This article examines the sources of municipal variation in the number of civil suits filed against police. Although a number of theoretical explanations are plausible, existing data and research are limited to a handful of rational/ technical explanations. A secondary analysis of a national survey data set reveals a moderate but manageable problem with missing data among the predictors. However, the problem is more severe for the outcome variable, the ratio of civil suits per 1,000 field officers, which is missing for about 70% of respondents. The authors explore a number of potential remedies based on recent advances in the statistical treatment of incomplete multivariate data. Their analysis suggests that data on civil suits are not missing randomly: There are important and quantifiable differences between respondents and nonrespondents to this survey item. These findings suggest a modest case of sample selection bias, which cannot be corrected using even the most recent statistical advances. The remedy lies not in statistics but in the use of new and creative research methods that account for the sensitive nature of the topic.
ASJC Scopus subject areas
- Social Sciences (miscellaneous)