Machine Learning Analysis of Non-Marital Sexual Violence in India

Abstract

Background: Non-marital sexual violence (NMSV) in India is a health and human rights concern that disproportionately affects adolescents, is under-reported, and not well understood or addressed in the country. Machine learning techniques can explore low prevalence data to offer insight into identification of factors associated with NMSV. Methods: We applied machine learning methods to retrospective cross-sectional data from India’s nationally-representative National Family Health Survey 4, a demographic and health study conducted in 2015-16, which offers 4000+ variables as potential independent variables. We used Least Absolute Shrinkage and Selection Operator (lasso) and L-1 regularized logistic regression models as well as L2 regularized logistic regression or ridge models; and conducted a thematic analysis of variables generated from an iterative series of regularized models. Findings: Thematic analysis of regularized models highlight that exposure to violence and attitudes accepting of violence were most predictive of NMSV, and poor sexual and reproductive health knowledge second most predictive. After these, indicators largely related to resources and autonomy (e.g., income generating, freedom of movement, media access) were more likely to have experienced NMSV, but in a context of social marginalization (e.g., poorer, rural, marginalized caste). Exploratory analysis with the subsample of adolescents 15-19 years, a population with higher representation of recent NMSV, yielded similar findings, but showed higher risk for NMSV among urban girls from wealthy households. Interpretation: Upwardly mobile women and girls with greater freedom of movement, but otherwise vulnerable due to young age or circumstances, are more likely to have experienced NMSV. Funding Statement: This study was funded under a grant from the Bill and Melinda Gates Foundation (Grant number OPP1179208; PI: Anita Raj). Declaration of Interests: None to declare.