16
AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE* NEAL SHOVER University of Tennessee CAROL Y. THOMPSON East Carolina University We specify an individual-level model linking crime desistance to esti- mates of legal risk, differential expectations, degree ofpast success at legit- imate and criminal pursuits, and age. OLS and logistic regression procedures are used to estimate the model using longitudinal data on seri- ous, previously imprisoned offenders. As predicted, age decreases esti- mates of the likely payoffs from crime and legitimate employment. Contrary to predictions, age is unrelated to the perceived legal risk of renewed criminal participation. Age, past success at avoiding confine- ment, expectations of success from crime, and level of education are sig- nificant predictors of crime desistance. Neither the perceived legal risk of crime nor expectations of success through straight pursuits significantly predict desistance. We suggest an interpretation for these anomalous findings. The past decade has witnessed substantial renascent interest in the variable careers and criminal participation of street offenders (e.g., Blumstein et al., 1988; Petersilia, 1980). This movement has focused new attention on crime desistance, that is, the termination of criminal careers (e.g., Ohlin et al., 1988). No one disputes that the overwhelming majority of street offenders, including those whose criminal participation extends into adulthood, eventu- ally desist from serious criminal activities. The positive relationship between age and desistance among adults is supported by official crime statistics, self- report studies, cohort follow-up investigations, and offender autobiographies (Farrington, 1986; Gartner and Piliavin, 1988; Shover, 1985). Historically, the desistance phenomenon has been approached inferentially (Ohlin et al., 1988). The lion’s share of research on the later stages of crimi- nal careers has focused on recidivism or the failure to desist from renewed criminal participation. Because most studies of recidivism are motivated by interest in parole prediction or other policy questions, there are few theoreti- cal explanations for desistance (e.g., Glaser, 1964, 1980). We draw primarily ~ ~~~~ The data used in the paper were collected originally by the Rand Corporation and were made available by the Inter-University Consortium for Political and Social Research. Neither the collector of the original data nor the consortium bears any responsibility for the analyses or interpretations presented here. We are grateful to Michael Benson and anonymous reviewers for comments on an earlier draft. CRIMINOLOGY VOLUME 30 NUMBER 1 1992 89

AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

Embed Size (px)

Citation preview

Page 1: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE*

NEAL SHOVER University of Tennessee

CAROL Y. THOMPSON East Carolina University

We specify an individual-level model linking crime desistance to esti- mates of legal risk, differential expectations, degree ofpast success at legit- imate and criminal pursuits, and age. OLS and logistic regression procedures are used to estimate the model using longitudinal data on seri- ous, previously imprisoned offenders. As predicted, age decreases esti- mates of the likely payoffs from crime and legitimate employment. Contrary to predictions, age is unrelated to the perceived legal risk of renewed criminal participation. Age, past success at avoiding confine- ment, expectations of success from crime, and level of education are sig- nificant predictors of crime desistance. Neither the perceived legal risk of crime nor expectations of success through straight pursuits significantly predict desistance. We suggest an interpretation for these anomalous findings.

The past decade has witnessed substantial renascent interest in the variable careers and criminal participation of street offenders (e.g., Blumstein et al., 1988; Petersilia, 1980). This movement has focused new attention on crime desistance, that is, the termination of criminal careers (e.g., Ohlin et al., 1988). No one disputes that the overwhelming majority of street offenders, including those whose criminal participation extends into adulthood, eventu- ally desist from serious criminal activities. The positive relationship between age and desistance among adults is supported by official crime statistics, self- report studies, cohort follow-up investigations, and offender autobiographies (Farrington, 1986; Gartner and Piliavin, 1988; Shover, 1985).

Historically, the desistance phenomenon has been approached inferentially (Ohlin et al., 1988). The lion’s share of research on the later stages of crimi- nal careers has focused on recidivism or the failure to desist from renewed criminal participation. Because most studies of recidivism are motivated by interest in parole prediction or other policy questions, there are few theoreti- cal explanations for desistance (e.g., Glaser, 1964, 1980). We draw primarily

~ ~~~~

The data used in the paper were collected originally by the Rand Corporation and were made available by the Inter-University Consortium for Political and Social Research. Neither the collector of the original data nor the consortium bears any responsibility for the analyses or interpretations presented here. We are grateful to Michael Benson and anonymous reviewers for comments on an earlier draft.

CRIMINOLOGY VOLUME 30 NUMBER 1 1992 89

Page 2: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

90 SHOVER AND THOMPSON

from ethnographic investigations to construct an individual-level model link- ing age, degree of past success at legitimate and criminal pursuits, expecta- tions about the likely payoffs from criminal and noncriminal behavior, estimates of legal risk, and crime desistance. We then evaluate the model using longitudinal data from a sample of serious offenders. We note points of similarity between our findings and reports by other investigators who have employed similar samples and longitudinal data.

THE DESISTANCE PROCESS

Two logically complementary constructions of the theoretical link between age and clime desistance have been offered. The first genre of explanation posits a direct, positive relationship between the two. Many age-related bio- psychosocial factors thought to be related to criminal participation presuma- bly contribute to crime desistance as well (Gove, 1985). Walsh (1986: 150) suggests, for example, that aging makes offenders “less audacious” and, there- fore, less interested in crime and other high-risk pursuits. Perhaps the best known example of a direct-effects hypothesis stresses the importance of aging- produced maturation and its dampening effects on criminal proclivities (Glueck and Glueck, 1937). Although this hypothesis has been criticized on several grounds (e.g., Gartner and Piliavin, 1988), the possibility of other direct links between age and crime desistance cannot be ruled out (Wilson and Hermstein, 1985). Shover (1985) suggests, for example, that aging improves offenders’ ability and inclination to calculate more precisely and carefully the results of past and prospective criminal involvement and the result is an increased probability of desistance.

Grounded securely in findings from ethnographic investigations, a second type of explanation hypothesizes significant indirect links between age and desistance. Compatible with social learning theory (Akers, 1985), the funda- mental assumptions underlying this construction are simple and straightfor- ward: To the extent offenders meet with self-defined success from crime, they will be optimistic about the potential payoffs from continued criminal partici- pation and will be unlikely to desist; to the extent they meet with failure from crime, they will be unlikely to expect a reversal of this pattern and will be more likely to desist.

Success at criminal pursuits strengthens commitment to criminal others and criminal lines of action and erodes the perceived formal risk of crime. Using the metric of thieves and hustlers, those who earn well from crime while serving little time in prison are successful (Shover, 1973). Despite whatever short-term monetary success they may enjoy, however, very few street offenders attain long-lasting financial success illegally. Their criminal financial gains are dissipated quickly on alcohol and other drugs, ostentatious consumption, and “good times.” Their performance avoiding imprisonment

Page 3: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 91

is no better; aging and criminal experience do not improve significantly their odds of avoiding arrest, conviction, and confinement. For the overwhelming majority of street offenders, extended involvement in crime brings only pen- ury, interspersed with modest, quickly depleted criminal gains and repeated imprisonment. This is the most important reason they eventually lose confi- dence in prospects for achieving success by committing street crimes (Walsh, 1980).

But these are not the only reasons crime becomes less attractive with increasing age. Shover (1983, 1985) contends that aging offenders gradually become aware of time-until-death as a finite, diminishing resource and that they become increasingly unwilling to risk wasting their remaining years in prison. Experience managing criminal pursuits eventually causes them also to grow “progressively weary” at the “hassle of everyday boundary mainte- nance and their feelings of being expatriated from conventional society” (Adler, 1985: 132). Even the daily routines of managing criminal involvement become tiring and burdensome to aging offenders. Many lower their material aspirations and find increasing interest and satisfaction in emergent goals, such as contentment and peace of mind that are experienced as antithetical to criminal participation. Consequently, the allure of crime diminishes substan- tially as offenders get older. According to Maguire (1982:89),

The impetus to think seriously about [desistance] seems to come in many cases from a gradual disenchantment with the criminal life in its totality: the inability to trust people; the frequent harassment by the police; the effects on wives and children when the offender is in prison; and [other hassles]. As people grow older such a process can become more painful and depressing and the optimistic outlook can give way to a feeling of being caught in a trap.

This growing disenchantment with the criminal life also causes offenders to lower their expectations for achieving success via criminal means.

What is true of the effects of criminal performance is also true of offenders’ performance in legitimate roles and employment: The degree of success they have known in the past determines their estimates of the likely payoff from more of the same. Most street offenders have known little success in the legit- imate work world, in part because it is difficult to reconcile sustained criminal involvement with the demands of a @-hour workweek. Confinement further expropriates the young, energetic years and may leave offenders ill-prepared and demographically mismatched for many conventional occupations and career timetables. Prospects can be particularly bleak for those who have served multiple prison terms. Too late they see how repeated confinement severely constrains prospects for a successful and rewarding straight life.

Page 4: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

92 SHOVER AND THOMPSON

Their underclass background and blue-collar or menial employment exper- iences also cause many street offenders eventually to scale down their legiti- mate expectations. Coupled with growing disenchantment with the criminal life, alternative and noncriminal life-styles become increasingly apparent and attractive to street offenders as they get older. This does not mean, however, that the expected rewards of noncrime increase since aging offenders come to have a very accurate assessment of their legitimate prospects (Shover, 1985).

In sum, we suggest theoretically that increasing age and past performance in straight and criminal pursuits determine the offender’s diflerential expecta- tions. These are “factors that reliably influence the decision to engage or not to engage in criminal acts” (Hirschi, 1986:116). They include general, pan- situational constraints on decision making and situational components that are determined by one’s momentary perceptions of needs, opportunities, and risks (Glaser, 1980). We hypothesize that age and minimal success at crime cause offenders eventually to reduce their expectations for achieving success by continued or renewed criminal participation. Although age may be unre- lated to expectations for achieving success by straight pursuits, past success at legitimate employment does increase straight expectations.

Changes in the perceived formal risk of criminal participation also may contribute to desistance as street offenders get older (Glassner et al., 1983). Some investigators suggest that offenders grow increasingly preoccupied with and fearful of the legal risks of crime. Cusson and Pinsonneault (1986:76) assert that “it is clear that, with age, criminals raise their estimates of the certainty of punishment.” Fear of reimprisonment was the “primary motive” of self-defined desisters in a sample of adult offenders (Meisenhelder, 1977:322). The same is reported by Cromwell et al. (1991:83), who state that “for most of the desisters we interviewed . . . the final decision to terminate a criminal lifestyle was primarily the result of their increasing fear of punish- ment.” As these authors see it, age increases the perceived legal risk of crimi- nal participation and that increases the odds of desistance. Shover cautions (1985:126) that the relationship between age and estimates of legal risk may not be direct but mediated by differential expectations; risk may increase only if criminal expectations are reduced or straight expectations are raised. Nev- ertheless, a direct and positive link between age and risk reflects the reports of most investigators who have explored desistance. It also reflects the results of research on samples drawn from the general population (e.g., Grasmick and Bursik, 1990; Grasmick and Milligan, 1976; Tittle, 1980).

Past research has shown that aging offenders who manage to establish a secure and rewarding social niche and to develop commitment to conven- tional lines of activity significantly improve their odds of desistance (Glaser, 1964; Irwin, 1970; Meisenhelder, 1977; Reitzes, 1955; Shover, 1985; West, 1978). Establishment of this stake in conformity “.ijve[s] meaning to life and provide[s] an incentive for respecting the law” (Cusson and Pinsonneault,

Page 5: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 93

j Criminal Expectations

1986:SO). Recent analysis of longitudinal data collected by the Gluecks finds support again for the importance of these conventional social bonds (Samp- son and Laub, 1990). Due to the lack of data on the number and strength of postrelease social bonds and legitimate activities, we were unable to test this part of the theory of desistance (Shover, 1985).1

Figure 1 depicts the theoretical model we have elaborated. This ethno- graphically based explanation as yet “has not been confirmed by deductive analyses of hard data” (Gartner and Piliavin, 1988:300). The remainder of this paper rectifies this omission. We explore empirically the causal connec- tions among age, offenders’ criminal and straight expectations, perceived for- mal risks of criminal participation, and desistance from crime.

Figure 1 The Desistance Process

Straight Success

Criminal Success

Age

Controls -.

‘1 Risk /

\ 1 Straight I Expectations

rn Desistance

DATA Data for the independent variables in Figure 1 were collected in 1978,

when investigators from the Rand Corporation administered the Rand Inmate Survey (RIS) to 1,469 male inmates of 12 prisons in Texas, Michigan,

This contention is counter to the argument presented by Gottfredson and Hirschi 1. (1990).

Page 6: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

94 SHOVER AND THOMPSON

and California (Peterson et al., 1982). On average the subjects were approxi- mately 27 years of age and had two previous felony convictions. Approxi- mately two-thirds of them were black, Hispanic, or another minority. Although the precise relationship of this sample to larger populations of theo- retical and substantive interest is unknown, the sample does capture serious offenders, the object of considerable public and official concern.

In addition to questions about past criminal behavior, the RIS measured respondents’ confinement history, earnings from legitimate employment dur- ing their most recent stay in the free community, earnings from crime during the same period, and estimates of the risks and rewards of future straight and criminal activities. The questionnaire was administered to groups of between 20 and 30 prisoners in classrooms, visiting rooms, or other available space inside the prisons. It was not anonymous.

In 1984, Rand investigators examined state “rap sheets” and correctional records for follow-up data on the post release criminal activities of the inmates who completed the RIS. Klein and Caggiano (19865) report that 184 of the original cohort either were deceased or still incarcerated in 1984. A comparison of the original RIS inmates with the remaining subjects indi- cated the two groups were similar in age, number of prior felony convictions, current conviction offense, and percent minority (1986:6). Of the 1,285 sub- jects who were available for follow-up, 1,023 were released at least 36 months before rap sheets were generated. Our analysis utilizes only the 948 inmates from this 36-months cohort who were 18 years of age or older at the time of the original RIS.

Rand investigators collected three categories of follow-up data: arrests, convictions, and incarcerations that occurred during the 36 months following release from prison. Because arrests are events that are closer to former pris- oners’ behavior than conviction or incarceration, we decided to use arrest data to construct a dichotomous measure of desistance: 0 = at least one arrest within 36 months of release and 1 = no arrests within 36 months of release. The longitudinal design produced by merging the RIS and follow-up data allows us to specify and examine properly relationships among key vari- ables in the theory of desistance we have outlined.

The RIS questionnaire was lengthy and very detailed. Because 50% of the respondents had 11 years of education or less, it may have required knowl- edge and skills beyond those of many of the respondents. Field notes from researchers indicate that poor reading skills caused some respondents to code multiple responses to some items and that many were confused about some questions. We considered two strategies for handling missing data produced by these and other problems: imputation and available case analysis. Impu- tation procedures treat substituted values as real data in statistical analysis. The resulting tests of significance for filled in data are underestimated, partic- ularly when errors are large (Little and Rubin, 1989). Although imputation

Page 7: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 95

is preferred when estimating the mean of a variable, it is generally regarded as inappropriate for estimating regression coefficients or probit and logit models (Dubin and Rivers, 1989; Little and Rubin, 1989; Hosmer and Lemeshow, 1989). Available case analysis, on the other hand, uses only cases without missing values. When the complete cases can be considered a random sub- sample of the original sample, subsequent analyses can be considered unbi- ased. If missing values are pervasive, however, statistical estimates may not be representative of the original sample (Little and Rubin, 1987). Both impu- tation and available case analysis require analytic compromises. Because ordinary least squares (OLS) and logistic regression techniques are used in this analysis, we decided in favor of using the unbiased statistical estimates provided by listwise deletion of cases with missing data.*

As a first step in exploring the links between age and desistance, we created measures of past success at straight and criminal pursuits, straight and crimi- nal expectations, and the estimated legal risk of criminal participation. The variable straight success represents the offender’s success at legitimate pur- suits in the immediate preimprisonment period, and the variables criminal financial success and confinement avoidance measure the offender’s past suc- cess at crime. Although Rand investigators did not collect data on the post- release employment and social circumstances of cohort members, data are available on their preimprisonment marital status. We included it as a control variable in the analysis. Race, level of education, and past legitimate employ- ment (work) also were included as controls. The operationalization of all variables is described in the Appendix.

ANALYSIS AND RESULTS

Our primary analytic objective is to explore the viability and causal nature of the proposed relationships in Figure 1. Formal methods of statistical mod- eling, such as covariance analysis, require complete data and exactly specified a priori substantive and measurement theory. When these requirements are not met, models cannot be estimated reliably. Therefore, to achieve our pri- mary goal, the major theoretical variables in the desistance process are expli- cated individually using regression analysis. The findings then may be used to refine the model and as a guide for future investigations.

We begin by determining the predictors of straight and criminal expecta- tions. The OLS regression results in Table 1 reveal age to be the only signifi- cant predictor of straight expectations (Beta = --.loo). As predicted,

2. The percentage of data missing for each variable is: marital status (0.6%). race (0.6%), education (0.5%), work (1.2%), criminal success (10.7%). straight success (5.3%), confinement avoidance (4.2%). straight expectations (8.6%), criminal expectations (8.9%), risk (8.0%). age (0%) and desistance (0%).

Page 8: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

96 SHOVER AND THOMPSON

Table 1 . Standardized Estimates from OLS Regression for Straight Expectations, Criminal Expectations, and Risk

Independent Variable

Criminal Expectations

Straight Expectations

Straight Success

Criminal Financial

Confinement Avoidance

Success

Marital Status

Race

Education

Work

RZ N

Dependent Variable

Straight Expectations

.047 (*0003)

(- .OOol)

(-.164)

- .052

-.041

.034 (.241)

-. loo* (-041)

(- .204) - .034

.066 (.352)

--.012 (- .077)

.02

71 1

Criminal Expectations Risk

- .033 (- .0002)

( . o o w .087*

(.396)

- .045 (- .364)

- .134***

.162***

(- .062)

- .065 (- .436)

.050 (.298)

- .063 (- .477)

.08

709

- .089* (- .054)

.139*** (.094) .045

(-00w

(-.oooo)

(-006)

-.018

.002

- .047 (- .226)

.04 1 (.011)

.105** (.417)

(.361) .101**

- .007 (- .033)

.06

663

NOTE: Standardized regression coefficients shown with unstandardized coefficients in paren theses.

* p < .05 * + p < .01

* * * p < ,001

increasing age dampens optimism for achieving success via legitimate pur- suits. Table l also includes the regression results for criminal expectations. It shows that age (Beta = -. 134), criminal financial success (Beta = .162),

Page 9: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 97

and confinement avoidance (Beta = .087) are significant predictors of crimi- nal expectations. These statistical relationships are in the predicted direc- tions, thus confirming the negative effect of age on expectations of criminal success. They also confirm that success at crime increases criminal expectations.

Table 1 also shows the predictors of risk. Risk increases as straight expec- tations increase (Beta = .139), and it decreases as criminal expectations increase (Beta = -.089). Increases in risk are associated also with increas- ing education (Beta = .101) and with being white (Beta = .105). All of these relationships are in the predicted direction. Contrary to predictions, we did not find age to be a significant predictor of risk. The findings, however, are consistent with the suggestion by Maguire (1982) and Shover (1985) that the effect of aging on risk is mediated by differential expectations. It is also possi- ble that this finding is an artifact of studying an already incarcerated cohort with little variation in risk levels.

The logistic regression findings in Table 2 indicate that age (b = .026), education (b = .388), criminal expectations (b = - .OH), and confinement avoidance (b = -.295) significantly predict desistance. The odds of desis- tance increase with age and education, and they decrease as confinement avoidance and criminal expectations increase.3

DISCUSSION AND CONCLUSIONS

Nearly all of the relationships reported here are in the predicted direction, and they lend support to the model of crime desistance developed from ethno- graphic research. Moreover, they are consistent generally with reports by other investigators. Using a longitudinal research design and a sample of serious offenders similar to those of this study, Piliavin et al. (1986: 1 18) show that the effect of age on renewed criminal participation is mediated by offend- ers’ belief that expected earnings from crime “are greater than or equal to expected earnings from a straight job.” We find that the probability of desis- tance from criminal participation increases as expectations for achieving friends, money, autonomy, and happiness via crime decrease (criminal expec- tations b = - .OH) . Piliavin et al. (1986) further show that self-reported

3. In a separate logistic regression analysis, we operationalized desistance as 0 = at least one criminal conviction in the 36 months following release from prison and 1 = no convictions. The results using this measure are substantially similar. The principal differ- ence was the absence of a significant relationship between confinement avoidance and desis- tance. This suggests that past success at avoiding confinement may increase the resolve of prosecutors, probation investigators, and judges to see that defendants pay for their mis- deeds by imprisonment thereby reducing to insignificance the advantage it affords offenders at the front end (arrest) stage of the criminal process.

Page 10: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

98 SHOVER AND THOMPSON

Table 2. Logistic Regression Predicting Desistance

Independent Variable b

Work

Education

Marital Status

Race

Criminal Expectations

Straight Expectations

Criminal Financial Success

Confinement Avoidance

Straight Success

Risk

Intercept

.348 (.239) .388* (.155) .026* (.012) - .37 1 (.212) .oo 1

(. 172)

(.027) .018 (.029) - .oooo2

- .064*

(.477)

(.121)

(.oO(-W

(.043)

(.779)

- .295*

-.OoO1

- .079

--.014

NOTE: Unstandardized regression coefficients shown with standard errors in parentheses. -2 log likelihood = 839.798; df = 651: p = .oooO; N = 663. Correctly predicted 63.8%. McFadden’s R 2 = .04. *2.0 times the standard error.

desistance is not affected by offenders’ average monthly income from legiti- mate employment in the year preceding incarceration. Our findings are similar.

Failure to find a link between perceived legal risk and desistance is consis- tent with Piliavin et al. (1986) but contrary to our prediction. One possible explanation for this may lie in a distinction between distal measures of risk

Page 11: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 99

perception and risk perceptions occumng in the immediate context of deci- sion-making about specific criminal opportunities. It is evident increasingly that measures of the latter are better predictors than more remote ones. Piliavin et al. (1986) show, for example, that risk perceptions fluctuate “sub- stantially” over a nine-month period of time. Risk perceptions, they suggest, “may be conditioned by elements within the immediate situation confronting the individual . . . [such that] perceptions of the opportunity, returns, and support for crime within a given situation may influence . . . perceptions of risks and the extent to which those risks are discounted’’ (1986: 1 15). Others have demonstrated the potential merit of a situational approach to risk esti- mates (e.g., Ekland-Olson et al., 1984; Rankin and Wells, 1983; Shover and Honaker, 1991).

Piliavin et al. (1986) failed to find a predicted positive relationship between straight expectations and crime desistance. So did we; straight expectations fail to make a difference in postrelease desistance. Although we can only speculate as to the reasons for this, we emphasize that our measures of differ- ential expectations are distal measures. We cannot rule out the possibility that in the RIS sample straight expectations more proximate to subjects’ actual criminal decision-making situations constrained their chances of desistance.

The fact that the RIS sample consists of prison inmates, many of them recidivists, means we cannot determine how much their behavior reflects innate differences in decision-making styles or experiential effects, that is, the effects of past success in committing crime and avoiding arrest (Gottfredson and Hirschi, 1990; Nagin and Paternoster, 1991). This may explain why we do not find a direct relationship between age and risk. The lack of compara- tive data on never-incarcerated offenders makes it impossible to examine the merits of these interpretations. More important, it could be argued that the criminal calculus and behavior of RIS subjects, precisely because they had demonstrated a willingness to commit serious crimes and had done so suc- cessfully in the past, limit the external validity of these findings. Generaliza- tions beyond the study population must be made, therefore, with caution. Data and sample limitations notwithstanding, the findings support theoretical explanations that emphasize both direct and indirect links between age and desistance.

REFERENCES

Adler, Patricia 1985 Wheeling and Dealing. New York: Columbia University Press.

Akers, Ronald L. 1985 Deviant Behavior: A Social Learning Approach. 3d ed. Belmont, Calif.:

Wadsworth.

Page 12: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

100 SHOVER AND THOMPSON

Blumstein, Alfred, Jacqueline Cohen, and David P. Famngton

Cromwell, Paul F., James N. Olson, and DAunn Wester Avary

1988 Criminal career research: Its value for criminology. Criminology 26:l-36.

1991 Breaking and Entering: An Ethnographic Analysis of Burglary. Newbury Park, Calif.: Sage.

Cusson, Maurice and Pierre Pinsonneault The decision to give up crime. In Derek B. Cornish and Ronald V. Clarke (eds.), The Reasoning Criminal. New York: Springer-Verlag.

Selection bias in linear regression, logit and probit models. Sociological Methods and Research 18:360-390.

1986

Dubin, Jeffrey A. and Douglas Rivers 1989

Ekland4lson, Sheldon, John Lieb, and Louis Zurcher 1984 The paradoxical impact of criminal sanctions: Some microstructural

findings. Law & Society Review 18:159-178.

Famngton, David P. 1986 Age and crime. In Michael Tonry and Norval Morns (eds.), Crime and

Justice: An Annual Review of Research. Vol. 7. Chicago: University of Chicago Press.

Gartner, Rosemary and Irving Piliavin The aging offender and the aged offender. In P.B. Baltes, D.L. Featheman, and R.M. Lerner (eds.), Life-Span Development and Behavior. Vol. 9. Hillsdale, N.J.: Lawrence Erlbaum.

1988

Glaser, Daniel 1964

1980

Effectiveness of a Prison and Parole System Indianapolis, Ind.: Bobbs-Merrill. The interplay of theory, issues, policy, and data. In Malcolm Klein and Katherine Tielmann (eds.), Handbook of Criminal Justice Evaluation. Beverly Hills, Calif. : Sage.

A note on the deterrent effect of juvenile vs. adult jurisdiction. Social Problems 3 1 :2 19-22 1 .

Glassner, Barry, Margaret Ksander, Bruce Berg, and Bruce D. Johnson 1983

Glueck, Sheldon and Eleanor Glueck 1937 Later Criminal Careers. New York: Commonwealth Fund.

Gottfredson, Michael R. and Travis Hirschi

Gove, Walter R.

1990 A General Theory of Crime. Stanford, Calif.: Stanford University Press.

The effect of age and gender on deviant behavior: A biopsychosocial perspective. In Alice S. Rossi (ed.), Gender and the Life Course. New York: Aldine.

1985

Grasmick, Harold G. and Robert J. Bursik, Jr. 1990 Conscicnce, significant others, and rational choice: Extending the deterrence

model. Law & Society Review 24:837-861.

Grasmick, Harold G. and Herman Milligan, Jr. 1976 Deterrence theory approach to socioeconornic/demographic correlates of

crime. Social Science Quarterly 57:608-617.

Page 13: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 101

Hirschi, Travis 1986 On the compatibility of rational choice and social control theories of crime.

In Derek B. Cornish and Ronald V. Clarke (eds.), The Reasoning Criminal. New York: Springer-Verlag.

Hosmer, David and Stanley Lemeshow

Irwin, John

Klein, Steven P. and Michael N. Caggiano

1989 Applied Logistic Regression. New York: John Wiley Bi Sons.

1970 The Felon. Englewood Cliffs, N.J.: Prentice-Hall.

1986 The Prevalence, Predictability, and Policy Implications of Recidivism. Santa Monica, Calif.: Rand.

Statistical Analysis with Missing Data. New York: John Wiley & Sons. The analysis of social science data with missing values. Sociological Methods and Research 18:293-325.

Little, Roderick and Donald Rubin 1987 1989

Maguire, Mike, in collaboration with Trevor Bennett

Meisenhelder, Thomas N. 1982 Burglary in a Dwelling. London: Heinemann.

1977 An exploratory study of exiting from criminal careers. Criminology 153 19-334.

Nagin, Daniel S. and Raymond Paternoster 1991 On the relationship of past to future participation in delinquency. Criminol-

ogy 29:163-189.

Ohlin, Lloyd, Alfred Blumstein, Kenneth Adams, Douglas Anglin, Arnold Barnett, Robert Boruch, Peter Greenwood, Albert Reiss, and Lawrence Sherman 1988 Final report of the desistance-persistence working Group Program on

Human Development and Criminal Behavior. Castine Research Corpora- tion, Maine. Photocopy.

Petersilia, Joan 1980 Criminal career research: A review of recent evidence. In Norval Moms

and Michael Tonry (eds.), Crime and Justice: An Annual Review of Research. Vol. 2. Chicago: University of Chicago Press.

Survey of Prison and Jail Inmates: Background and Method. Santa Monica, Calif.: Rand.

Peterson, Mark, Jan Chaiken, Patricia Ebener, and Paul Honig 1982

Piliavin, Irving, Rosemary Gartner, Craig Thornton, and Ross Matsueda 1986 Crime, deterrence, and rational choice. American Sociological Review

51:101-119.

Rankin, Joseph H. and L. Edward Wells

Reitzes, Dietrich C. 1983

1955

The social context of deterrence. Sociology and Social Research 67:18-39.

The effect of social environment upon former felons. Journal of Criminal Law, Criminology and Police Science 46:22&231.

Crime and deviance over the life course: The salience of adult social bonds. American Sociological Review 55:602-627.

Sampson, Robert J. and John H. Laub 1990

Page 14: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

SHOVER AND THOMPSON

Shover, Neal 1973 1983

1985

1991

The social organization of burglary. Social Problems 20:499-5 14. The later stages of ordinary property offender careers. Social Problems

Aging Criminals: Beverly Hills, Calif.: Sage.

The socially bounded decision making of persistent property offenders. Knoxville: University of Tennessee, Department of Sociology. Photocopy.

Sanctions and Social Deviance: The Question of Deterrence. New York: Praeger.

3 1 :208-2 18.

Shover, Neal and David Honaker

Tittle, Charles R. 1980

Walsh, Dermot 1980 Break-Ins: Burglary from Private Houses. London: Constable. 1986 Heavy Business. London: Routledge & Kegan Paul.

1978 West, W. Gordon

The short term careers of serious thieves. Canadian Journal of Criminology 20~169-190.

Wilson, James Q. and Richard J. Hermstein 1985 Crime and Human Nature. New York: Simon & Schuster.

Neal Shover is Professor of Sociology at the University of Tennessee. His current research explores the social psychology of criminal careers, criminal decision making, and corporate crime.

Carol Y. Thompson is Assistant Professor of Sociology at East Carolina University. Her current research examines the relationship between community context and property crime victimization.

Page 15: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

CRIME DESISTANCE 103

APPENDIX OPERATIONALIZATION OF VARIABLES

Age

Race

Work

Marital Status

Education

Desistance

Criminal Financial Success

Confinement Avoidance

Straight Success

Criminal Expectations

Response to the question: How old were you on your last birthday? 1 = white 0 = other 0 = no jobs during last period of time in free

community 1 = one or more jobs 0 = not mamed 1 = married 0 = 6th grade or less 1 = 7th through 11th grade 2 = high school or more 0 = at least one arrest within 36 months of

1 = no arrests within 36 months of release Self-reported average monthly income from crime during last period of time in the free community Total number of adult felony convictions/total number of prison terms ever served (including the current term), plus one-half the total number of jail terms ever served Self-reported average monthly income from legitimate employment during last period of time in the free community Summated index (alpha = .67) created from the matrix question, What are the chances each of these things would happen to you from doing crimes?: Having friends, Being my own man, Having a lot of money, and Being happy. Response options were 1 = no chance, 2 = low chance, 3 = even chance, 4 = high chance, 5 = certain.

release

Page 16: AGE, DIFFERENTIAL EXPECTATIONS, AND CRIME DESISTANCE

104 SHOVER AND THOMPSON

Straight Expectations Summated index (alpha = .67) created from the matrix question, What are the chances each of these things would happen to you if you did not do crimes?: Having friends, Being my own man, Having a lot of money, and Being happy. Response options were 1 = no chance, 2 = low chance, 3 = even chance, 4 = high chance, 5 = certain.

Summated index (alpha = .76) created from the matrix question, What are the chances each of these things would happen to you from doing crimes?: Getting arrested, and Going to prison for years. Response options were 1 = no chance, 2 = low chance, 3 = even chance, 4 = high chance, 5 = certain.

Risk