Panel Methods in Criminology

by

Julie Phillips

• LAST REVIEWED: 28 May 2013
• LAST MODIFIED: 28 May 2013
• DOI: 10.1093/obo/9780195396607-0142

Introduction

Panel data, sometimes called repeated measures or multilevel data, have a hierarchical or nested structure. Multiple observations gathered over a series of time periods are nested or clustered within units. The units may be individuals or may be geographic in nature (e.g., states or cities). In recent years, panel data have become increasingly available, software capabilities have expanded, and criminologists use such data sets to answer a variety of questions. For example, researchers use panel data to understand factors that affect time trends in crime across states or cities. Criminal career researchers and developmental criminologists use panel data on individuals to identify and explain individual trajectories of offending over time. Others have explored the factors that predict the time to a criminal event, such as a first offense, (re)arrest, or incarceration. Panel data have a number of advantages over cross-sectional or time-series information alone. They are more informative as they contain greater variability and more degrees of freedom, so that regression models can produce more efficient estimates. They enable researchers to sort through questions of causal ordering because they include information on time-ordering of events. Finally, models using panel data are able to control for unobserved heterogeneity (omitted variable bias). This article offers an overview of the extensive literature in criminology on panel data analysis. After a brief overview of the methodological advantages to longitudinal data, the article reviews four major analytical approaches that rely on panel data: cross-sectional time-series analysis, growth trajectory models, group-based trajectory models, and event history analysis. For each methodological approach, works are selected that provide overviews and introductions to the method, contain more detailed information on the method in textbook form, and showcase applications in criminology. The article concludes with some general resources (software and writing) for panel data analysis.

Advantages and Disadvantages to Longitudinal Data

Panel data, also known as longitudinal data, are characterized by multiple measurements over time, and, thus, they present both advantages and disadvantages for the researcher. Davies 1994 highlights the strengths of longitudinal data compared to cross-sectional data. Deschenes 1990 provides an overview of the issues criminologists should consider when deciding on a research design and also touches on the historical debate within criminology regarding the necessity of longitudinal data. For a useful overview on the kinds of longitudinal data sets and models used by criminologists studying delinquency as well a review of the key findings, readers will find Thornberry and Krohn 2003 valuable.

• Davies, Richard B. 1994. From cross-sectional to longitudinal analysis. In Analyzing social and political change: A casebook of methods. Edited by Angela Dale and Richard B. Davies, 2–40. Thousand Oaks, CA: SAGE.

  Good overview discussion of the advantages of longitudinal data analysis over cross-sectional methods.

• Deschenes, Elizabeth P. 1990. Longitudinal research designs. In Measurement issues in criminology. Edited by Kimberly L. Kempf, 152–166. New York: Springer-Verlag.

  DOI: 10.1007/978-1-4613-9009-1

  Discusses some of the major issues researchers should consider when deciding on a research design. Strengths and weaknesses of longitudinal research designs are examined, and the history of the debate regarding the use of longitudinal research in criminology is reviewed.

• Thornberry, Terence P., and Marvin D. Krohn, eds. 2003. Taking stock of delinquency: An overview of findings from contemporary longitudinal studies. New York: Kluwer Academic-Plenum.

  Comprehensive synthesis of findings of seven important ongoing longitudinal studies of delinquency. Examines the extent to which these studies answer questions concerning the origins of delinquent and criminal careers. Highlights the varying guiding theories and methods used to analyze the data. The first chapter describes the development of panel studies in delinquency.