UNPACKING THE BLACK BOX OF PEER SIMILARITY IN DEVIANCE: UNDERSTANDING THE MECHANISMS LINKING PERSONAL BEHAVIOR, PEER BEHAVIOR, AND PERCEPTIONS

• Author: JACOB T. N. YOUNG,CESAR J. REBELLON,J. C. BARNES,FRANK M. WEERMAN
• Date: 01 February 2014
• Published date: 01 February 2014
• DOI: http://doi.org/10.1111/1745-9125.12029

UNPACKING THE BLACK BOX OF PEER SIMILARITY

IN DEVIANCE: UNDERSTANDING THE MECHANISMS

LINKING PERSONAL BEHAVIOR, PEER BEHAVIOR,

AND PERCEPTIONS∗

JACOBT.N.YOUNG,

1CESAR J. REBELLON,2J. C. BARNES,3

and FRANK M. WEERMAN4

1School of Criminology and Criminal Justice, Arizona State University

2Department of Sociology, University of New Hampshire

3Criminology Program, University of Texas at Dallas

4Netherlands Institute for the Study of Crime and Law Enforcement

KEYWORDS: influence, deviant peers, social networks, theory testing, mechanism

The strong correlation between measures of personal and peer deviance occurs with

near “law-like” regularity. Yet, as with other manifestations of peer similarity (often

referred to as homophily), the mechanisms generating this relationship are widely de-

bated. Specific to the deviance literature, most studies have failed to examine, simul-

taneously, the degree to which similarity is the consequence of multiple causes. The

current study addresses this gap by using longitudinal network data for 1,151 indi-

viduals from the Netherlands Institute for the Study of Crime and Law Enforcement

(NSCR) School Project. Structural equation modeling is used to address these issues

by adapting Jussim and Osgood’s (1989) model of deviant attitudes in dyadic pairs to

the current data. Across two separate behavioral domains (substance use and property

offending), the results provide strong support for the prediction that individuals project

their own deviant tendencies inaccurately onto their peers. Conversely, the results pro-

vide little or no support for the predictions that respondents accurately perceive their

peers’ deviance or that their perceptions of peer deviance influence their own behavior.

Implications for understanding the role of peer behavior in the etiology of adolescent

deviance are discussed.

Relationships among similar individuals occur more frequently than relationships

among dissimilar individuals, a phenomenon often referred to as homophily (McPherson,

Smith-Lovin, and Cook, 2001). Considerable attention has been paid to this reality,

and social scientists have posited several explanations for why we observe homophily

(Lazersfeld and Merton, 1954) on demographic characteristics, attitudes, behaviors, and

psychological traits. Homophily might occur because actors seek out relationships with

those to whom they are similar or because environmental circumstances (e.g., Feld, 1982;

∗Additional supporting information can be found in the listing for this article in the Wiley Online

Library at http://onlinelibrary.wiley.com/doi/10.1111/crim.2011.52.issue-1/issuetoc.

Direct correspondence to Jacob T. N. Young, School of Criminology and Criminal Justice, Ari-

zona State University, 411 N. Central Ave, Suite 600, Phoenix, AZ 85004 (e-mail: jacob.young.1@

asu.edu).

C2013 American Society of Criminology doi: 10.1111/1745-9125.12029

CRIMINOLOGY Volume 52 Number 1 60–86 2014 60

UNPACKING THE BLACK BOX OF PEER SIMILARITY 61

Schaefer, 2012) affect multiple people similarly, even if those people do not choose to

associate with one another. For malleable attributes, similarity might reflect a process of

social influence, where individuals adopt the beliefs, values, or behaviors of those with

whom they affiliate (e.g., Akers, 2009). Overall, the variability in these explanations un-

derlines the fact that similarity among individuals in the same social network is likely the

consequence of multiple, complex processes.

The question of why individuals associate with similar others has been a focal concern

for the field of criminology. Nearly a quarter century ago, Agnew (1991: 47) emphasized

the empirical regularity of homophily with respect to deviance: “[T]he most consistent

finding in the literature on the causes of delinquency is that adolescents with delinquent

peers are more likely to be delinquent themselves.” Yet, research seeking to examine the

validity of various explanations of homophily has continued to generate debate. Recently,

a corpus of research centered on methodological concerns has produced two important

findings, creating additional complexity for an accurate interpretation of the observed

similarity in deviance.

First, measures of peer deviance derived from respondent perceptions of their peers’

behavior seem to be subject to false consensus (Marks and Miller, 1987) and projection

(Rebellon and Modecki, 2013) biases, whereby individuals impute their own behaviors

inaccurately to others (Young and Weerman, 2013; Young et al., 2011). By extension,

given that most studies of “peer influence” have not technically included peer self-reports,

most of the empirical literature has failed to demonstrate peer influence per se. Studies

that have employed peer-reported deviance measures have tended to suggest a much

weaker link between personal and peer deviance than traditional studies that have relied

on respondent perceptions of peer behavior (e.g., Prinstein and Wang, 2005; Weerman

and Smeenk, 2005). Insofar as perceptions of peer deviance remain important in their

own right (Akers, 2009), preliminary research has suggested that personal behavior might

be more responsible for perceptions than vice versa, likely reflecting false consensus and

projection bias (e.g., Matsueda and Anderson, 1998; Rebellon, 2012).

Second, recent studies using both dyadic (Rebellon and Modecki, 2013) and network

(Young et al., 2013) data have demonstrated that respondent perceptions of peer deviance

and peer-reported deviance reflect two fundamentally separate constructs. Most existing

research, however, does not view these measures as indicators of two fundamentally sepa-

rate constructs. Rather, these studies have tended to use either perceptual measures (e.g.,

Akers et al., 1979) or peer-reported measures (e.g., Haynie, 2001), thereby conflating

respondent perceptions of peer deviance and peer-reported deviance as alternative rep-

resentations of the same underlying theoretical construct. Yet the reliance on one type of

measure over the other might systematically alter the findings gleaned from any particu-

lar study. Moreover, the existing literature has yet to explore systematically the relation

of each measure to deviance simultaneously in the same overall model.

In sum, the proper interpretation of the correlation between personal and peer delin-

quency remains ambiguous and recent research has suggested multiple measurement is-

sues might account for part of the continuing confusion. The current study, therefore,

attempts to identify more precisely “the microfoundations ...through which the social

facts to be explained are brought about” (Hedstrom and Bearman, 2009: 9). In particular,

it attempts to explicate the “cogs and wheels” (Elster, 1989) that account for the corre-

lations among personal deviance, perceived peer deviance, and peer-reported deviance.

We build on the work of Jussim and Osgood (1989), who developed and tested a model