Logo
Munich Personal RePEc Archive

Ethnicity Matters: Correcting Cultural Bias in Social Models

Waheed, Muhammad Shahid and Taj, Naveera (2025): Ethnicity Matters: Correcting Cultural Bias in Social Models.

[thumbnail of MPRA_paper_126591.pdf]
Preview
PDF
MPRA_paper_126591.pdf

Download (451kB) | Preview

Abstract

Ethnicity Matters: Correcting Cultural Bias in Social ModelsThis study examines the determinants of domestic violence (DV—physical or sexual) against women in Pakistan using nationally representative data from the 2017–18 Pakistan Demographic and Health Survey (PDHS). We estimate survey-adjusted logistic regression models, reporting logit coefficients, odds ratios, and marginal effects, to assess the influence of intergenerational, marital, demographic, and socioeconomic factors, and test whether excluding ethnicity biases results. Findings show that intergenerational transmission is a powerful driver: women whose fathers beat their mothers face more than double the odds of experiencing DV (OR = 2.39, marginal effect +12 pp, p<0.01). Marital dynamics are equally critical. Husband’s alcohol consumption (OR = 3.96, +19 pp) and women’s fear of their husbands (OR = 5.11, +23 pp) emerge as the strongest correlates of DV. Polygamy increases the odds substantially (OR ≈ 3.50), although estimates are less precise. Household structure also matters—each additional child increases DV risk (OR = 1.16, +2 pp). Socioeconomic variables show weaker and less consistent effects. Wife’s and husband’s education, as well as consanguinity, are not robust predictors. However, household wealth displays a protective gradient: women in the richest quintile face significantly lower odds of DV (OR = 0.60, marginal effect –7 pp, p<0.1). Model comparison confirms that ethnicity is jointly significant (Wald test, p<0.01) and improves model fit. Relative to Urdu-speaking women, Sindhi (OR = 0.41, –12 pp), Pashto (OR = 0.46, –11 pp), Brahui (OR = 0.24, –18 pp), Hindko (OR = 0.28, –16 pp), and Marwari (OR = 0.19, –19 pp) women report substantially lower odds of DV, underscoring the role of cultural norms and community structures. Overall, the results highlight that excluding ethnicity biases estimated effects and conceals heterogeneity across groups. Policy interventions must therefore combine universal protections with culturally tailored strategies that recognize ethnic variation in women’s vulnerability and reporting of domestic violence.

Atom RSS 1.0 RSS 2.0

Contact us: mpra@ub.uni-muenchen.de

This repository has been built using EPrints software.

MPRA is a RePEc service hosted by Logo of the University Library LMU Munich.