Aggregation Bias and Woman Abuse

Many researchers have been attracted to broad, national-level surveys as an antidote to the more usual practice of studying woman abuse in one location or campus and presuming that the results generalize to the entire population. However, the reverse error is also possible: presuming that one national rate may adequately represent a variety of different regions, types of schools, and cultural groups. This article analyzes the Canadian National Survey data to compare geographic regions, types of schools, and whether the students took the survey in French or English. None of these factors influenced the results. Male peer support measures, as hypothesized, did strongly affect male behavior in both physical and sexual abuse.