Peer-reviewed article

Small area geographic variation in girl and boy child marriage in India: a district-level longitudinal analysis, 2016 and 2021.

Gausman J, Eom YJ, Kim R, Subramanian SV · 10 Oct 2025

Abstract

Background Sustainable Development Goal (SDG) target 5.3 aims to 'eliminate all harmful practices, such as child, early, and forced marriage'. India is a signatory to the SDGs and tracks this indicator to measure national progress. Geographic inequity in the change in the district-level prevalence of child marriage over time in India has not been previously explored. Methods We used data from the National Family Health Survey (NFHS) in India from 2016 (NFHS-4) and 2021 (NFHS-5). We aligned districts to be consistent over time. We included 720 districts for girls and 716 for boys. We used a four-level logistic regression model to partition the total variation of boy (<21 years) and girl child marriage (<18 years) in each year. We computed a precision-weighted prevalence of girl and boy child marriage, as well as within-district variation, using standard deviation measures. Results Across districts, the average prevalence of child marriage decreased from 22.65% to 18.89% among girls and from 16.17% to 13.57% among boys. Inequality across districts in the prevalence of girl and boy child marriage decreased between 2016-21, and there was a negative association between district prevalence in 2016 and the change in district prevalence between 2016-21. There was a positive correlation between the difference in prevalence in boy (r = 0.72; P < 0.001) and girl (r = 0 · 60; P < 0.001) child marriage and the change in standard deviation across districts, indicating increasing inequality in the worst-performing districts. Conclusions We provide new insight into recent progress and setbacks in eliminating child marriage across India. We found considerable district-level heterogeneity and clustering of poorly performing districts, coupled with increasing local inequality, especially within the worst-performing districts. Intervention and policy may be more effective if informed by local inequality.