The Implications of Using Group-Based Offenses versus Non-Group-Based Offenses in Peer Deviance Scales

ABSTRACT

A key distinction from Warr’s research is “groupy” (deviant acts commonly committed with peers) versus “non-groupy” (acts that are likely committed alone) deviance. Drawing on Warr’s important (and underutilized) distinction, this study’s goal is to determine whether measures of perceptual groupy peer deviance more accurately reflect a peer’s true deviance than perceptions of nongroupy peer deviance. Using dyadic data, results demonstrate that perceptions of a peer’s non-groupy deviance are largely inaccurate but perceptions of a peer’s groupy deviance are quite accurate. Despite this discrepancy, groupy and non-groupy perceptual measures function similarly in multivariate models and consistently outperform peer self-reports.

Acknowledgments

We cordially thank Scott Culhane, Adrienne Freng, Eric Wodahl, and our anonymous reviewers for providing useful comments on an earlier version of this article.

Notes

¹ Realistically, research demonstrates that nearly all offenses are characterized by both solo-person offending and offending with others present (e.g., Reiss 1986). Our use of these terms is simply to distinguish that some deviant and criminal actions frequently occur in the presence of peers and other actions frequently occur without peers.

² Depending on how deviance is captured in any particular dataset, an additional possibility to what could be classified as groupy deviance arises. Methodologically, it may appear that groupy deviance has occurred if friends offended without each other present but in similar patterns. That is, Persons A and B committed the same type of offense, but at different times and in different places. While it may seem logical that these perceptions could be of low accuracy since neither A nor B witnessed the other’s behavior, there is substantial evidence to suggest that people project their own deviant (or nondeviant) behavior into the perception of their friend’s behavior (e.g., Boman and Ward 2014). In the hypothetical case of Persons A and B, the projection of Person A’s behavior into his/her perception of B’s behavior—despite it being a guess—would actually be accurate, leading one to expect that groupy deviance could potentially be a very accurately perceived construct regardless of the friends’ behavioral similarity.

³ Although a trivial amount of missing data were observed (less than 1% missing on all variables), we chose to impute missing values using to a Markov-Chain Monte Carlo (MCMC) procedure with 1,000 random draws. Using available information and a maximum likelihood procedure, the MCMC procedure iteratively generates different datasets with the goal of finding the most likely missing data value.

⁴ The average correlations for the individual scale items follows: perceptual groupy and actor’s self-reported groupy = .577; perceptual groupy and peer’s self-reported groupy = .286; perceptual non-groupy and actor’s self-reported non-groupy = .382; perceptual non-groupy and peer self-reported non-groupy = .089.

⁵ Interestingly, actor perceptions of a peer’s groupy deviance are slightly more strongly related to the peer’s non-groupy deviance (r = .21, p ≤ .001) than perceptions of non-groupy deviance are related to the peer’s non-groupy deviance (r = .19, p ≤ .001). This could speak to the possibility that indirect measures of groupy peer deviance could outperform indirect measures of non-groupy deviance regardless of what behaviors the peer is actually committing. We return to this issue with a series of supplementary analyses at the end of the Results section.

⁶ Because of the MCMC imputation’s iterative procedure, Stata produces an F-statistic in mixed models instead of a χ² statistic.

⁷ Using the coefficient and standard error comparison test developed by Clogg, Petkova, and Haritou (1995) and used by Paternoster and colleagues (1998), we compared the coefficient strength of the indirect and direct groupy peer deviance measures to crime. Specifically, we compared models one and two, models three and four, and models five and six. Without exception, the indirect groupy measures were consistently related to crime more strongly than the direct groupy measures (all z-scores > 10, all p’s ≤ .001).

⁸ Again using coefficient and standard error comparison tests, we compared the strength of the indirect and direct non-groupy measures of peer deviance between models one and two, three and four, and five and six. The three comparisons all illustrated that the perceptual measures are much more strongly related to crime outcomes than the peer self-reports (all z-scores > 16, all p’s ≤ .001).

⁹ Our results also extend to gender, one of the strongest correlates of crime (e.g., Steffensmeier and Allan 1996; see LaGrange and Silverman 1999). In particular, gender was considerably stronger in models controlling direct groupy peer deviance as opposed to direct non-groupy peer deviance. Considering the pragmatic importance of gender to criminologists, reasons for this finding must be explored.

Additional information

Notes on contributors

John H. Boman

JOHN H. BOMAN IV is an Assistant Professor in the Department of Criminal Justice at the University of Wyoming. His research is primarily theoretical and focuses on the nature of peer relationships and friendships across the life-course, measurement, validity and construct validation, and perceptual processes that may influence crime. Some of his recent work appears in the Journal of Quantitative Criminology, Criminal Justice and Behavior, the Journal of Criminal Justice, and Crime and Delinquency.

Chris L. Gibson

CHRIS L. GIBSON is a Research Foundation Professor and Associate Professor of criminology in the Department of Sociology and Criminology & Law at the University of Florida. He is also a faculty affiliate in the Institute for Child Health Policy, College of Medicine at the University of Florida. He was a recipient of the W.E.B. Du Bois Fellow for the National Institute of Justice and the 2013 recipient of the Tory Caeti award from the Academy of Criminal Justice Sciences. His book (w/Marv Krohn), titled Handbook of Life-Course Criminology: Emerging Trends and Directions for Future Research, was published in 2013 by Springer-Verlag. He teaches graduate courses on crime over the life course and quantitative methods.