Deterrence: Testing the Effects of Perceived Sanction

Certainty on Probation Violations

Sheila Roy0 Maxwell, Michigan State University

M. Kevin Gray, Michigan State University

Deterrence theory has been repeatedly used to explain individuals’ choices to com-
mit vs. not commit crimes. Its utilitarian premise is that individuals will choose to com-
mit or not commit crimes based on the certainty, swiftness, and severity of punishments.
This study examines the perceived certainty of sanctions among offenders mandated into
intensive probation and the effects of varying levels of perceived sanction certainty in
deterring offenders from violating probation requirements. Data from an Intensive
Supervision Probation program in New Jersey are used. Results support the salience of
offenders’ perceptions of sanction certainty on their intensive probation outcomes, and
implications for policy are discussed.

Deterministic theories about law-abiding and lawbreaking behaviors have

dominated explanations of crime and deviance for much of the first half of
the twentieth century (Gottfredson and Hirschi 1990; Maxwell 1997). These in-
cluded explanations such as Lombroso-Ferrero’s Criminal Man (191 l), Merton’s
Strain (1938), Sutherland’s Differential Association (1947), or Cloward and
Ohlin’s Diflerential Opportunity (1960). Generally, these theories emphasized
the development of criminal motivations through socialization within deviant
subcultures, the failure of conventional socialization processes, psycho-dynamic
problems, or external criminogenic drives generated by the social milieu (Maxwell
1997; Tittle 1980a; Tittle 1980b).
Within the last thirty years, however, these explanations have proven limited
in explaining and preventing crime (Andenaes 1975; Glaser 1979). Particularly
in the 1960s, rising crime rates, increasing drug use and general disorder led re-
searchers who were interested in crime control to redirect their attention from
searching for the seemingly elusive “causes” of crime to examining instead the
utility of formal and informal sanctions in controlling crime (Maxwell 1997;
Tittle 1980b). This paradigmatic shift was also fueled by well-publicized results
of a study that challenged the utility of expensive rehabilitation and treatment
programs for offenders, claiming these were ineffective in reducing crime rates
(i.e., the famous “Nothing Works” by Lipton, Martinson, and Wilks 1975;
Martinson 1974). The result of this study was an affront to then popular policies
of preventing crime or managing offenders through treatment, policies that were

Socio/ogicul Inquiy, Vol. 70, No. 2, Spring 2000, 117-36

02000 by the University of Texas Press, P.O. Box 78 19, Austin, TX 787 13-78 19
1 18 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

direct legacies of the deterministic paradigms of offending behaviors (Maxwell

1997; Palmer 1992).
The renewed focus on using threats of sanctions to control behavior derives
from the age-old utilitarian philosophy of the Classical School of Criminology
and underlies deterrence theory. The Classical School views human beings as ra-
tional individuals who efficiently weigh the costs and benefits of their actions in
attaining desired goals (Akers 1994; Gottfredson and Hirschi 1990; Zimring and
Hawkins 1973). It posits that “actions are taken and decisions are made by per-
sons in the rational exercise of free will” (Akers 1994). As a derivative of the
Classical School, deterrence theory postulates that individuals choose to obey or
violate the law by making rational calculations about the potential costs and
benefits of such acts (Bridges and Stone 1986). The deterrence model argues
that once legal sanctions are applied, individual rule-breaking will desist among
those directly sanctioned (specific deterrence) and among those who were only
aware of the sanctions (general deterrence). Individuals will be deterred from
committing crimes because they dislike the intrinsic or extrinsic consequences
associated with the sanctions or because they fear the potential applications of
these costly sanctions on their fiture misconduct (Maxwell 1997).
The Relevance of the Deterrence Philosophy in Crime Policies
Deterrence is the theoretical principle upon which many aspects of the jus-
tice system are grounded. For example, Mueller (1977) argued that the sole pur-
pose of punishments levied by criminal laws is to deter future crimes. Similarly,
Zimring and Hawkins (1973) and Morris (1966) asserted that the deterrent and
preventive elements of criminal law are its primary and most important postulates.
Currently, the deterrence doctrine dominates criminal justice philosophy
and practice (Maxwell 1996; Sabol and Lynch 1997; Tonry 1995; Wichiraya
1996). This is especially seen in augmented prison terms for many types of
offenses (e.g., drug offenses, offenses committed with firearms), in mandatory
sentences or enhanced penalties (Maxwell 1995; Sabol and Lynch 1997), in in-
creasing numbers of juveniles waived to adult courts, and in “three strikes” pro-
grams (Maxwell 1996). In addition, the justice system in the United States has
increased the likelihood of detection of law violators. Policies ranging from more
police officers on the streets to electronic monitoring of offenders in the com-
munity (Byrne and Pattavina 1992) and more intensive probation programs
(Petersilia, Turner, and Deschenes 1992b) indicate to potential violators that their
behaviors will not go unchecked. Contrary to the rehabilitative approaches of ad-
dressing crimes that were prominent in the 1960s, these policies are premised on
deterrence and the belief that increasing punishment will make crime commis-
sion less profitable, thus decreasing crime rates. However, evidence of the deter-
rent effects of sanctions have been meager (Blumstein, Cohen, and Nagin 1976;
 PROBATION VIOLATIONS 119

Piliavin, Thorton, Gartner, and Matsueda 1986; Paternoster 1987).Austin, Jones,

and Kramer (1994), for example, recently examined crime trends in four states
that had sentencing enhancements and found these trends to be similar to na-
tional crime trends and to those states without sentencing enhancement laws.
Similarly, Wichiraya (1996) examined thirty-one states that have implemented
sentencing enhancement laws but found no significant decreases in crime rates
in most of these states and even noted increases in some. Moreover, Petersilia et
al. (1992b) compared the re-offense rates of offenders given intensive probation
and those given regular probation, and found higher rates of re-offense among
the former. Several other researchers have examined the effects of one or more
recent punitive trends on crime rates and on recidivism rates of offenders, and
found no substantial effects of enhanced penalties on crime rates and recidivism
(Tonry 1996). Others found that enhanced penalties were related to increased re-
cidivism (see Fagan 1994; MacKenzie 1994).
Certainty of Punishments and Deterrence
Driven by many research results that have failed to support the deterrent
elements of sanctions, scholars in the last few years have re-examined and refined
the measures of deterrence (Paternoster 1987; Williams and Hawkins 1986) to
include not only the severity of the punishment, which refers to the extent of un-
pleasant consequences resulting from a criminal sanction, but also the celerity or
swiftness of punishment and the certainty of punishment (Erickson, Gibbs, and
Jensen 1977). Celerity refers to the proximity of the dispensation of a sanction
following the commission of a crime (Clark 1988), and certainty refers to the
probability that a punishment will ensue following commission of a crime
(Gibbs 1975; Meier 1978; Zimring and Hawkins 1973). Several studies have
examined each of these dimensions (see, for example, Paternoster 1987;
Paternoster, Saltzman, Waldo, and Chiricos 1983; Schneider and Ervin 1990;
Williams and Hawkins 1986) and found that compared to the other dimensions
of deterrence, Certainty of punishment is a more forceful factor in the calculation
of costs and benefits associated with criminal acts. Grasmick and Bryjack (1980)
and Erickson et al. (1977), for example, found more clear and consistent findings
across studies that examined the certainty of sanctions in deterring crimes.
Likewise, Beccaria (1963) noted that the certainty of punishment will always
make a stronger impression than the fear of more severe sanctions that is com-
bined with the hope of impunity. Singer (1970) and Gottfredson and Hirschi
(1990) further contended that offenders are often “presently oriented” such that
their behaviors are more likely affected by consequences of crimes that are im-
mediate and certain rather than severe but uncertain. In their exhaustive review
of deterrence research, Blumstein, Cohen and Nagin (1978) also contended that
deterrence first depends on the perceived certainty of punishments and then on
120 SHEILA ROY0 MAXWELL AND M.KEVIN GRAY

consideration of the severity of punishments. They further contended that the

lack of evidence supporting the effects of punishment severity on deterrence may
be due to the lack of attention to the certainty dimension. That is, severity may
matter but “only when persons believe there is a high probability that they will
receive that punishment” (1978). It is clear from previous research that certainty
factors play a central role in deterrence and warrants a closer examination. This
paper focuses specifically on the role that certainty plays in deterring probation
violations.
Testing the Certainty Dimension of Deterrence
The certainty dimension comprises two related probabilities: the objective
probability of the sanction application and the perceived probability of the sanc-
tion application (Ekland-Olson and Kelly 1993; Gibbs 1982). The first series of
studies that examined the role of certainty in deterring crime have relied exclu-
sively on the objective measures of certainty, such as number of arrests and num-
ber of convictions. Certainty was then calculated by dividing the number of
arrests or convictions by the number of reported crimes (Antunes and Hunt 1978;
Gibbs 1982). The effectiveness of deterrence was determined based on how
these aggregate numbers negatively related to crime rates across geographical
areas (Grasmick and Bryjack 1980; Klepper and Nagin 1989; Piliavin et al.
1986). Though informative, these studies were limited when predicting indi-
vidual behaviors due to the aggregate measures that were used. To address this
limitation, the next wave of studies utilized surveys of individuals, first cross-
sectionally and then later longitudinally, asking individuals about their percep-
tions of punishment certainty and their past or future criminal activities (Klepper
and Nagin 1989; Paternoster 1987). These studies then correlated individuals’
perceptions of punishment certainty with their self-reported criminal activities
(Klepper and Nagin 1989; Paternoster 1987; Paternoster et al. 1983; Schneider
and Ervin 1990; Williams and Hawkins 1986). Individual-level analysis was
thought more profitable in analyzing perceptions of individuals about sanction
certainty because it permitted a more direct examination of the sociological and
the psychological correlates of these perceptions (Cohen 1978).
Results of cross-sectional studies showed negative associations between
perceptions of sanction certainty and criminal behavior (Paternoster et al. 1983;
Schneider and Ervin 1990), lending support for the effect of perceived certainty
on criminal behavior. However, these cross-sectional studies were also limited in
at least one respect-their inability to sequentially order the measurements of
perceived certainty and individual criminal behaviors (Piliavin et al. 1986). In
other words, respondents were asked to give current estimates of certainty on
their involvement in prior criminal activities (Paternoster 1987; Paternoster,
Saltzman, Waldo, and Chiricos 1982a). Paternoster (1987) noted that this tem-
 PROBATION VIOLATIONS 12 1

poral difficulty may have resulted in researchers finding a “deterrent effect”

when in fact what they were finding was an “experiential effect.” Experience,
Paternoster (1987) claimed, is an important but often overlooked factor in as-
sessing risk in that persons with previous violations may subsequently have
lower estimates of risk than those less experienced or without experiences.
To address the issue of temporal ordering, subsequent studies have used
longitudinal samples to examine the effect of perceived certainty on criminal be-
havior (Paternoster, Saltzman, Waldo, and Chiricos 1982b; Piliavin et al. 1986).
Among these studies, Paternoster and Saltzman (1982b) found that juveniles
who have had experiences in the juvenile justice system had lower perceptions of
sanction risk on their future delinquent activities. Similar findings were found by
Minor and Harry (1982) and Horney and Marshall (1992), indicating that indi-
viduals who have had experiences with law violation had lower estimates of the
risk of punishment than those with no such experiences. On the other hand, other
perceptual deterrence studies, such as those by Burkett and Hickman (1982) and
Thomas and Bishop (1984), found no significant changes in the perceptions of
certainty among juveniles who had been sanctioned, compared to those who had
not. They found that juveniles who had been sanctioned did not display an “ex-
periential effect” as reflected in their perceptions of sanction risk. The issue of
an experiential effect is one of the main reasons to carefilly consider the tempo-
ral ordering between perceptions of sanction certainty and criminal behavior in
doing perceptual deterrence research. This study considers this important time
dimension in the subsequent analysis.
Another problem with previous perceptual deterrence studies was the lim-
ited generalizability of the samples used. Prior studies examining perceptual de-
terrence have had as their sample college students or delinquents involved in
minor infractions (Klepper and Nagin 1989; Ward and Tittle 1993). Piliavin et al.
(1986) noted that previous studies have analyzed “restricted populations of con-
ventional persons and nonserious crimes” and ignored “serious offenders . . .
the crimes and criminals our public fears most” (1986). On the same note,
Klepper and Nagin (1989) cautioned that researchers who have used conven-
tional persons in their samples may be asking these respondents questions about
criminal activities that may be outside their “scheme of conscious calculations.”
That is, the respondents may have never contemplated the act about which a risk
question is asked, thereby making their assessments of risk untenable. This
would raise serious concerns about validity.
The Study
This paper examines the effect of perceived punishment certainty on the
success (completion) or failure of offenders in an intensive drug probation pro-
gram. The “failure” of offenders in this program is deemed similar to the
122 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

commission of a criminal act in that this “failure” (through drug or alcohol use
while in the program, or absconding) is punishable by incarceration and should
therefore induce on the part of offenders the same utilitarian assessments of costs
and benefits as other criminal acts. In other words, the same principle of assessing
the costs and benefits of committing crimes proposed by deterrence theorists
should apply to the assessments of the costs (punishment) and benefits (temporary
high, freedom) of “failing” (using drugs, alcohol, absconding).
This study is an improvement over previous studies of perceptual deterrence
in that it addresses some of the methodological shortcomings found in those ear-
lier studies. First, the information available concerns individuals’ perceptions of
sanctions on a micro-level. Key perceptual variables are available for exami-
nation in terms of possible deterrent effects. A second improvement with the
current study is with regard to temporal ordering. The certainty measure used in
this study was a question asked when offenders first entered the Intensive
Supervision Program, about the risk of getting caught for a violation (drug or al-
cohol use, absconding). The measure therefore was collected prior to the of-
fenders completing or failing the program. Third, the sample is comprised of
individuals who were already in the criminal justice system, where punishment
was imminent and not merely a vague “theoretical” possibility. It is also com-
prised of persons committing serious crimes. Both of these elements have sel-
dom been found in previous perceptual deterrence studies (Piliavin et al. 1986).
Besides providing a test of deterrence, this study also provides important
policy information by assessing the predictors of “successes” or “failures” in
Intensive Probation Programs. Intensive drug probation programs are among a
host of alternatives-to-prison sanctions (often referred to as intermediate sanc-
tions) that have proliferated over the last few years in response to prison over-
crowding (Pearson 1991; Pearson 1988; Petersilia and Turner 1993; Ryan 1997).
These sanctions were designed to mete out punishments that were tough yet had
“rehabilitative” elements, in hopes of giving offenders chances for reform. As an
intermediate punishment, Intensive Supervision Programs (ISP) are character-
ized by more strict and intensified supervision of offenders in the community
compared to ordinary probation and stiffer penalties for probation violations
(Clear, Clear, and Braga 1993; Pearson 1988). Intensive Supervision Programs
attempt to alleviate problems associated with prison overcrowding while at the
same time satisfy a demand for punishments that are more harsh than traditional
probation.
Evaluations of ISPs are meager, and many are still underway, but data on
completed studies have asserted higher revocation rates for offenders in ISP pro-
grams compared to those in ordinary probation (Byrne 1990; Petersilia, Turner,
and Deschenes 1992a). This often means higher returns to prison because of the
 PROBATION VIOLATIONS 123

stiffer penalties attached to ISPs (Clear et al. 1993; Petersilia and Turner 1990).
Despite these high rates of revocation and the potential for more incarceration,
few studies have examined predictors of “successes” and “failures” in ISP (Clear
and Hardyman 1993). Researchers who have examined ISP outcomes have gen-
erally focused on “static” measures like demographics, criminal records, and
some programming measures, but glossed over the offenders’ attitudes about ISP,
especially their perceptions of the certainty of punishment if they fail the pro-
gram. Because ISP is characterized by stiffer penalties for violations, it is only
logical to test how individuals subjected to these potential penalties perceive the
penalties and what their behaviors are given these perceptions (Petersilia and
Turner 1990).

Sakple and Measures

Sample
This analysis uses data collected by the Institute for Criminological Re-
search at Rutgers University on all individuals who entered New Jersey’s In-
tensive Drug Probation program between January 1989 and April 30, 1990
(Pearson 1991). The total number of individuals who entered the program was
546, of which 5 16 were interviewed. The respondents were typical of offenders
serving their first prison term in the late 1980s and early 1990s (Pearson 1991),
most were convicted of drug crimes (62.5% for drug sales) or property offenses
(29.7%), and most were in their early to mid-twenties.
The goal of the original research was to predict a drug use relapse in
Intensive Supervision Probation (hereafter, ISP). Several measures relevant to re-
lapse were gathered, including the attitudes of offenders toward ISP, their atti-
tudes toward self, perceived certainty of punishments, criminal history, drdg use,
and alcohol use (Pearson 1991). Baseline measures were gathered, as well as
post-failure or post-success measures. The original study found that the objective
measure of punishment significantly predicted drug use while in the program and
that subjective assessments of the certainty of punishment (specifically a question
on the perceived number of positive urine tests before revocation) inversely pre-
dicted a drug-use relapse. These findings, however, used generally bivariate sta-
tistical techniques, and the outcome measure examined was limited to drug use.
This present analysis includes all types of “failures.” This includes drug
use, alcohol use, and absconding. Likewise, the analysis examines several factors
that have been identified as correlates of criminal activities and program failures.
Only baseline interviews are used as this ensures a sequential time-order be-
tween the characteristics and attitudes of offenders before treatment and their
outcomes.
124 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

Independent Measures
Of primary concern in the original study was the effect of the “perceived
certainty” of sanctions on the likelihood of successes (completions) or failures
(revocations) among offenders who were court-ordered into an intensive drug-
diversion program. Many questions that were asked of offenders concerned their
perceptions of the certainty of getting caught if they violated program regula-
tions. (Refer to Table 1 for a listing of the dependent and the independent mea-
sures.) Two of these measures are included in this paper because they had the
lowest numbers of missing data and because these measures were highly corre-
lated with the other perceptions of certainty measures. The first measure is the
offenders’ estimation of the number of times they can have a positive urine test
(positive for illicit substances) before they are revoked. Answers here ranged
from 0 “not sure” to 3 “3 or more times.” The second perceived certainty mea-
sure is the offenders’ estimation of the probability that a “street smart” person is
detected for using drugs while in the program. This measure vicariously esti-
mated the offenders’ perceptions of the risk of detection through a hypothetical
“street smart” individual. This measure was used because there was no question
in the original database that directly asked the offenders the probability of their
own detection (Pearson 1991). This measure also correlated highly with other
certainty questions, such as whether an “average” person will get caught and
whether an “average” person or a “street smart” person will get revoked. The
range of answers for this measure was 0 for ‘bo chance” through “100 percent
certainty” of getting caught. For the purposes of the bivariate analysis, this vari-
able was recoded and categorized as low ( 0 4 9 % chance), medium (5&69%
chance), and high (70-100%) probabilities of being caught, although the inter-
val-level measure was used in the multivariate models.
Also included in the analysis are the offenders’ feelings about two specific
requirements of the intensive drug-diversion program: the curfew requirement
and the community service requirement. The range of options for these variables
was 0 for “hate it” through 100 for “love it.” Again, for the purposes of the bi-
variate analyses, these measures were recoded to represent three categories:
“hate it” (049%), “in the middle” (50-69%), and “love it” (70-loo%), but the
interval-level measure was used in the multivariate models.
Besides the offenders’ perceptions of the certainty of sanctions and their at-
titudes toward the program, other domains of information that are included as in-
dependent measures are the offenders’ history of drug use (marijuana, cocaine,
crack and heroin), the percent of their friends using drugs, their criminal history,
and demographic information. Regarding drug use, the offenders were asked
if they have used marijuana, cocaine, crack or heroin in the year prior to enter-
ing the program. These questions were coded 0 for “no” and 1 for “yes.” The
 Table 1
Independent Variables and Results of Bivariate Analysis

Total Failure (0) Success (1) Chi Sq.

Total Sample 100% (455) 50.1% (228) 49.9% (227)

RaceiEthnicity 3 1.5***
Hispanic 17.0% (77) 41.6% (32) 58.4% (45)
White 47.4% (215) 40.0% (86) 60.0% (129)
Black 35.7% (162) 67.9% (1 10) 32.1% (52)
Sex 0.742
Male ( 0 ) 87.3% (397) 50.9% (202) 49.1 Yo (195)
Female (1) 12.7% (58) 44.8% (26) 55.2% (32)
Age 40.2***
18-22 22.2% (101) 71.3% (72) 28.7% (29)
23-25 22.0% (100) 47.0% (47) 53.0% (53)
26-30 22.4% (102) 55.9% (57) 44.1% (45)
3 1-35 14.3% (65) 44.6% (29) 55.4% (36)
Above 35 19.1% (87) 26.4% (23) 73.6% (64)
Number of Positive to Revoke 0.92
0 21.9% (93) 52.7% (49) 47.3% (44)
1 25.9% (110) 50.0% (55) 50.0% (55)
2 32.9% (140) 47.1% (66) 52.9% (74)
3 or more 19.3% (82) 52.4% (43) 47.6% (39)
Prob. Street Smart Caught 9.78**
Low 8.9% (38) 73.7% (28) 26.3% (10)
Medium 16.6% (71) 49.3% (35) 50.7% (36)
High 74.5% (318) 46.9% (149) 53.1% (169)
Feel about Curfew 5.6
Hate It 26.7% (115) 58.3% (67) 41.7% (48)
Middle 41.5% (179) 44.1% (79) 55.9% (100)
Love It 31.8%(137) 50.4% (69) 49.6% (68)
Feel about Comm. Svc. 0.284
Hate It 25.1% (108) 51.9% (56) 48.1% (52)
Middle 42.1% (181) 48.6% (88) 51.4% (93)
Love It 32.8% (141) 49.6% (70) 50.4% (71)
Percent of Friends Use Drugs 5.3
None 48.1% (201) 46.3% (93) 53.7% (108)
< Half 26.1% (109) 46.8% (51) 53.2% (58)
Half to All 25.8% (108) 59.3% (64) 40.7% (44)
Marij. before ISP 1.6
No (0) 57.3% (247) 47.4% (117) 52.6% (130)
Yes (1) 42.7%(184) 53.3% (98) 46.7% (86)
Cocaine before ISP 1.9
No (0) 47.3% (204) 46.6% (95) 53.4% (109)
Yes (1) 52.7% (227) 52.9% (120) 47.1%(107)
(continued)
126 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

Table 1 (continued)
Total Failure (0) Success (1) Chi Sq.

Crack before ISP 9.8**

No (0) 87.7% (378) 47.1%(178) 52.9% (200)
Yes (1) 12.3% (53) 69.8% (37) 30.2% (16)
Heroin before ISP 16.1***
No (0) 84.2%(363) 45.7% (166) 54.3% (197)
Yes (1) 15.8% (68) 72.1% (49) 27.9% (19)
Juvenile Convictions 10.54**
No (0) 84.2% (354) 46.0% (163) 54.0% (191)
Yes (1) 15.8% (101) 64.4% (65) 35.6% (36)
Indictable Convictions 4.05*
No (0) 63.7% (290) 46.6% (135) 53.4% (155)
Yes (1) 36.3% (165) 56.4% (93) 43.6% (72)
~

* = p < .05, ** = p < .01, *** = p < ,001

offenders were also asked to estimate the percentages of their friends outside of
prison who regularly used drugs. This measure is used in the paper to represent
the level of the offenders’ social networks who use drugs to examine any impact
this may have on the offenders’ outcomes in the program. Criminal history is
represented by two measures, the number of the offenders’juvenile convictions
and the number of indictable convictions. Demographic measures included age,
race and sex. These three variables have been shown in several other studies as
significant correlates of illicit activities (Sampson and Laub 1993; Steffensmeier
and Streifel 1991), as well as predictors of outcomes in some treatment modali-
ties (Smith and Silverman 1994). Thus, these measures are examined in the bi-
variate and the multivariate models.*
Dependent Measure
The dependent measure used is the offender’s status at the end of the
program. This took two forms for each of the multivariate analyses conducted.
For the Logistic Regression model, status at the end of the program was di-
chotomized into “failures” (0) and “successes” (1). “Successes” includes offend-
ers who completed the program or maxed-out of the program. “Failures”
includes offenders who were revoked to prison, were in jail awaiting revocation
hearing, or those who absconded from the program. Offenders who were still in
the program at the completion of the study (n = 87) were not included in the
Logistic Regression analysis (since they neither succeeded nor failed) but were
included in the Cox Regression model. The total number of cases for the Logistic
model was 455. The dependent variable for the Cox Regression model was the
 PROBATION VIOLATIONS 127

number of days offenders had stayed in treatment. This included all the offend-
ers referred to treatment, i.e., those who completed, failed, and those who were
still in treatment at the end of the study.
Two multivariate models are used in the succeeding analyses. First, the
Logistic Regression model is used to predict the characteristics of offenders
“most likely” to complete the program, and examines the effects of the measures
of perceived certainty on outcome status while holding all the other independent
variables constant. However, because this technique can only handle a dichoto-
mous dependent measure, offenders who were still in the program at the time of
the study’s completion are excluded from the analysis. The Cox Regression
model is then used to examine the predictors of the number of days offenders
stayed in treatment. Cox regression is part of a host of survival models that is
used with censored observations; these are obseivations for which the event of
interest (failure) has not yet occurred. Therefore, cases where neither a “success”
nor a “failure” is observed, are included in this analysis. In the Cox models, the
risk (proportional hazards) of failure for each offender’s discrete time in the pro-
gram is estimated using the baseline probability of their dropping out of the pro-
gram at specific time intervals. Cox regression predicts this “risk” as a function
of a set of independent measures.

Results
Bivariate Model
Table 1 outlines the bivariate results between the independent measures and
the dichotomous dependent measure, the outcome status. Of outcome status,
49.9% (227) of offenders successfully completed the program while 50.1% (228)
did not, and had either been revoked or were pending revocation at the comple-
tion of the study’s observation period. The table also shows that among the de-
mographic measures, race and age were statistically related to the program
outcome. Of race, White and Hispanic participants had significantly higher suc-
cess rates than African American participants (60% and 58% respectively, vs.
32%). Of age, older participants had higher success rates.
As to certainty measures, a significant association was found between one
of the measures (the participants’ opinions of a “street smart” person getting
caught), and success rates. A significant proportion of offenders who thought a
“street smart” person would have a low probability of getting caught, failed
(74%), compared to those who thought there was a medium or high probability
of getting caught. As to the offenders’ attitudes about specific program require-
ments, none had any bivariate relationships with the offenders’ outcome status.
Some of the drug use measures, however, were significantly related to the
offenders’ success or failure in the program, specifically crack use and heroin use
128 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

the year before program initiation. Offenders who used crack prior to program
initiation had failure rates of up to 70% compared to about 47% of those who did
not use crack. Likewise, offenders who used heroin the year before program ini-
tiation had failure rates of approximately 72% compared to 43% among those
who did not. Another measure, the percentage of a participant’s friends who used
drugs had proportions in the expected direction and almost significant at p = .05.
Here, participants who claimed to have no friends who used drugs had the low-
est proportions of failures (46%); those who claimed to have 50% or more
friends who used drugs had the highest proportions of failures (59%).
Finally, for both measures of criminal history, the numbers of prior juvenile
convictions and the numbers of prior indictable convictions were significantly re-
lated to program outcomes. Sixty-four percent of those who have had a juvenile
conviction failed in the program compared to 46% of those who didn’t. Also,
56% of those who have had an indictable conviction failed compared to 47% of
those who never had an indictable conviction
Multivariate Models
Logistic Regression. Logistic regression is used to examine the characteris-
tics of offenders who are likely to succeed or complete the program. To examine
the incremental effects of the independent measures, the logistic model first in-
cluded only the demographic measures, then program perception measures were
added, then the perceptions of certainty measures, friends using drugs, drug use
measures, and finally, criminal history. Table 2 presents the final model, which
includes all the independent measures. This model correctly classifies 7 1% of
“failures” and 70% of “successes,” which is better than the baseline distribution.
In the logistic regression, two demographic variables, race and age are sig-
nificant. These variables were significant when initially entered into the logistic
model, and remained significant even after controlling for additional measures.
The effects of these measures are similar to the direction observed in the bivari-
ate model with Whites and Hispanics significantly more likely than African-
Americans to complete the program, and older offenders significantly more
likely to succeed. Three other measures appeared to significantly predict out-
come status: one measure of the offenders’ perceptions of certainty (a “street
smart” individual getting caught), the offenders’ crack use prior to program ini-
tiation, and the offenders’ heroin use prior to program initiation. As observed in
the bivariate models, the offender’s vicarious estimation of detection (of a hypo-
thetical “street smart” individual) was significantly related to outcome status,
even after controlling for other measures. The more offenders believed that a
“street smart” person would get caught violating the rules of the program, the
more they themselves were likely to succeed in the program. This points directly
to the relevance of their estimation of the certainty of detection on their perfor-
 PROBATION VIOLATIONS 129

Table 2
Full Model of the Predictors of ISP Outcome

B S.E. EXP (B)

Race/Ethnicity
Race 1 (Hispanic) 1.43 0.35 4.18***
Race 2 (White) 1.37 0.26 3.93***
Sex 0.27 0.36 1.31
Age 0.08 0.02 1.09***
Number Positive Urines to Revoke 0.17 0.11 1.18
Prob. Street Smart Person Revoked 0.01 0.00 1.1 1”
How Feel about Curfew 0.0 1 0.01 1.oo
How Feel about Comm. Svc. -0.01 0.01 0.99
Percent Friends Using Drugs -0.01 0.00 0.99
Juvenile Convictions -0.44 0.29 0.64
Indictable Convictions -0.41 0.25 0.67
Use Marijuana before ISP 0.08 0.25 1.09
Use Cocaine before ISP 0.31 0.25 1.37
Use Crack before ISP -0.79 0.36 0.45*
Use Heroin before ISP -3.92 0.81 0.29***
-2 Log Likelihood 454.15
Model Chi-square 108.7***

* = p < .05, ** =p <.01, *** = p < .001.

mance in the program. Offenders who perceived that detection was imminent
(even for a “street smart” individual), were likely to complete the program.
Among other significant measures, use of crack or heroin before program initia-
tion is a significant predictor of outcomes, even when other variables are con-
trolled. However, with regard to criminal history, it appears that the initial
bivariate relationships observed are spurious and the initial relationships ob-
served are accounted for by the other measures included in this model. Table 3
presents a reduced model, which includes only significant measures shown
above, and two other measures that were near significant. As shown, the same
variables significant in the full model remained significant in this reduced model.
Cox Regression. The logistic models above inform us of the characteristics
of offenders who are likely to succeed or fail the intensive diversion program.
However, the previous analysis only allowed a dichotomous dependent variable,
130 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

Table 3
Reduced Model of the Predictors of ISP Outcome

RacelEtknicity
Hispanic 1.4 0.4 4.23***
White 1.3 0.3 3.57***
Black (Comparison Category)
Age 0.1 0.0 1.08***
Number of Positive Urines to Revoke 0.2 0.1 1.16
Prob. Street Smart Person Revoked 0.0 0.0 1.10*
Percent of Friends Using Drugs -0.0 0.0 1.oo
Use Crack before ISP -0.7 0.4 0.52
Use Heroin before ISP -1.2 0.3 0.29***
-2 Log Likelihood 4,654.7
Model Chi-square 97.02***

* =p < .05, ** =p <.O1, *** =p < .001.

making it unfeasible to predict the likelihood of success or failure among those

offenders who were still in the program at the time the study ended. Cox regres-
sion models address this problem. As a survival technique, Cox models allow for
the prediction of a dependent variable (here, length of time to failure/dropping-
out) given censored observations.
Results are presented in Table 4. As shown, the same variables that were
significant in the Logistic models are significant here, reinforcing the initial find-
ings. Because this model predicts time to failure, a negative beta coefficient (B)
indicates that the variable “slows down” time to the event (failure), and a posi-
tive coefficient means faster time to failure (increased hazards). The same direc-
tions that were observed in the Logistic models were also observed here. Whites
and Hispanics, compared to African Americans had slower times-to-failure,
older offenders had slower times-to-failure, and the same is true of offenders
who perceived that “street smart” offenders would get caught. Variables that ap-
peared to accelerate the offenders’ times-to-failure were crack and heroin use be-
fore treatment initiation as well as greater percentages of fiends using drugs.
The latter is the only measure that is significant in this model that was not sig-
nificant in the logistic models. Therefore, although the effect of this measure was
not large enough to predict outcome status, friendships with individuals who
used drugs appeared to pose a significant “risk” of accelerated failure.
 PROBATION VIOLATIONS 13 1

Table 4
Cox Regression Model of Time to ISP Failure

RaceIEthnicity
Hispanic -0.79 0.23 0.45* **
White -0.67 0.16 0.51***
Black (Comparison Category)
Sex -0.28 0.23 0.75
Age -0.04 0.01 0.96**
Number of Positive Urines to Revoke -0.10 0.07 0.90
Prob. Street Smart Person Revoked -0.01 0.03 OM*
How Feel about Curfew -0.01 0.00 1.oo
How Feel about Comm. Svc. 0.00 0.00 1.oo
Percent Friends Using Drugs 0.01 0.00 1.01*
Use Marijuana before ISP -0.07 0.16 0.93
Use Cocaine before ISP -0.06 0.15 0.94
Use Crack before ISP 0.51 0.19 1.67**
Use Heroin before ISP 0.59 0.18 1.81**
Juvenile Convictions 0.33 0.17 1.40
Indictable Convictions 0.20 0.15 1.22
-2 Log Likelihood 2,138.5
Model Chi-square 105.62***

Discussion
Deterrence researchers have recently begun to recognize the importance of
the offenders’ perceptions of punishment certainty on their assessments of the
risks and benefits of criminal activities. However, tests of deterrence theory,
particularly those that examine the offenders’ perceptions of the risks of sanc-
tions, are still limited in at least two ways: first in establishing the time-order
between perceptions of the certainty of punishments and criminal behaviors;
and second, in using samples that were not generalizable to the offending
population.
This paper addressed these two issues by examining the effects of perceived
sanction certainty on the outcome status of individuals who were mandated
to complete an intensive drug diversion program. The time-order issue was
132 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

addressed by taking the offenders’ responses to perceived certainty of sanctions

as soon as they initiated the program and therefore before any of their subse-
quent outcome statuses. The sampling issue was addressed by examining behav-
iors of individuals who were already under criminal justice custody and have
committed serious crimes. Threats of sanctions were therefore imminent and not
merely a “theoretical” possibility.
The “criminal behavior” studied was the offenders’ failure to comply with
program requirements, thereby risking revocation and re-incarceration. As stated
in the beginning of this paper, although the failure to comply with program re-
quirements is not substantively “criminal,” it solicits the same types of formal
sanctions as criminal behaviors, if not more so given the stiff penalties often at-
tached to violations of court-ordered programs (Clear Todd and Hardyman 1993;
Petersilia and Turner 1990). Thus, the same logic that is applied to the offenders’
calculations of the risks and benefits of criminal behaviors also applies to the
calculations of the risks and benefits for violating court-mandated programs.
The results shown in the logistic and the Cox regression models provide
support for the positive effect of the offenders’ perceptions of the certainty of
sanctions on their outcome status and their lengths of time in the program.
Controlling for measures that previously have been found to correlate with
criminal behaviors and program attrition, the offenders’ perceptions that a
“street smart” individual will get caught if this individual violates the require-
ments of the program remains a significant predictor of program completion by
the offender. This finding has implications for policy. Attempts by policymakers
to reduce “failures” in alternatives-to-prison programs may be efficiently ad-
dressed by increasing the offenders’ perceptions of sanction certainty. However,
more studies are needed to careklly examine how offenders’ perceptions of
sanction threats are affected by law, criminal justice policy, and other social
processes. This study is only a start of several more studies that need to be com-
pleted to better assess the role of perceived sanction certainty on offenders’
avoidance of criminal behaviors or compliance with criminal justice-based pro-
gramming and rehabilitation.
Besides perceptions of the certainty of sanctions, the demographic mea-
sures of race and age were also found to be significant predictors of outcome sta-
tus. The significance of race may perhaps be due to variables not controlled for
in this model, such as human capital or family stability, that were previously
found to correlate with race and ethnicity (Fagan 1993). However, these mea-
sures were not available in this study. Specific interactions were nevertheless
considered in the logistic and the Cox regression models to see if the effects of
ethnicity were clarified. The authors examined the interactions between ethnicity
and friends using drugs, as well as ethnicity and criminal history, but these in-
teractions were far from significant and did not contribute to the explanatory
powers of the models.
 PROBATION VIOLATIONS 133

Besides ethnicity, age and use of crack or heroin were also significant pre-
dictors of outcome status. The age effect was in the expected direction: older of-
fenders were less likely to “fail.” This trend reinforces the existing empirical and
theoretical knowledge about this element, which has often been shown to nega-
tively correlate with criminal behaviors and program attrition. Nevertheless, it
poses a compelling issue, that younger offenders need more interventions to
keep them in the programs than older offenders, given their propensity for fail-
ures. Similarly, the effects of the use of crack or heroin on the outcome status of
offenders call for some intervening measures for offenders who have had histo-
ries of crack and heroin abuse, as these attributes were also significant predictors
of noncompletion and accelerated failure.

ENDNOTE

*Tolerance statistics and Pearson’s correlations were performed to check for multicollinearity
across the independent measures. No multicollinearity was observed. Tolerance levels ranged from
.78 to .94 which are acceptable levels, and correlation coefficients ranged from ,013 to .399.

REFERENCES

Akers, R. L. 1994. Criminological Theories: Introduction and Evaluation. Los Angeles, CA:
Roxbury Publishing Company.
Andenaes, J. 1975. “General Prevention Revisited: Research and Policy Implications.” The Journal
of Criminal Law and Criminology 66(8):33845.
Antunes, G., and A. L. Hunt. 1978. “Certainty and Severity of Punishment.” Pp. 381-89 in Justice
and Corrections, edited by N. B. Johnston and L. D. Savitz. New York: John Wiley and Sons.
(Reprinted from “The Impact of Certainty and Severity of Punishment on Levels of Crime in
American States; An Extended Analysis.” The Journal of Criminal Law and Criminology
64[1973]:486-93.)
Austin, J., C. Jones, and J. Krmer. 1994. National Assessment of Structured Sentencing. San
Francisco: National Council on Crime and Delinquency.
Beccaria, C. 1963. On Crimes and Punishment (Henry Paolucci, Trans.). London: Bobbs-Merrill Co.
Blumstein, A,, J. Cohen, and D. Nagin (eds.). 1978. Estimating the Effects of Criminal Sanctions on
Crime Rates. Washington, DC: National Academy Press.
. 1976. “The Dynamics of a Homeostatic Punishment Process.” Journal of Criminal Law and
Criminology 67(3):317-34.
Bridges, G. S., and J. A. Stone. 1986. “Effects of Criminal Punishment on Perceived Threat of
Punishment: Toward an Understanding of Specific Deterrence.” Journal of Research in Crime
and Delinquency 23(3):207-39.
Burkett, S. R., and C. A. Hickman. 1982. “An Examination of the Impact of Legal Sanctions on
Adolescent Marijuana Use: A Panel Analysis.” Journal of Drug Issues 12(1):73-87.
Byrne, J. M. 1990. “The Future of Intensive Probation Supervision and the New Intermediate
Sanctions.” Crime and Delinquency 36:&41.
134 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

Byme, J. M., and A. Pattavina. 1992. “The Effectiveness Issues: Assessing What Works in the Adult
Community Corrections System.” Pp. 1840 in Smart Sentencing, edited by J. Byme, A.
Lurigio, and 1. Petersilia. Newbury Park, CA: Sage Publications.
Clark, R. D. 1988. “Celerity and Specific Deterrence: A Look at the Evidence.” Canadian Journal of
Criminology 30: 109-120.
Clear, T., and P. L. Hardyman. 1993. “The New Intensive Supervision Movement.” Crime and
Delinquency 36:42-60.
Clear, T., V B. Clear, and A. A. Braga. 1993. “Correctional Alternatives for Drug Offenders in an Era
of Overcrowding.” The Prison Journal 73(2): 178-98.
Cloward, R., and L. E. Ohlin. 1960. Delinquency and Opportunity: A Theory ofDelinquent Gangs.
New York The Free Press.
Cohen, L. 1978. “Problems of Perception in Deterrence Research.” Pp. 84-99 in Quantitative Studies
in Criminoloa, edited by C. Wellford. Sage Research Progress Series in Criminology, vol. 8.
Beverly Hills, CA: Sage Publications, Inc.
Ekland-Olson, S., and W. R. Kelly. 1993. “Justice under Pressure.” In Research in Criminology,
edited by A. Blumstein and D. P. Farrington. New York, NY: Springer-Verlag.
Erickson, M. L., J. P. Gibbs, and G. F. Jensen. 1977. “The Deterrence Doctrine and the Perceived
Certainty of Legal Punishments.” American Sociological Review 42:305-3 17.
Fagan, J. 1994. “Do Criminal Sanctions Deter Drug Crimes?“ Pp. 188-214 in Drugs and Crime:
Evaluating Public Policy Initiatives, edited by D. L. MacKenzie and C. D. Uchida. Thousand
Oaks, CA: Sage Publications.
. 1993. “Set and Setting Revisited Influences of Alcohol and Illicit Drugs on the Social
Context of Violent Events.” Pp. 161-92 in Alcohol and Interpersonal Violence: Fostering
Multidisciplinary Approaches, NIAAA Research Monograph #24, edited by S. Martin.
Rockville, MD: U S . Public Health Service.
Gibbs, J. P. 1982. “Law as a Means of Social Control.” Pp. 83-1 13 in Social Control: Views from
Social Sciences, edited by .IP.. Gibbs. Beverly Hills: Sage Publications, Inc.
. 1975. Crime, Punishment and Deterrence. New York: Elsevier.
Glaser, D. 1979. “A Review of Crime Causation Theory and Its Application.” Pp. 203-238 in Crime
and Justice: An Annual Review of Research (Vol. 13, edited by N. Morns and M. Tomy.
Chicago: The University of Chicago Press.
Gottfredson, M. R., and T. Hirschi. 1990. A General Theory of Crime. Palo Alto, CA: Stanford
University Press.
Grasmick, H. G., and Bryjack, G. 1980. “The Deterrent Effect of Perceived Severity of Punishment.”
Social Forces 59:471-91.
Homey, J., and I. H. Marshall. 1992. “Risk Perceptions among Serious Offenders: The Role of Crime
and Punishment.” Criminology 30(4):575-94.
Klepper, S., and D. Nagin. 1989. “The Deterrent Effect of Perceived Certainty and Severity of
Punishment Revisited.” Criminology 27(4):72 1 4 6 .
Lipton, D., R. Martinson, and J. Wilks. 1975. The Effectiveness of Correctional Treatment.New York
Praeger.
Lombroso-Ferrero, G. 1911. Criminal Man, According to the ClassiJication of Cesare Lombroso.
New York and London: G. P. Putman.
MacKenzie, D. L. 1994. “Shock Incarceration as an Alternative to Drug Offenders.” Pp. 229-44 in
Drugs and Crime: Evaluating Public Policy Initiatives, edited by D. L. MacKenzie and C. D.
Uchida. Thousand Oaks, CA: Sage Publications.
Martinson, R. 1974. “What Works?--Questions and Answers about Prison Reform.” The Public
Interest 35:22-54.
Maxwell, C . D. 1997. “The Specific Deterrent Effect of Arrest on Aggression between Intimates and
Spouses.” Dissertation. Newark, NJ: Rutgers University
 PROBATION VIOLATIONS 135

. 1995. “Estimating the Effects of Neighborhood Context on the Development of

Aggression.” Analysis prepared for the Metropolitan Area Child Study. Ann Arbor, MI.
Maxwell, S. R. 1996. “Conservative Correctional Policies and Practices in the U.S.: Rhetoric,
Reality, and Future Prospects.” Presentation to the International Conference on Corrections.
Seoul, Korea.
Meier, R. 1978. “The Deterrence Doctrine and Public Policy: A Response to Utilitarians.” Pp.
233-47 in Preventing Crime, edited by J. Cramer. Sage Criminal Justice System Annuals, vol.
10. Beverly Hills: Sage Publications, Inc.
Merton, R. K. 1938. “Social Structure and Anomie.” American Sociological Review 3:672-82.
Minor, W. W., and J. Harry 1982. “Deterrent and Experiential Effects in Perceptual Deterrence
Research: A Replication and Extension.” Journal of Research in Crime and Delinquency
19:190-203.
Morris, N. 1966. “Impediments to Penal Reform.” University of Chicago Law Review 33:631.
Mueller, G. 0. W. 1977. Sentencing: Process and Purpose. Publication of the Criminal Law
Education and Research Center, vol. 12. Springfield, IL: Charles C. Thomas.
Palmer, T. 1992. The Re-emergence of Correctional Intervention. Newbury Park, CA: Sage
Publications.
Paternoster, R. 1987. “The Deterrent Effect of the Perceived Certainty and Severity of Punishment:
A Review of the Evidence and Issues.” Justice Quarterly 4(2):173-217.
Paternoster, R., and L. E. Saltzman. 1982a. “Casual Ordering in Deterrence Research: An
Examination of the Perceptions-Behavior Relationship.” Pp. 55-70 in Deterrence
Reconsidered: Methodological Innovations, edited by J. Hagan. Beverly Hills, CA: Sage
Publications.
. 1982b. “Perceived Risk and Deterrence: Methodological Artifacts in Perceptual Deterrence
Research.” Journal of Criminal Law and Criminology 73:1238-58.
Paternoster, R., L. E. Saltzman, G. P.Waldo, and T. G. Chiricos. 1983. “Perceived Risk and Social
Control: Do Sanctions Really Deter?” Law and Society 17:457-79.
Pearson, E S. 1991. Deterring Drug Use with Intensive Supervision. Rutgers University.
. 1988. “Evaluation of New Jersey’s Intensive Supervision Program.” Crime and Delinquency
34:43748.
Petersilia, J., and S. Turner. 1993. “Intensive Probation and Parole.” Pp. 281-336 in Crime and
Justice: An Annual Review of Research (Vol. l,), edited by N. Morris and M. Tonry. Chicago:
The University of Chicago Press.
. 1990. “Comparing Intensive and Regular Supervision for High-Risk Probationers: Early
Results from an Experiment in California.” Crime and Delinquency 36:87-111.
Petersilia, J., S. Turner, and E. P. Deschenes. 1992a. “The Costs and Effects of Intensive Supervision
for Drug Offenders.” Federal Probation 56(4): 12-17.
. 1992b. “Intensive Supervision Programs for Drug Offenders.” Pp. 18-40 in Smart
Sentencing, edited by J. Byme, A. Lurigio, and J. Petersilia. Newbury Park, CA: Sage
Publications.
Piliavin, I., C. Thorton, R. Gartner, and R. L. Matsueda. 1986, February. “Crime, Deterrence, and
Rational Choice.” American Sociological Review 51(l):101-119.
Ryan, J. 1997. “Who Gets Revoked? A Comparison of Intensive Supervision Successes and Failures
in Vermont.” Crime and Delinquency 43:104-118.
Sabol, W. J., and J. P. Lynch. 1997, August. “Crime Policy Report: Did Getting Tough on Crime
Pay?’ [www.urban.org/crime/crime,htm]. The Urban Institute (24/09/97).
Sampson, R. J., and J. H. Laub. 1993. Crime in the Making: Pathways and Turning Points through
Life. Cambridge, MA: Harvard University Press.
Schneider, A. L., and L. Ervin. 1990. “Specific Deterrence, Rational Choice, and Decision
Heuristics: Application in Juvenile Justice.” Social Science Quarterly 71(3):585-601.
136 SHEILA ROY0 MAXWELL AND M. KEVIN GRAY

Singer, B. F. 1970. “Psychological Studies of Punishment.” California Law Review 58(2):405-443.

Smith, L. G., and M. L. Silverman. 1994. “Functional Literacy Education for Jail Inmates: An
Examination of the Hillsborough County Jail Education Program.” The Prison Journal
74(4):41+32.
Steffensmeier, D., and C. Streifel. 1991. “The Distribution of Crime by Age and Gender across Three
Historical Periods, 1935, 1960, 1985.” Social Forces 69:869-94.
Sutherland, E. H. 1947. Principles of Criminology (4th Ed.). Chicago: J. B. Lippincott Co.
Thomas, C. W., and D. M. Bishop. 1984. “The Effects of Formal and Informal Sanctions on
Delinquency: A Longitudinal Comparison of Labeling and Deterrence Theories.” The Journal
of Criminal Law and Criminology 75(4): 122245.
Tittle, C. R. 1980a. “Evaluating the Deterrent Effects of Criminal Sanctions.” Pp. 381402 in
Handbook of Criminal Justice Evaluation. Beverly Hills: Sage Publications, Inc.
. 1980b. Sanctions and Social Deviance. New York: Praeger Publishers.
Tonry, M. 1996. Sentencing Matters. Studies in Crime and Public Policy. New York, N Y Oxford
University Press.
Ward, D. A,, and C. R. Tittle. 1993. “Deterrence or Labeling: The Effects of Informal Sanctions.”
Deviant Behavior 14143-64.
Wichiraya, T. 1996. Simple Theov, Hard Reality: The Impact of Sentencing Reforms on Courts,
Prisons, and Crime. Albany, NY: State University of New York Press.
Williams, K. R., and R. Hawkins. 1986. “Perceptual Research on General Deterrence: A Critical
Review.” Law and Socieiy Review 20:544-72.
Zimring, F, E., and G. J. Hawkins. 1973. Deterrence: The Legal Threat in Crime Control. Chicago
and London: The University of Chicago Press.