Beyond the “Science of Sophomores”

• Author: Jeff Bouffard,Shamayne Smith,Rhonda Bry,Jeff Bry
• Published date: 01 December 2008
• Date: 01 December 2008
• DOI: http://doi.org/10.1177/0306624X07309721
• Subject Matter: Articles

International Journal of
Offender Therapy and
Comparative Criminology
Volume 52 Number 6
December 2008 698-721
Beyond the “Science of
© 2008 Sage Publications
10.1177/0306624X07309721
http://ijo.sagepub.com
Sophomores”
hosted at
http://online.sagepub.com
Does the Rational Choice Explanation
of Crime Generalize From University
Students to an Actual Offender Sample?
Jeff Bouffard
Washington State University, Pullman
Jeff Bry
Minnesota State Community and Technical College, Moorhead
Shamayne Smith
North Dakota State University, Fargo
Rhonda Bry
North Dakota Division of Juvenile Services, Fargo
Much of the criminological literature testing rational choice theory has utilized
hypothetical scenarios presented to university students. Although this research generally
supports rational choice theory, a common criticism is that conclusions from these studies
may not generalize to samples of actual offenders. This study proceeds to examine this
issue in two steps. First, a traditional sample of university students is examined to determine
how various costs and benefits relate to their hypothetical likelihood of offending. Then
the same data collection procedures are employed with a somewhat different sample of
younger, adjudicated, and institutionalized offenders to determine whether the conclusions
drawn from the student sample generalize to this offender sample. Results generally suggest
that the content and process of hypothetical criminal decision making differ in the sample
of known offenders relative to the university students. Limitations of the current study, as
well as suggestions for future research, are discussed.
Keywords:
rational choice; student samples; generalizability
Most criminological research from the rational choice perspective has examined
the effect of various crime costs and benefits on offending decisions, commonly
using one of three methodologies. First, ethnographic approaches have been employed
Authors’ Note: Please address correspondence to Jeff Bouffard, PhD, Criminal Justice Program,
Washington State University, PO Box 644880, Pullman, WA 99164; e-mail: jbouffard@wsu.edu.
698

---

Bouffard et al. / Beyond the “Science of Sophomores”
699
that involve interviewing known offenders about their past criminal decision-making
processes (Shover, 1996; Tunnell, 1992; Wright & Decker, 1994). This line of
research has discovered, for instance, that residential burglars attend to such factors
as the ease of entry to a house when considering where to commit their crimes
(Wright & Decker, 1994). Although this research confirms that offenders are responsive
to the potential risks and costs associated with crime, it has tended to suggest that
offenders focus on the potential rewards of crime. However, there are indications
from the body of ethnographic research that offenders sometimes ignore the potential
risks and costs, focusing instead on the immediate benefits of crime. Consider, for
example, Shover’s interviews (1996) with persistent thieves:
I wasn’t worried about getting caught or anything, you know. I was a positive thinker
through everything, you know. I didn’t have no negative thoughts whatsoever. (p. 156)
I didn’t think about nothing but what I was going to do when I got that money, how I
was going to spend it, what I was going to do with it, you know. See, you’re not thinking
about those things [possibility of being arrested]. You’re thinking about that big paycheck
at the end of thirty to forty-five minutes worth of work. (p. 159)
A second type of research design, employing longitudinal analyses, has been used
to examine perceptions of the certainty and severity of costs and benefits at one point
and then relate them to self-reported offending at a later date (e.g., Piliavin, Gartner,
Thornton, & Matsueda, 1986). This type of research has also found some support for the
contentions of rational choice theories of crime. For instance, Piliavin and colleagues
(1986), in one of the earliest and best-known longitudinal rational choice studies,
employed a sample of previously incarcerated adult offenders, drug users, and young
adults who had dropped out of school. The authors found little support for the deterrent
effect of either formal or informal sanction risks, but they did find some support for the
impact of financial benefits (and opportunities) on criminal offending.
Finally, a large number of rational choice studies have used either simple
hypothetical offending questions (e.g., “How likely is it that you would commit a
burglary?”) or more detailed hypothetical scenarios involving brief fictional depic-
tions of potential offending situations to elicit intentions to offend and thus relate
those intentions to the perception of various consequences (Bachman, Paternoster, &
Ward, 1992; Bouffard, 2002a, 2002b; Decker, Wright, & Logie, 1993; Exum, 2002;
Grasmick & Bursik, 1990; Klepper & Nagin, 1989; Nagin & Paternoster, 1993, 1994;
Paternoster & Simpson, 1996). Although many such studies have been conducted,
few of them have used samples of actual offenders, unlike the ethnographic and various
longitudinal studies previously described.
Strengths and Weaknesses of Hypothetical Scenarios
The detailed hypothetical scenario design has been commonly employed, relatively
speaking, in part because it avoids the problem of incorrect temporal ordering that has

---

700
International Journal of Offender Therapy and Comparative Criminology
plagued some longitudinal studies (Paternoster, Saltzman, Waldo, & Chiricos, 1983)
while providing more consistent context for the would-be offender’s decision than
that of the simple hypothetical offense question design, thus minimizing one source
of measurement error (Klepper & Nagin, 1989). Nevertheless, research using all
three designs has generally found at least some support for the rational choice
contention that perceptions of various costs and benefits (formal and informal
sanctions, perception of the act as immoral, and several indicators of rewards) influence
the decision to offend (Bachman et al., 1992; Bouffard, 2002a; Exum, 2002; Grasmick
& Bursik, 1990; Klepper & Nagin, 1989; Nagin & Paternoster, 1993; Paternoster &
Simpson, 1996; Piliavin et al., 1986; Wright & Decker, 1994). The inherent impact
of these rational choice concepts (costs and benefits) on offending has been some-
what unclear though, at least partly because of the different samples (rarely, actual
offenders) examined in these variously designed studies.
Despite its advantages over other measurement approaches, the hypothetical scenario
design continues to draw criticism. In particular, questions have been raised toward
whether those intentions to offend solicited through the use of hypothetical scenarios
accurately represent offending behavior. Pogarsky (2004) recently reviewed the
literature demonstrating the relationship between intentions to offend and both past
offending and future criminal behavior (see also, Azjen, 1991; Murray & Erickson,
1987). He extended this research by presenting results from his own work that
demonstrated a relationship between intentions to offend solicited from hypothetical
scenarios and a contemporaneous measure of deviance (cheating behavior) within
the study setting (Pogarsky, 2004). Thus, the appropriateness of using hypothetical
offending intentions in place of actual behavior seems to have been given a good
deal of empirical attention in the existing literature (although some will no doubt still
be unmoved by the research produced by Pogarsky and others).
A second criticism of research on rational choice and perceptual deterrence
theories—especially, those studies employing hypothetical scenarios—has not received
as much empirical attention, at least within the field of criminology. This criticism
centers on the generalizability of conclusions reached from the use of nonoffender
samples in hypothetical scenario studies. Specifically, this body of research has been
criticized (see Piliavin et al., 1986, as an example) for utilizing convenient but potentially
nongeneralizable samples of students (typically, high school or university students). In
fact, with the exception of the study of residential burglars by Decker and colleagues
(1993), little if any rational choice research employing hypothetical scenario designs
has been conducted using samples of known offenders, let alone comparing those
results to findings from more commonly examined student samples. The current
study thus begins the process of examining this commonly leveled criticism that
research using student samples cannot be generalized to actual offenders. This is
important, given that so little existing research has attempted to examine the merits
of this frequent criticism.

---

Bouffard et al. / Beyond the “Science of Sophomores”
701
Comparing Student Samples to Known Offenders
In one of the few studies attempting to address this issue, Decker and colleagues
(1993) found that members of their comparison sample of nonoffenders were
substantially less likely to report any hypothetical likelihood of burglary (even when
presented with relatively low levels of cost certainty) than were the known offenders
in their study. More important, they found that the groups responded differently to
the potential costs and benefits of offending. For instance, nonoffenders were found
to be largely unaffected by changes in the certainty or severity of costs or in the value
of the potential financial gain (i.e., they were not particularly responsive to either
costs or benefits). Burglars’ intentions to offend,...