Perceived arrest risk, psychic rewards, and offense specialization: A partial test of rational choice theory†

• Author: Kyle J. Thomas,Benjamin C. Hamilton,Thomas A. Loughran
• Date: 01 August 2020
• Published date: 01 August 2020
• DOI: http://doi.org/10.1111/1745-9125.12243

Received: 26 June 2019 Revised: 27 February 2020 Accepted: 27 February 2020

DOI: 10.1111/1745-9125.12243

ARTICLE

Perceived arrest risk, psychic rewards, and offense

specialization: A partial test of rational choice

theory*

Kyle J. Thomas1Thomas A. Loughran2Benjamin C. Hamilton3

1University of Colorado—Boulder

2The Pennsylvania State University

3University of Missouri—St. Louis

Correspondence

KyleJ. Thomas, Department of Sociology,

Universityof Colorado Boulder Ketchum Arts

andSciences, Boulder, CO 80309.

Email:kyle.thomas@colorado.edu

∗Additionalsupporting information

canbe found in the full text tab for this

article in the WileyOnline Librar y at

http://onlinelibrary.wiley.com/doi/10.1111/

crim.2020.58.issue-3/issuetoc.

Wewould like to thank Dan Nagin, Holly

Nguyen,the anonymous reviewers and, the

Editor,David McDowall for their helpful feed-

backon earlier versions of this article.

Abstract

In prior tests of Beckerian rational choice theory, the notion

that individuals are responsive to the (dis)incentives associ-

ated with crime has been supported. Much of this research

has comprised composite scores of perceived rewards and

risks of multiple, often disparate, crime types that are then

used to predict “general” offending behavior. Although the

results of such prior tests are informative, we believe that

this tendency has resulted in two shortcomings. First, a

central component of mathematical rational choice the-

ory is overlooked, namely, that responsivity to incentives

will be crime specific. That is, offenders should prefer

crime types that subjectively offer greater rewards and

fewer risks relative to other crimes. Second, individual dif-

ferences in offending specialization are not addressed, of

which Clarke and Cornish (1985) and Shover(1996) argued

rational choice theories are well suited to explain. Using a

sample of serious offenders, we find that in a given time

period, individuals are more likely to engage in crime types

they viewed as more intrinsically rewarding and less risky

compared with other crimes. Furthermore, individuals dis-

played greater specialization in violence to the extent they

view violence as more rewarding and less risky than prop-

erty offenses

KEYWORDS

offender decision-making, rational choice theory, specialization

Criminology. 2020;58:485–509. wileyonlinelibrary.com/journal/crim © 2020 American Society of Criminology 485

486 THOMAS ET AL.

The rational choice perspective in criminology differs from other traditions in its explicit focus on

the decision-making processes underlying offending (Becker,1968; McCart hy, 2002; see also Hechter

& Kanazawa, 1997). Although numerous variations of rational choice theories exist (Becker, 1968;

Cornish & Clarke, 1986; McCarthy, 2002), these perspectives all share the common assumption that

individuals are responsive to incentives. Most quantitativetests have consisted of the formal mathemat-

ical interpretation grounded in the traditional expected utility (von Neumann & Morgenstern, 1945)

model originally proposed by Becker (1968), including constituent elements suchas Bayesian updating

(Anwar & Loughran, 2011; Matsueda, Kreager, & Huizinga, 2006), and have included a regression of

self-reported offending on rational choice inputs to examine whether the risks and rewards associated

with crime predict offending (e.g., Loughran, Paternoster, Chalfin, & Wilson, 2016). The findings

from extant research show support for rational choice predictions, suggesting that potential offend-

ers consider the consequences associated with crime when deciding whether to offend (see Loughran

et al., 2016; Matsueda et al., 2006; Nagin, 2007; Paternoster, 2010). Much of this research has utilized

composite scores of risk and rewards to predict variation in a general offending outcome (Paternoster,

1986; Piliavin, Gartner, Thornton, & Matsueda, 1986). That is, key constructs like perceived arrest

risk and psychic rewards are typically measured as an aggregation of subjective perceptions across

multiple, often disparate crime types,1which are then linked to aggregate offending. The implication

is that monolithic measurements of “risk” and “rewards” are general, rather than crime-specific factors

in offending decisions. As Shover (1996, p. 65) noted, however, “the existence of distinct crime pref-

erences …[tends] to draw [individuals] back to …the type of crime they feel most at home with.” In

other words, individuals hold crime-specific perceptions and preferences that make them more likely

to commit certain types of crime relative to others.2

The findings from prior research has shown that many individuals view some crimes as thrilling

and exciting but see other types of crime as eliciting little in terms of intrinsic rewards (Katz, 1988;

Shover, 1996; Steffensmeier, 1986; Steffensmeier & Ulmer, 2005; Topalli, 2006; Wright & Decker,

1997). Furthermore, Anwar and Loughran (2011) found that perceived arrest risk was updated in a

nuanced manner within persons across different types of crime and highlighted that individuals tend

to view some crimes as less risky than others (see also Thomas, Hamilton, & Loughran, 2018). To

the extent that individuals are responsive to the risks and rewards associated with offending, scholars

drawing on the rational choice perspective have explicitly suggested that subjective expectations of

different crimes should vary within persons (Nagin, 1998) and, thus, affect the qualitative nature of

individuals’ offending by allowing fordifferent choice calculi for different crimes (i.e., an individual’s

tendency to commit specific crimes over others; Anwar & Loughran, 2011; Cornish & Clarke, 1985;

Matsueda et al., 2006; Shover, 1996). Todate, however, little research has been conducted to examine

this prediction directly (Nagin, 1998).

The notion that the types of crimes3individuals commit may be a function of within-person variation

in the risks and rewards associated with crime has several implications. First, when keeping perceived

1For example, it is common for authors to report a Cronbach’s alpha combining all crimes for these types of measures (see

https://www.pathwaysstudy.pitt.edu/codebook/personal-and-social-costs-and-rewards-sb.html).

2Others have openly questioned the assumption that perceptions are global as opposed to crime specific (Nagin, 1998).

3We use the terms “crime-specific,” “crime types,” and “types of crime” to reflect both specific criminal acts (e.g., fighting

and theft) as well as broader offense categories such as violent and property crimes. Both can be contrasted to a general overall

offendingmeasure. Alt hough at times this mayseem confusing, there is no consensus in t he field as to the appropriate granularity

when assessing qualitative dimensions of offending(Mazerolle, Brame, Paternoster, Piquero, & Dean, 2000; Sullivan, McGloin,

Pratt, & Piquero, 2006). Even rational choicet heorists havediscussed both specific acts and broader crime-type categor ies when

considering crime-specific effects, and both outcome operationalizations are consistent with a rational choice approach.