The research team set out to solve a stubborn methodological problem. Although leading causal frameworks of intimate partner violence have long conceptualized patriarchy as a structural-level driver of abuse, most empirical studies have been forced to rely on individual-level proxy measures—such as a woman’s own attitudes toward wife-beating, or her husband’s controlling behavior. Those proxies conflate what a person believes with the social architecture surrounding her, making it difficult to isolate the independent contribution of structural patriarchy itself. To get around this, the team turned to a validated composite measure known as the India Patriarchy Index, or IPI, which quantifies patriarchy at the structural level across geographic units, allowing researchers to ask whether the society a woman lives in predicts her risk of violence above and beyond her own characteristics.

The data source was India’s National Family Health Survey, specifically the fifth round, known as NFHS-5, a nationally representative household survey conducted under the auspices of the International Institute for Population Sciences and India’s Ministry of Health and Family Welfare. The analytic sample included 61,180 women aged 18 to 49 who were asked about their experiences of violence at the hands of a male partner in the preceding 12 months. The survey captures multiple distinct forms of abuse: emotional violence, physical violence, and sexual violence. This granularity matters, because different forms of violence may respond to different drivers, and interventions designed to prevent one form may not automatically protect against another.

The statistical approach was deliberately rigorous. The researchers used survey-weighted mixed effects logistic regression models, a technique that accounts for the hierarchical structure of the data—women nested within districts—and adjusts for the complex survey design that makes NFHS estimates representative of the national population. Crucially, the models did not simply compare raw rates of violence across districts. Instead, they adjusted for an extensive battery of potential confounders spanning other social stratification systems, including caste, religion, and household wealth, as well as attitudinal and behavioral indicators and a full set of socio-demographic characteristics. This adjustment strategy is what allows the analysis to disentangle structural patriarchy from other structural and individual-level determinants of violence—many of which, such as gender attitudes, have historically been used as stand-ins for patriarchy itself.

The headline result is striking. Compared with women living in districts with lower levels of structural patriarchy, women in districts with the highest levels of structural patriarchy had 1.9 times the odds of reporting experience of intimate partner violence of any type. In statistical terms, the adjusted odds ratio was 1.92, with a 95 percent confidence interval of 1.58 to 2.34. Because that entire interval sits well above 1.0, the association is unlikely to be a statistical fluke. The researchers also found a positive linear pattern, meaning that risk rose steadily across the spectrum of structural patriarchy rather than appearing only at the extreme end—a dose-response relationship that strengthens the case for a genuine association.

Importantly, the team disaggregated their models by the type of violence reported, examining any IPV, emotional-only violence, physical-only violence, sexual-only violence, and the experience of all forms simultaneously. This disaggregation reflects a growing recognition in the field that intimate partner violence is not a monolith. Emotional abuse, physical assault, and sexual coercion can have different etiologies, different correlates, and different consequences for survivors. By modeling each form separately, the study provides a more textured picture of how structural patriarchy relates to the full landscape of abuse, rather than collapsing everything into a single binary outcome.

What makes this study methodologically distinctive is the way it separates levels of analysis. The intraclass correlation coefficient—a statistic that quantifies how much of the variation in an outcome occurs between clusters rather than within them—plays a key role in this kind of multilevel modeling, and the mixed effects framework allowed the researchers to partition variance in violence risk across individual and district levels. The researchers also examined the spatial patterning of patriarchy using Local Indicators of Spatial Association, or LISA, a technique borrowed from spatial statistics that identifies geographic clustering of high and low values on a map. Together, these tools let the team describe where structural patriarchy concentrates in India and test whether that concentration translates into measurable differences in women’s safety.

The study’s intellectual lineage is also notable. The authors acknowledge Siegfried Gruber and Mikołaj Szołtysek, whose development of the original Patriarchy Index in the European context laid the groundwork for the India-specific adaptation used here. That index approach represents a conceptual shift: rather than treating patriarchy as an attitude to be surveyed, it treats it as a measurable property of social systems, assembled from indicators that capture how power, resources, and authority are distributed between men and women across a population. Applying such an index to a survey as large and methodologically mature as NFHS-5 allows researchers to test structural theories of violence at a scale that was previously impractical.

The implications for policy are significant. The authors argue that their findings underscore the need for district-level planning in violence prevention, and for the use of structural indicators alongside individual- and couple-level gender norms transformative programming. In practical terms, this means that programs aimed at changing the attitudes and behaviors of individual couples—valuable as they are—may be insufficient if the surrounding social structure continues to concentrate power and normalize male authority. India’s national health and rights initiatives, the study suggests, should incorporate structural measures of patriarchy when targeting resources and designing interventions, so that prevention strategies are calibrated to the structural realities of each district rather than applied uniformly across a vast and heterogeneous country.

The research also carries a broader message for global public health. Intimate partner violence affects millions of women worldwide, and the field has made enormous progress in cataloging its individual-level risk factors, from exposure to violence in childhood to alcohol use and economic dependency. But this study demonstrates that the social structures within which individuals are embedded carry independent predictive weight. A woman’s risk of being beaten, coerced, or humiliated by her partner depends in part on how patriarchal her district is—after accounting for her caste, her religion, her wealth, her own attitudes, and her personal circumstances. That is a sobering conclusion, but also an actionable one: if structures can be measured, they can be monitored, compared, and ultimately changed. The study, a secondary analysis of publicly available data that received ethical exemption from the University of California San Diego Institutional Review Board, was supported by the Bill & Melinda Gates Foundation and the National Institute on Alcoholism and Alcohol Abuse, with no funder involvement in design, analysis, or interpretation. As India and other countries confront the scale of gender-based violence, the evidence points toward a prevention agenda that operates at every level—from the couple, to the community, to the structure of society itself.

Subject of Research: The association between structural-level patriarchy and intimate partner violence in India using the India Patriarchy Index and national demographic and health survey data.

Article Title: Examining the association between structural-level patriarchy and intimate partner violence: evidence from India’s national demographic and health survey

Article References: Barker, K. M., Patwardhan, V., Chokhandre, P., Singh, A., Singh, A., & McDougal, L. (2026). Examining the association between structural-level patriarchy and intimate partner violence: evidence from India’s national demographic and health survey. International Journal for Equity in Health. https://doi.org/10.1186/s12939-026-03010-x

Image Credits: AI Generated

DOI: 10.1186/s12939-026-03010-x

Keywords: intimate partner violence, patriarchy, India, National Family Health Survey, gender norms, domestic violence, structural determinants, public health, multilevel modeling, district-level planning, gender equity, NFHS-5

The research, conducted by Susanta Sen, Asif Ali, Amit Banerjee, and Namita Chakma of the Department of Geography at The University of Burdwan in West Bengal, drew on two rounds of India’s National Family Health Survey: the fourth round, conducted in 2015–16 with 122,955 women, and the fifth round, conducted in 2019–21 with 118,700 women. The analysis focused on women aged 20 to 24 years, the standard age group used to assess child marriage prevalence under international indicators, since these women had passed beyond the age at which a legal marriage before 18 could have occurred. By comparing the two survey rounds, the researchers were able to track both the level of child marriage and the distribution of that practice across the wealth spectrum over roughly a five-year interval.

To measure inequality rigorously, the team employed the Erreygers Concentration Index, a widely used summary measure in health equity research that captures whether a health or social outcome is systematically concentrated among richer or poorer groups. A negative value of the index indicates that the outcome is concentrated among the economically disadvantaged. In both survey rounds, the Erreygers Concentration Index for child marriage was negative and statistically significant, confirming that the burden of marrying before 18 falls disproportionately on girls from poorer households. The index moved from −0.281, with a 95 percent confidence interval of −0.289 to −0.273, in the fourth round to −0.257, with a 95 percent confidence interval of −0.265 to −0.249, in the fifth round, indicating a modest narrowing of wealth-related inequality but no fundamental change in its direction or magnitude.

The absolute picture tells a similarly nuanced story. The absolute gap in child marriage prevalence between the poorest and richest groups narrowed from 20.65 percentage points in NFHS-4 to 19.04 percentage points in NFHS-5. Yet the relative measure moved in the opposite direction: the ratio of child marriage prevalence among the poor compared with the non-poor actually increased, from 2.07 to 2.20. This combination of a shrinking absolute gap and a widening relative gap is a classic pattern in epidemiology and social statistics, arising when prevalence declines faster in absolute terms among groups that already have lower baseline levels. In practical terms, even though fewer girls overall are marrying as children, a girl born into a poor household today remains substantially more exposed to the risk of early marriage than one born into affluence, and in relative terms that exposure has grown.

Beyond describing the inequality, the study’s central methodological contribution lies in its use of regression-based decomposition analysis, a technique that partitions the observed concentration index into contributions attributable to individual socio-demographic characteristics. Rather than simply noting that child marriage correlates with poverty, decomposition analysis quantifies how much of the measured inequality each factor statistically explains. The results were striking in their consistency: educational attainment dominated the explained component of wealth-related inequality in both survey rounds, accounting for 77.06 percent in NFHS-4 and 77.41 percent in NFHS-5. In other words, differences in schooling levels between rich and poor women are by far the largest statistical driver of why early marriage clusters among the disadvantaged.

The composition of the remaining explained inequality shifted over time in ways that offer clues about the changing social landscape. In the fourth survey round, place of residence, meaning the divide between urban and rural areas, was the second-largest contributor to wealth-related inequality in child marriage. By the fifth round, however, the contribution of mass media exposure had risen substantially, overtaking residential location as a key explanatory factor. This shift suggests that access to information, whether through television, radio, newspapers, or increasingly digital channels, has become a more salient axis of inequality as connectivity spreads unevenly across Indian society. It also implies that the messages, norms, and alternatives that media exposure conveys about marriage, gender roles, and women’s life courses play a measurable role in shaping when and whether girls marry.

The study also documented substantial variation across India’s states and Union Territories, underscoring that child marriage is not a uniform national phenomenon but one embedded in highly diverse regional contexts of culture, economy, and policy implementation. States differ enormously in baseline prevalence, in the steepness of the wealth gradient, and in the pace of change between the two survey rounds. This heterogeneity means that a single national policy lever, applied uniformly, is unlikely to produce equitable results. Regions where the wealth gradient is steepest may require intensively targeted approaches, while regions with lower but persistent prevalence may benefit more from broad norm-change campaigns and enforcement of existing legal prohibitions on marriage below the age of 18.

The public health significance of these findings extends well beyond the institution of marriage itself. Child marriage is internationally recognized as a driver of adverse outcomes including early pregnancy, elevated maternal and infant mortality risks, interrupted schooling, restricted economic agency, and increased vulnerability to domestic violence. When the burden of child marriage concentrates among the poorest girls, it functions as a mechanism through which poverty and disadvantage reproduce themselves across generations. The study’s conclusion that reductions in overall prevalence have not been matched by equivalent reductions in socioeconomic inequality echoes a broader theme in development research: aggregate progress can mask persistent, and sometimes widening, disparities that demand explicit equity-oriented policy attention.

The authors point toward a set of policy directions grounded in their decomposition results. Because education explains such an overwhelming share of the inequality, interventions that support girls’ secondary education, reduce the economic pressures that push families to withdraw daughters from school, and keep adolescents enrolled through the years when marriage decisions are typically made stand out as the highest-leverage options. The growing contribution of mass media exposure suggests that improving access to information, and using media channels deliberately to shift social norms around early marriage, could complement educational investments. The researchers also highlight strengthening social protection for economically disadvantaged households and explicitly prioritizing the poorest families in programmatic targeting, since the data confirm that the risk of child marriage remains most intense at the bottom of the wealth distribution.

Ultimately, the study delivers a message that is both encouraging and cautionary. India has made genuine progress: child marriage prevalence declined measurably between 2015–16 and 2019–21, and the concentration index softened slightly. But the persistence of a large absolute wealth gap, the rise in the poor-to-non-poor prevalence ratio, and the stark regional differences all indicate that the finish line remains distant for the girls whom the practice continues to touch most directly. As India and the global community pursue the elimination of child marriage under international development goals, this research argues that success should be judged not only by how far national averages fall, but by how quickly the gap closes for those at the bottom, where the consequences of marrying too young are felt most severely.

Subject of Research: Wealth-related socioeconomic inequality in child marriage among young women in India

Article Title: Wealth-related inequality in child marriage in India using decomposition analysis

Article References: Sen, S., Ali, A., Banerjee, A., & Chakma, N. (2026). Wealth-related inequality in child marriage in India using decomposition analysis. Discover Social Science and Health. https://doi.org/10.1007/s44155-026-00488-2

Image Credits: AI Generated

DOI: 10.1007/s44155-026-00488-2

Keywords: child marriage, India, wealth inequality, Erreygers Concentration Index, decomposition analysis, National Family Health Survey, girls' education, mass media exposure, social protection, public health, NFHS-5, socioeconomic disparity