Exploring the Paradox of Intimate Partner Violence and Increased Contraceptive Use in sub-Saharan Africa.

We question the positive effect of intimate partner violence on women's modern contraceptive use in sub-Saharan Africa found in previous studies. The explanations offered for this counter-intuitive result are either that women make greater efforts to avoid childbearing in conflictual relationships, or that endogeneity bias exists. Endogeneity bias stems from the inability of researchers to attribute a specific cause to one variable when they are unable to control for related missing covariates. Demographic and Health Survey data from 13 countries in sub-Saharan Africa provide evidence for the latter but not the former. Indeed, using simple probit regression models, we observe a positive relationship between intimate partner violence and modern contraceptive use in Burkina Faso, Mali, Nigeria, Tanzania, and Zimbabwe. This effect remains unchanged when controlling for various measures of women's autonomy in the household, showing that these two variables interact with contraceptive use independently. However, the use of recursive bivariate probit and Rosenbaum bounds sensitivity analysis to control for endogeneity biases erodes the initial positive effect in the five countries, although only partially in Burkina Faso. Our research shows that the previously reported findings arise from poor model specification and highlights the need for more appropriate data to assess the effect of intimate partner violence on modern contraceptive use in sub-Saharan Africa.