational training and employment, will provide women with resources that will in turn decrease their risk of adverse reproductive health outcomes. However, there is a growing recognition that these strategies may conflict with prevailing social norms and expectations that are relatively slow to change, and may instead result in increased violence against women.^7,16,17 In particular, women's exercise of autonomy through, for example, participation in micro-savings and credit groups in conservative social settings may be ‘viewed as provocative because they undertake actions that challenge prevailing norms governing women's propriety and modesty’.¹⁶ Thus, it is critical that research examining factors associated with violence be grounded in an in-depth understanding of prevailing sociocultural norms.

The Samata Health Project is an ongoing women's health research study looking at issues of gender, sexuality, reproductive health, violence and sexually transmitted infection (STI) and HIV prevention among vulnerable women and men in Bangalore, the capital city of Karnataka state in southern India. In this article, we use data from this study to examine the relationships between economic and social factors that represent different dimensions of gender-based power and reported experience of physical domestic violence among young, married women in low-income urban communities. In addition, the current research explores whether certain aspects of empowerment, such as women's employment, vocational training and social group participation, affect risk of violence differentially if they occurred prior to or after the time of marriage.

Methods

Study setting and design

Between August 2005 and February 2006, a convenience sample of 744 young married women was recruited from two low-income communities in Bangalore. The working class communities, classified as ‘slums’ by the Bangalore municipal government, were originally established in the mid-20th century, when trade and employment opportunities drew large numbers of migrants from the neighbouring state of Tamil Nadu into Bangalore.¹⁸ These communities have continued to grow and are now home to almost one-fifth of Bangalore's population.

Trained field staff approached potential participants in two government primary health centres and surrounding communities through outreach activities and door-to-door visits. Given the poorly demarcated and often unnamed lanes characteristic of these densely populated slum communities, community mapping and purposive sampling were not feasible. Eligibility criteria included being a married female between 16 and 25 years of age, fluency in one of the local languages (Tamil or Kannada), and anticipating residence in the community for the duration of the 2-year study. Guardian consent was obtained for women under 18 years of age.

Enrolled women participated in face-to-face interviews conducted in private rooms in the health centre by trained interviewers. The survey collected information on sociodemographics; household and relationship characteristics; economic activity and assets; sources of social support; and reasons for and responses to marital conflict.

Data collection is ongoing, with participants completing surveys once a year for 2 years. Data used in this analysis are from the baseline visit. The protocol was approved by the human subjects’ protection committees of the University of California, San Francisco and the Indian Institute of Management, Bangalore.

Measures and hypotheses

This study is guided by Kabeer's conceptualization of women's empowerment in which women's power emerges as a result of the dynamic links between their access to and control over resources, including economic resources, such as education and employment, and social resources, such as social support, norms and expectations.¹⁹ According to this framework, women's limited access to resources combined with constraints in decision making heightens their risk of adverse health outcomes, including domestic violence. Measures for inclusion in this analysis (described in Table 1) were guided by Kabeer's framework, the relevant literature and formative qualitative research conducted between 2002 and 2004 prior to the initiation of this study. The latter involved 18 focus group discussions and 40 in-depth interviews with married women and men aged 18–49.²⁰

Table 1.

Description of measures

Variable	Coding	Description
Sociodemographic factors
Age at enrollment	Discrete, in years	Age of participant at baseline. Self-reported.
Age at marriage	Discrete, in years	Age participant was married. Self-reported.
Duration of marriage	Discrete, in years	Years between age of participant at marriage and at baseline.
Age of husband	Discrete, in years	Participant's husband's age at baseline. Reported by participant.
Education	Discrete, in years	Number of years of education completed.
Household asset score	Continuous. Normally distributed.	Score calculated using factor analysis of reported possession of a set of household items.a
Religion	Hindu	Religion. Self-reported.
	Muslim
	Christian
Hypothesized protective social/economic factors
Dowry given at marriage	Yes/no	Participant's natal family gave household items or money to the married couple and/or husband's family at the time of marriage.
Natal family provides any financial support	Yes/no	Participant's natal family has ever given the married couple money or other support.
Husband has occupational stability	Yes/no	Participant's husband has had no trouble keeping a stable job over the past year and has had a full-time job (as opposed to part-time or seasonal work).
Social group participation before marriage	Yes/no	Participated in a group (youth group, self-help group or savings group) at any time prior to marriage.
Social group participation after marriage	Yes/no	Participated in a group (youth group, self-help group or savings group) at any time since marriage.
Hypothesized risky social/economic factors
Marriage type	‘Love’ marriage Arranged marriage	A love marriage is a marriage in which the participant reported that she and her husband decided themselves to get married. An arranged marriage is one in which the participant reported that the couple's elders (family members) determined that the couple would be married.
Household composition	Nuclear only Nuclear with only in-laws Nuclear with any natal family member	Composition of household in which participant lives. Nuclear only means the participant lives only with her husband and children. Nuclear with in-laws means any member of the husband's family resides with the participant. Nuclear with any natal family members means any of the participant's own family live with her.
Dowry requested after marriage	Yes/no	In-laws requested dowry of the participant's natal family after the time of marriage.
Vocational training before marriage	Yes/no	Participated in a vocational or skills-building training at any time prior to marriage.
Vocational training after marriage	Yes/no	Participated in a vocational or skills-building training at any time since marriage.
Work before marriage	Yes/no	Worked for money at any point prior to marriage.
Work after marriage	Yes/no	Worked for money at any point since marriage.
Outcome
Spousal violence	Yes/no	Participant's husband hit, kicked or beat her for any reason in the 6 months prior to study enrolment.

^aAsset score based on Vyas and Kumaranayake, 2006.²¹

The outcome of interest, experience of domestic violence, was based on self-report of having been ‘hit, kicked or beat’ by one's husband (yes/no) for any reason in the 6 months prior to study enrolment.

Independent variables were selected on the basis of being social and economic factors with either the potential to put women at increased risk of domestic violence or the potential to serve as protective resources. Measures of factors both prior to marriage and after marriage were included for certain variables to assess whether resources that a woman brings into her marriage have a different relationship to domestic violence than those acquired within the marital context. The examination of resources from before marriage allows us to draw stronger conclusions about the temporality of the relationship of these exposures to violence within marriage.

It was hypothesized that the following factors would be associated with an increased risk of domestic violence: receiving dowry requests after marriage and marrying by own choice (a ‘love’ marriage). Because the formative data suggested that mothers-in-law sometimes encourage and participate in abuse of their sons’ wives, the participant's household composition was also examined, and it was hypothesized that having members of the husband's family in the household would increase risk for women. The formative research also indicated that paid employment was driven by economic necessity, specifically by husbands’ failure to provide adequately for the family and associated with marital conflict. Thus, measures of whether the participant had engaged in paid employment or participated in vocational training prior to or after marriage were included.

Protective social and economic factors that were considered included a woman's family providing dowry at marriage and financial support post-marriage and husband's occupational stability. Women repeatedly described during pre-study in-depth interviews that marital conflicts arose from their husbands’ inability to contribute adequate income to meet household needs as well as their natal families' inability to offer support. It was therefore hypothesized that financial support from the wife's natal family and husband's occupational stability would be associated with decreased domestic violence.

Participation in a social group, such as a savings and loan or youth, religious or political group, both prior to and after marriage, was also assessed as a potential resource that could protect women from marital conflict and abuse. This hypothesis is based in our conceptual model in which access to resources, including social resources, may serve to protect women from negative health outcomes.¹⁹

Sociodemographic variables that have been associated with domestic violence (i.e. age, age at marriage, religion and education) were also included in the analysis. To measure household socioeconomic status, an asset score was created based on questions assessing household possessions and facilities using factor analysis.²¹

Analysis

The relationships among related independent variables were assessed using contingency tables, chi-square analyses and Student's t-tests to ensure that two highly correlated variables were not both included in multivariable analyses. The strength of the independent association of each social and economic predictor variable with the outcome was then assessed using a univariable logistic regression model. A full multivariable logistic regression model was then run to examine the association of each variable with domestic violence, adjusting for the others in the model. All analyses were conducted using Stata 10 (College Station, TX, USA).

Results

The average age of study participants at baseline was 22 years, while their husbands were on average 5 years older. Education levels were low, with 18% reporting no formal education. Over half of participants are Hindu (54%), 35% are Christian and 11% are Muslim. Two of every three participants reported being in arranged marriages; the remaining third reported having chosen their spouse. Two-thirds of participants reported working before marriage, mostly as garment factory workers or domestic workers. Over half (56%) of participants reported having ever experienced physical domestic violence; about a quarter (27%) reported physical domestic violence in the past 6 months. The sample characteristics are summarized in Table 2.

Table 2.

Participant characteristics at baseline (n = 744)

	Mean (years)	SD
Sociodemographic factors
Age at enrollment	22.3	2.3
Age at marriage	18.0	2.4
Duration of marriage	4.3	2.9
Age of husband	27.6	3.7
Education	5.9	3.6

	n	Percentage
Interview language
Kannada	219	29.4
Tamil	525	70.6
Religion
Hindu	400	53.8
Muslim	82	11.0
Christian	261	35.1
Parity
Nulliparous	131	17.6
1	221	29.7
2+	392	52.7
Hypothesized protective social/economic factors
Dowry given at marriage	590	79.3
Natal family provides any financial support	568	76.3
Husband has occupational stability	113	15.2
Social group participation before marriage	35	4.7
Social group participation after marriage	164	22.0
Hypothesized risky social/economic factors
Marriage type
‘Love’ marriage	248	33.3
Arranged marriage	496	66.7
Household Family Structure
Nuclear with only in-laws	288	38.7
Nuclear only	350	47.0
Nuclear with any natal family members	106	14.3
Dowry requested after marriage	161	21.7
Vocational training before marriage	230	31.0
Vocational training after marriage	55	7.4
Worked before marriage	501	67.3
As a garment factory worker a	147	29.3
As a domestic worker a	154	30.7
Worked after marriage	293	39.4
As a garment factory worker a	44	15.0
As a domestic worker a	162	55.3
Spousal violence
Ever experienced spousal violence	418	56.2
Experienced spousal violence, past 6 months	199	26.8

^aDenominator of percentage is women who worked.

Table 3 summarizes the results of logistic regression analyses. Many of the factors hypothesized to increase risk were indeed positively associated with domestic violence. In the unadjusted model, women in ‘love’ marriages had almost twice the risk (OR = 2.0, 95% CI 1.5–2.8); this association remained strong in the model adjusting for other variables (OR = 1.7, 95% CI 1.1–2.5). As expected, women whose families were asked to pay additional dowry after marriage had higher levels of violence in both the unadjusted (OR = 2.6, 95% CI 1.8–3.8) and adjusted (OR = 2.3, 95% CI 1.5–3.4) models.

Table 3.

Unadjusted and adjusted multivariable logistic regression analyses for experience of domestic violence in the previous 6 months,^a at enrolment (n = 738)

	Unadjustedb	Adjustedb
	OR (95% CI)	OR (95% CI)
Sociodemographic factors
Age at enrollment (years)	1.01 (0.94–1.09)	1.05 (0.96–1.14)
Age at marriage (years)	0.93 (0.87–1.00)	0.93 (0.86–1.01)
Religionc
Muslim	0.75 (0.42–1.32)	0.73 (0.39–1.36)
Christian	0.95 (0.67–1.35)	0.86 (0.58–1.27)
Education (years)	0.97 (0.93–1.02)	1.00 (0.94–1.08)
Asset score	0.76 (0.64–0.90)	0.78 (0.63–0.96)
Hypothesized protective social/economic factors
Dowry given at marriage	0.61 (0.41–0.89)	0.80 (0.51–1.25)
Natal family provides any financial support	1.06 (0.72–1.58)	1.04 (0.67–1.59)
Husband has occupational stability	0.58 (0.35–0.96)	0.81 (0.47–1.41)
Social group participation before marriage	1.89 (0.94–3.80)	1.99 (0.93–4.27)
Social group participation after marriage	1.51 (1.04–2.21)	1.55 (1.01–2.38)
Hypothesized risky social/economic factors
Love marriagec	2.03 (1.45–2.83)	1.68 (1.13–2.48)
Household compositionc
Nuclear with in-laws only	0.89 (0.62–1.28)	1.18 (0.79–1.79)
Nuclear with any natal family members	1.25 (0.77–2.00)	1.69 (1.00–2.88)
Dowry requested after marriage	2.61 (1.80–3.77)	2.30 (1.54–3.41)
Vocational training before marriage	1.28 (0.91–1.81)	1.42 (0.96–2.11)
Vocational training after marriage	2.68 (1.54–4.69)	3.11 (1.67–5.79)
Work before marriage	1.65 (1.14–2.38)	1.47 (0.98–2.22)
Work after marriage	1.61 (1.16–2.24)	1.19 (0.82–1.72)

^aDependent variable is domestic violence, defined as having been ‘hit, kicked or beat’ by one's husband (yes/no) for any reason in the 6 months prior to study enrolment.

^bUnadjusted ORs represent the independent effect of each variable on the outcome. Adjusted ORs are from a model including all variables in the table.

^cReference category for religion is ‘Hindu’. Reference category for ‘love marriage’ is ‘arranged marriage’. Reference category for household composition is ‘nuclear family only’.

In terms of employment, women who worked before marriage, worked after marriage and participated in vocational training after marriage were more likely to report domestic violence, based on individual unadjusted models. Vocational training participation after marriage was the only employment variable that was associated with violence in the adjusted model; women who received training had three times the odds of reporting recent domestic violence (OR = 3.1, 95% CI 1.7–5.8).

Results pertaining to potentially protective social and economic resources were not entirely consistent with the hypotheses. Women whose families paid dowry at the time of marriage were at reduced odds of violence in the unadjusted model (OR = 0.6, 95% CI 0.4–0.9), although the relationship was attenuated in the adjusted model. Women whose husbands had more stable work situations were also at decreased risk (OR = 0.6, 95% CI 0.4–1.0); this association also weakened in the adjusted model. Finally, contrary to expectations, social group participation while married was associated with an increased risk of spousal violence in both the unadjusted (OR = 1.5, 95% CI 1.0–2.2) and adjusted (OR = 1.6, 95% CI 1.0–2.4) models. Among participants, 22% reported being part of a group, and the most common types of groups were self-help, savings fund or women's groups. Social group participation prior to marriage also was weakly associated with increased odds (OR = 1.9, 95% CI 0.9–3.8).

Women who entered into marriage at later ages were at slightly decreased risk of violence in both models (OR = 0.9, 95% CI 0.9–1.0). In addition, higher household assets were associated with decreased risk of spousal violence in both the unadjusted (OR = 0.8, 95% CI 0.6–0.9) and adjusted (OR = 0.8, 95% CI 0.6–1.0) models.

Discussion

The limitations of this study should be noted outright. First, all data were self-reported by participants and therefore subject to recall and social desirability biases, particularly data that relate to the private realm, such as experiencing domestic violence.²² We took measures to reduce participants’ hesitancy to report violence. Female interviewers were recruited from local communities and engaged in extensive and ongoing research methods training. Further, we conducted formative qualitative research over a 2-year period to develop rapport with potential participants and local community members. It is possible also that differential reporting by certain traits, such as age, is at least in part responsible for the associations we detect. However, our formative work indicated that domestic violence is considered to be normal and socially accepted across women of varying traits in study communities. As such, willingness to report violence is likely higher across women in the study communities than in other settings where violence is more stigmatized. Regardless, our results should be interpreted carefully and deemed suggestive rather than definitive.

Second, because cross-sectional data were used, it was not possible to establish causal relationships between the factors studied and domestic violence, underscoring the need for longitudinal data. Finally, our sample is relatively small and may differ from slum communities as a whole. However, since 95% of urban women in Karnataka access antenatal care²³ and the majority of poor women obtain care in public primary health care facilities,²⁴ we believe our recruitment strategy reached a population reasonably similar to that of slum communities.

In spite of these limitations, our study sheds light on potential determinants of violence in India. It is the first to examine the relationship between ‘love’ marriage and a women's health outcome. The finding that women in ‘love’ marriages were more likely to report recent physical violence than women whose marriages were arranged is intriguing. While the term ‘love’ connotes romance, ‘love’ marriages in the Indian context are not necessarily a mere result of romance and self-choice. A woman may decide or be forced to marry her boyfriend because a premarital relationship is discovered or because she became pregnant. Alternatively, youth who are attracted to one another simply may not consider or have any opportunity for dating and instead default into marriage. According to ethnographic research in Bangalore, in many of these circumstances, women lose important social and economic support from their natal families, including the provision of dowry. Therefore, they have fewer resources to draw on in their marital household.²⁰ Participants revealed that these conditions often precipitated marital conflict, including post-marriage dowry harassment by in-laws and led to domestic violence.

Unexpectedly, even after controlling for post-marriage dowry requests by the husband or his family and the receipt of natal family financial support after marriage, the relationship between ‘love’ marriage and domestic violence persisted. The social repercussions of ‘love’ marriages are likely to go beyond the act of dowry. Even if dowry is provided by women's families, social support and familial involvement in mitigating marital conflict may be severely limited. In fact, couples who choose to marry each other may be ‘uniquely vulnerable to the violence and excesses of … their families’.²⁵ Indeed, the large majority of women in ‘love’ marriages (84%) in our study reported that their parents and family members were angry with them for deciding to marry their spouse; over half (56%) were either disowned or hit, kicked or beaten by family members for the decision. Almost a third (30%) reported that neighbours and community members spoke badly about them. Individuals in such circumstances may be even less prepared to handle the demands and stresses of a marital relationship than those in an arranged marriage and therefore the risk of domestic violence may be greater. Further research on the characteristics of ‘love’ marriages, including the extent of parental approval obtained by couples, is needed to shed light on this association.

Despite the relatively long history of public and private activism against the practice of dowry in India, this study reveals its continued social value and importance.^26,27 Consistent with previous studies, a slightly reduced odds of reported violence was found among women who were given some form of dowry at the time of marriage, and post-marriage requests for dowry by the husband or his family were associated with increased violence.^3,8,28 The reduced effect of dowry at marriage in the adjusted model is likely due to the inclusion of ‘love’ marriage and post-marriage dowry harassment in the model. Still, our findings provide further evidence that the practice of dowry is a deep reflection of the many ‘forms of gender inequality that women experience’.¹¹ Given that its practice is so pervasive and routine in many communities, the effectiveness of anti-dowry and anti-violence laws, even if actively implemented, may be limited.³ Additional strategies which mobilize women, families and communities to challenge the widespread acceptance of dowry and violence and to promote gender equity are urgently needed.

Another finding of note is the increased odds of reported violence among women who participated in a social group after marriage, as well as the elevated odds associated with social group participation prior to marriage. These findings, taken together with the elevated odds of reported violence among women who attended vocational training, suggest that although there do not appear to be overt restrictions on women's mobility in these communities, unspoken norms pertaining to women's mobility may be operating. Women defying these norms may be risking domestic conflict and violence, suggesting that investigation of subtle forms of restrictions on young, married women's mobility and social interaction in these urban communities and their implications for women's access to economic and social resources is warranted.

Conclusion

This research illustrates the complexity of the meanings of and relationships between economic and social measures of women's resources and their experience of domestic violence. A woman's conformity to family and community norms and expectations, either by choice or as a result of prevailing circumstances, may serve to protect her against domestic violence. On the other hand, defiance of or an inability to conform to social norms is liable to increase her risk of experiencing domestic violence. This scenario suggests that the meaning of indicators of power needs to be interpreted within the social context in which they are being researched. Our findings also indicate that in settings where gender inequities are deeply entrenched, enhancing women's social or economic resources may result in increased domestic violence. Perhaps this is a temporary phenomenon, which disappears over time. Nonetheless, when combined with existing evidence, the findings point to the urgent need to engage with and co-opt men, families and communities with regards to gender justice and equity.

Funding

A gift from the Levis Strauss Foundation (LSF), USA; National Institute of Child Health and Human Development (R01 HD041731 to S.K.).