Dude, Who’s in My Cohort? Challenges and Solutions for Measuring Cohort Homicide from Period Data

• In Event: Issues and Advances in Measurement and Design

Abstract

Criminologists have long studied the role of age, period, and cohort in shaping homicide trends using a variety of modeling strategies and theoretical motivations. Despite this robust literature, fundamental measurement challenges remain pervasive in estimating age-specific homicide rates for birth cohorts using available period data. In this paper, we first demonstrate substantial differences in the observed trends in homicide rates between alternative approaches of measuring cohorts with period data. These discrepancies are particularly large when using categorical age and multi-year birth cohorts. Substantively, these discrepancies lead to different conclusions when and for which cohorts homicide rates peaked in the United States. Most troublingly, each of these alternative approaches to measuring the homicide rate for cohorts is equally valid theoretically and authors rarely specify which approach they are using. We provide a mathematical model for understanding why these discrepancies occur and, most importantly, propose multiple practical solutions depending on data availability. Lastly, we demonstrate that these measurement challenges can lead to different substantive conclusions in theoretically driven models predicting cohort trends in homicide. Overall, accurately measuring cohort homicide is an essential, yet often overlooked, part of understanding homicide trends.

Authors

• Jason Robey, University at Albany, SUNY

• Matt Vogel, University at Albany, SUNY