Publication:

Comparative analysis of regression models in predicting Intimate partner violence: a gender specific analysis using NFHS-5 data

Background Existing regression models for determining intimate partner violence (IPV) against women include characteristics of women and men individually but evidence lacking on couple characteristics and best fit models. We aimed to compare regression models in predicting IPV using characteristics of women and her partner separately and as a dyad.

Methods We used couple data from National Family Health Survey-5 containing data from both women and her partner. IPV considered as positive response for any of the questions on domestic violence. Multiple logistic regression models were built separately for characteristics of women and men and couple characteristics. Best fit models identified through r-squared value and AIC and BIC statistics.

Results We included 46445 couple data for final analysis. R-squared value for models including female characteristics were similar to models with couple characteristics and higher than the models with male characteristics. Odds Ratio were also higher for women characteristics compared to their counterparts’ characteristics. Family history of IPV among women, women’s attitude towards wife beating and controlling behavior of partner were associated with higher IPV.

Conclusions Individual characteristics of women in a couple alone can be an effective determinant of IPV than the traditional view of power relations between men and women in household.