Psychological abuse within intimate relationships is linked to negative health outcomes among women and is frequently identified as more wounding than physical or sexual violence. There is little agreement, however, on how to conceptualize or measure the phenomenon, despite measurement being necessary to estimate the prevalence of psychological abuse, establish its interaction with physical and sexual violence, assess its health impacts, and monitor progress towards global Sustainable Development Goals. To address this gap, we used latent class analysis (LCA), psychometric testing, and logistic regression to evaluate the construct and content validity of alternative methods for deriving a measure of psychological partner abuse, using pooled data from the first 10 countries and 15 sites of the World Health Organization Multi-Country Study on Domestic Violence and Women's Health (WHO MCS). Our analysis finds that psychological abuse is highly prevalent, ranging from 12% to 58% across countries. A three-class solution was supported for coding psychological abuse: none, moderate-intensity abuse, and high-intensity abuse. This three-level categorization, which can be coded without LCA, demonstrates a clear graded relationship with controlling behaviors and all measured health outcomes except physical pain. Factor analysis confirms that psychological abuse and male controlling behaviors are separate constructs as measured in the WHO MCS and the Demographic and Health Surveys and should not be combined. We conclude that this is a simple way to code psychological abuse for cross-country comparison. Its use could support urgently needed research into psychological abuse across settings and identify an appropriate threshold for defining psychological violence for surveys globally.
SSM - Population Health, Volume 9, December 2019.,