Identification of Associated Factors and Prediction for the Level of Intimate Partner Violence against Women in Sri Lanka

Abstract

Intimate partner violence (IPV) can be defined as a serious social problem rapidly increasing in Sri Lanka as same as in other countries in the world. The Sri Lanka Demographic and Health Survey (SLDHS) 2016 revealed that 17% of married women age 15-49 in Sri Lanka have become victims of IPV. The objectives of this study were to determine the factors associated with IPV against women in Sri Lanka and to develop an appropriate regression model and feedforward neural network (FNN) to predict the Violence Index which describes the level of IPV against women in the country. The data records of 2494 ever-married women that have experienced IPV were considered from Sri Lanka Demographic and Health Survey 2016. The Violence Index was estimated using Multiple Correspondence Analysis. Gamma regression analysis revealed that religion, education level of the woman, husband’s occupation, woman’s married time, the age difference between husband and wife, Empowerment Indicator, enough money for daily household expenses, and household alcohol consumption were significantly associated with IPV against women. The optimum FNN consists of one hidden layer with 3 neurons provided a better prediction on the Violence Index with the minimum mean squared error for the testing set. Based on the prediction accuracy, the FNN was found to be better than the gamma regression model. The findings of this study would support an effort to develop the current policies and implement prevention programs against IPV in Sri Lanka.