Testing a rational choice model of “desistance:” Decomposing changing expectations and changing utilities

• Author: Matt Vogel,Kyle J. Thomas
• Published date: 01 November 2019
• DOI: http://doi.org/10.1111/1745-9125.12223
• Date: 01 November 2019

Received: 16 April 2018 Revised: 17 May 2019 Accepted: 19 June 2019

DOI: 10.1111/1745-9125.12223

ARTICLE

Testing a rational choice model of “desistance:”

Decomposing changing expectations and changing

utilities∗

Kyle J. Thomas1Matt Vogel2,3

1University of Colorado Boulder

2University at Albany,SUNY

3TU Delft

Correspondence

KyleJ. Thomas, Department of Sociology,

Universityof Colorado Boulder UCB 327

Ketchum195, Boulder, CO 80309.

Email:kyle.thomas@colorado.edu

∗Wewould like to thank Tom Loughran, Lee

Slocum,Jody Miller, and the anonymous

reviewersfor their helpful comments on ear-

lierversions of this ar ticle.

Abstract

We argue that a rational choice framework can be used

to explain declines in offending from adolescence to

young adulthood in two ways. First, subjective expecta-

tions of offending can be age graded such that percep-

tions of rewards decrease and perceptions of risks and costs

increase. Second, the marginal (dis)utility of crime may be

age graded (e.g., preferences for risks, costs, and rewards).

We examine changes in offending from adolescence to

young adulthood among a subset of individuals from the

Pathways to Desistance Study (N =585) and employ a

nonlinear decomposition model to partition differences in

offending attributable to changing subjective expectations

(X) and changing marginal utilities (𝛽). The results indicate

that both have direct and independent effects on changes

in offending over time. The results of a detailed decom-

position on the subjective expectations also indicate that

differences exist across the type of incentives. That is,

the effect of changing expectations is attributed mainly to

changes in perceived rewards (both social and intrinsic).

Changing expectations of social costs and risk of arrest

from offending have weak effects on changes in criminal

behavior, which suggests that they must be accompanied

by increases in the weight placed on these expectations to

promote appreciable declines in offending.

KEYWORDS

decomposition, life course, rational choice

Criminology. 2019;57:687–714. wileyonlinelibrary.com/journal/crim © 2019 American Society of Criminology 687

688 THOMAS AND VOGEL

That criminal activity varies over the life course is an established fact in the social sciences (Hirschi

& Gottfredson, 1983; Steffensmeier, Allan, Harer, & Streifel, 1989). Of particular note is the seem-

ingly sharp decline in offending from adolescence to early adulthood. Numerous theories have been

proffered to explain such changes in offending (Agnew, 1997; Akers, 1998; Giordano, Cernkovich,

& Rudolph, 2002; Moffitt, 1993; Sampson & Laub, 1993). In much of this literature, scholars have

emphasized the importance of life-course transitions, identity transformation, and the role of emotions

in the desistance process (Giordano et al., 2002; Giordano, Schroeder, & Cernkovich, 2007; Laub &

Sampson, 2003; Maruna, 2001; Sampson & Laub, 1993; Warr, 1998). Although most of these per-

spectives incorporate an element of “choice,” it is usually treated as a narrow by-product of “human

agency,”invoked to explain residual fluctuations in offending trajectories not captured by keyprocesses

proposed by a given theory (Paternoster, 2017). Mostly absent from this literature is a formal treatment

of the decision-making processes that underpin developmental differences in offending across the life

course (cf. Gartner & Piliavin, 1988; Paternoster & Bushway, 2009; Steinberg, 2008).

Rational choice theorists traditionally view offending decisions as a function of both subjective

expectations and marginal utility, with the former representing the perceptions of rewards, costs, and

risks that go into the decision calculus and the latter reflecting how those expectations are weighted

(Watkins, 1915).1In drawing on this distinction, differences in offending between adolescence and

young adulthood can be explained through a decision-making perspective in multiple ways. First, sub-

jective expectations about the risks, costs, and rewards of offending may be age graded. A common

theme in theories of desistance is that the perceived risks and costs of crime increase as individuals

enter adulthood (Sampson & Laub, 1993). Less theoretical attention has been focused on how rewards

from crime vary over time; however, there are reasons to suspect that such incentives may change in

ways that promote declines in offending (Giordano, Cernkovich, & Holland, 2003). From this view,

the weight placed on rational inputs may be similar over time but individuals perceive the benefits of

crime as greater than the costs in adolescence relative to adulthood (Becker, 1968). A second possibil-

ity is that individuals weigh the inputs differently over time—that is, the marginal (dis)utility of crime

is age graded (Paternoster & Bushway, 2009). Adolescents may be particularly concerned about the

social and intrinsic rewards from crime but care less about such incentives byt he time theyreach young

adulthood (i.e., the marginal utility of offending decreases with age; see Steinberg & Monahan, 2007).

Similarly, in adulthood individuals may be more risk averse or increasingly sensitive to the negative

social costs of crime (e.g., loss of job). In this way, it may be the changes in the weights placed on

arrest risk and social costs of offending—rather than the changes in the perceptions of these inputs—

that explains declines in crime in young adulthood (i.e., the marginal disutility of offending increases

with age). These two processes can operate in tandem, with changes in both contributing to the decline

in criminal behavior from adolescence to young adulthood.

Although changes in subjective expectations and changesin marginal utility are both consistent with

a decision-making framework, determining the extent to which reductions in offending from adoles-

cence to early adulthood are a result of changes in subjective expectations or the utility derived from

these expectations is of theoretical importance. A fundamental disagreement among prominent desis-

tance theories concerns the extent to which declines in offending are a result of changes in perceptions

of risks, costs, and rewards from crime (Akers, 1998; Sampson & Laub, 1993), changes in preferences

that affect the weight that individuals place on these (dis)incentives (Paternoster & Bushway, 2009), or

some combination of both (Giordano et al., 2002). Indeed, several scholars have highlighted the impor-

tance of disentangling the relative contributions of changing perceptions and changing preferences for

1Marginal utility is more formally defined as 𝜕𝑈∕𝜕𝑥 or the change in expected utility (U) given changes in some subjective

input x.