Researching Domestic Violence Against Women: Methodological and Ethical Considerations

TLDR

The article compares three population‑based studies on violence against women in Nicaragua and investigates why their prevalence estimates differ using pooled multivariate logistic regression and focus‑group discussions. The comparison involves two regional studies (Leon and Managua) focused on women's experiences of violence and a nationally representative DHS, with data analyzed through pooled multivariate logistic regression and focus‑group discussions. Prevalence of partner‑perpetrated physical violence was 52 % in Leon, 69 % in Managua, and 28 % in the DHS, with differences largely attributable to ethical and safety procedures and interview settings, indicating that prevalence estimates are highly sensitive to methodology and that underreporting threatens validity.

Abstract

The results of three population‐based studies on violence against women in Nicaragua are compared in this article. Two of the studies were regional in scope (Leon and Managua) and focused specifically on women's experiences of violence, whereas the third study was a Demographic and Health Survey (DHS) conducted with a nationally representative sample of women. The lifetime prevalence estimates for women's undergoing physical violence from a partner were significantly higher in the Leon study (52 percent) and Managua study (69 percent), compared with that given in the DHS (28 percent). Possible explanations for the differences are examined through pooled multivariate logistic regression analysis, as well as analysis ofsix focus‐group discussions cairied out with field‐workers and staff from the three studies. The most important differences that were found concerned ethical and safety procedures and the interview setting. The results indicate that prevalence estimates for violence are highly sensitive to methodological factors, and that underreporting is a significant threat to validity.