Maturity of Judgment and Perceptual Deterrence

Abstract

Recent deterrence literature has found that the degree to which sanction threats are perceived to influence subsequent offending differs within individuals and between individuals over time. This study examines whether three psychosocial aspects (temperance, perspective, responsibility) relevant to the maturity of judgment predict within-individual and between-individual differences in levels of perceptual deterrence. Random effects regression models with fixed effects (hybrid models) are used to estimate the impact of maturity of judgment on the perceived risks, costs, and benefits of crime among a sample of serious juvenile offenders from the Pathways to Desistance study over 7 years of development. The results support both within-person effects and between-person effects. More mature judgment ability is generally associated with the perception of greater risks, heavier costs of punishment, and fewer rewards of crime. The rate of change in perceptual deterrence by maturity of judgment varies between individuals. Implications of the findings are discussed.

Keywords

perceptual deterrence rational choice psychological development

Both deterrence theory and rational choice theory assume potential offenders take into consideration the detection risk, as well as the costs and benefits associated with offending before they decide to commit crime. Deterrence theory argues that legal punishment inhibits future crime by scaring off would-be offenders (Beccaria, 1764/2009). The central premise of this process is that individuals fear sanction threats, meaning greater levels of certainty, severity, and celerity of punishment will lead to fewer crimes. In a similar vein, rational choice theorists argue that individuals make decisions to offend when the benefits of crime outweigh the costs of detection and punishment (Clarke & Cornish, 1985; Nagin & Paternoster, 1993). From their perspective, crime is likely to occur when $U (Benefits) - P [U (Costs)] > 0$ , where U(Benefits) are standardized units of benefits of crime, U(Costs) are standardized costs of crime, and P is the perceived probability of detection (Loughran, Paternoster, Piquero, & Pogarsky, 2011). This approach to explaining criminal decision making offers a seductively simple answer to the question of how to control crime. Specifically, harsher punishments should, all else equal, reduce crime rates at the aggregate level. For this reason, deterrence and rational choice frameworks have promoted “get tough” policies that emphasize the costs of criminal behavior (Cullen, Pratt, Miceli, & Moon, 2002).

One of the caveats in this framework is that it does not explicitly account for the possibility that individuals perceive the rational choice components of crime in a heterogeneous way. For example, given the same sanction, there may be variation in the level of perceived threat. Similarly, an individual may show a change in the perception of criminal reward at different time points. This is an important point to recognize because subjective perceptions of risks, costs, and rewards of crime must intervene between sanctions and criminal action (Decker, Wright, & Logie, 1993; Paternoster, 2010). Yet, early works on deterrence theory paid scant attention to psychological processes (Nagin, 1998; Zimring & Hawkins, 1968). Consequently, it is worth exploring whether youths’ psychological development is associated with variation in perceptual deterrence, including the perceptions of sanction risk, costs, and benefits of crime across and within individuals.

Framework

Recent deterrence research has focused on explaining two types of heterogeneity in perceptual deterrence: within-individual heterogeneity and between-individual heterogeneity. Within-individual heterogeneity notes that the same individual may perceive deterrence factors differently across time or situations. Such differences are theorized to stem from developmental or situational factors. Using longitudinal techniques, deterrence scholars have found evidence supporting within-person changes in perceptions over time (Loughran, Piquero, Fagan, & Mulvey, 2012; Pogarsky, Kim, & Paternoster, 2005). Also, situational factors such as the ambiguity in punishment certainty and the presence of delinquent peers have been found to influence offending probabilities (Loughran et al., 2011; Matthews & Agnew, 2008).

Between-individual heterogeneity relates to the efficacy of sanction threats across individuals at the same time point. This line of research mostly relies on cross-sectional studies and seeks to categorize individuals into groups based on their sensitivity to sanction threats (Loughran, Piquero, et al., 2012; Nagin & Paternoster, 1993; Parker & Grasmick, 1979; Pogarsky, 2002; Zimring & Hawkins, 1968). The categorization conceptualizes the continuum of criminal proclivity, and those in the middle of the continuum are labeled “deterrable” offenders, whereas those at both ends are termed “marginal” offenders (Jacobs, 2010; Zimring & Hawkins, 1968). The source of criminal proclivity is linked to personal traits such as gender, self-control, moral inhibition, and impulsivity (Pauwels, Weerman, Bruinsma, & Bernasco, 2011; Pogarsky, 2002; Schulz, 2014). Scholars have elaborated the concept with a three-group categorization (acute conformist, incorrigible, deterrable) or proposed the nonlinearity of the certainty–sanction relationship called the “tipping effect” (i.e., the detection probability has a minimum threshold that, once crossed, will have a deterrent effect; Loughran, Pogarsky, Piquero, & Paternoster, 2012; Pogarsky, 2002). The logic that deterrence-based policies have effects only for the deterrable group has motivated the identification of offending profiles for each type of offense.

In criminology, perceptual deterrence framework might integrate two perspectives that have been deemed parallel with regard to their views on the sources of crime: theories of crime and theories of criminality (Nagin & Paternoster, 1993, 1994; Pogarsky, 2002). Theories of crime assume individuals are equally motivated toward crime, but such motivation is differently triggered across environmental or social settings (Akers, 1985; Cloward & Ohlin, 1960). In contrast, theories of criminality argue that the propensity for crime is established early in life and varies from person to person (Glueck & Glueck, 1968; Gottfredson & Hirschi, 1990; Rowe, Osgood, & Nicewander, 1990). Perceptual deterrence theorists acknowledge that individuals make rational choices by comparing costs and benefits associated with offending. Thus, offending is viewed as a form of situational decision making rather than a manifestation of invariant pathology. At the same time, they posit that the sensitivity to sanction threats varies across individuals such that some individuals are more prone to offending than others because of their psychological traits as well as developmental and situational factors. In this sense, perceptual deterrence theory suggests that the two criminological frameworks are not wholly incompatible.

But the available literature on perceptual deterrence needs a few qualifications. Most studies emphasized sanction risks and punishment costs but neglected criminal rewards (Nagin, 1998; Pratt, Cullen, Blevins, Daigle, & Madensen, 2006, but see Decker et al., 1993). For this reason, critics argue that deterrence theory and rational choice theory justify punishment-oriented or cost-increasing correctional programs, which have been shown ineffective in reducing recidivism (Cullen et al., 2002). A few recent studies, however, have renewed interest in understanding the benefits of crime by identifying their significant association with criminal offending (Baker & Piquero, 2010). At the same time, studies supporting between-individual heterogeneity tend to classify individuals into deterrable and undeterrable groups and argue that deterrence policies only have effects for deterrable offenders. These studies imply that personal traits (i.e., deterrability) are unchangeable. However, this conclusion is troubling not only because no study has confirmed that invariability in sensitivity to sanction threats but also because it does not provide any insight for undeterrable offenders.

Furthermore, the assumption that individual traits endure across the life course is challenged by accumulating evidence in psychology and developmental neuroscience. Developmental theorists argue that criminal offending is a normative process. They posit that both involvement and desistance in illegal conduct are influenced by psychological (mal)functions. In their view, mid-adolescence is a vulnerable period for risky and reckless behavior because cognitive control is not developed enough to regulate a dramatic remodeling of the brain’s socioemotional system (Steinberg, 2008). The brain’s cognitive development, including the capacity for autonomy and self-regulation, peaks during the late teenage years and stabilizes in early adulthood, resulting in decreased risk-taking behavior (Fagan & Piquero, 2007; Moffitt, 1990, 1993; Paus, 2005). These findings suggest psychological development affects one’s likelihood of engaging in antisocial behavior.

Developmental psychologists have repeatedly shown that psychological traits such as sensation seeking, impulsivity, future orientation, and delay discounting follow developmental patterns across age, at least during adolescence (Steinberg et al., 2008; Steinberg et al., 2009). Steinberg and Cauffman (1996) suggested that a list of psychosocial (cognitive, noncognitive) factors are relevant to the “maturity of judgment,” or good decision making, and these factors converge on one of the three dispositions: temperance, perspective, and responsibility. According to the authors, temperance is the ability to limit impulsivity, avoid extremes in decision making, evaluate a situation thoroughly before acting, and seek appropriate advice from others. Perspective means the capacity to acknowledge the complexity of a situation and to frame a specific decision within a larger context. Responsibility refers to healthy autonomy, self-reliance, and clarity of identity. Empirical evidence shows that youths with higher scores in the above dimensions are less likely to engage in antisocial behavior in hypothetical scenarios regardless of age, indicating that maturity of judgment plays a role in criminal decision making (Cauffman & Steinberg, 2000).

In sum, research in psychology and developmental neuroscience indicates that individual traits, including those that affect good decision making, develop over time. Taken together with prior literature on perceptual deterrence, these findings suggest that influences on decision-making skills that affect offending may be heterogeneous among individuals and across time within the same individual. In addition, they indicate that heterogeneity in decision-making skills may be influenced by individual psychosocial development.

Hypotheses

This study tests whether variation in three domains of maturity of judgment—originally proposed by Steinberg and Cauffman (1996)—influences the perception of risks, costs, and rewards associated with offending. We will examine whether the impact of maturity of judgment on offending perceptions varies within individuals and across individuals. In particular, three research hypotheses will be tested:

Hypothesis 1: Within-individual differences in perceptual deterrence are affected by changes in levels of maturity of judgment over time after controlling for confounding influences.

Hypothesis 2: The rate of change in perceptual deterrence by the change in maturity of judgment will differ between individuals.

Hypothesis 3: The effect of within-person changes in maturity of judgment on perceptual deterrence will be contingent upon delinquent peer influence, prior offending, and age.

Method

Data

This study uses data from the Pathways to Desistance (Pathways) study, which is a longitudinal investigation of continuity and change in antisocial behavior among high-risk juvenile offenders (Mulvey, 2016). Between November 2000 and January 2003, the Pathways study drew from court files and recruited a sample of 1,354 adjudicated adolescents who were aged between 14 and 17 years at the time of committing an offense (predominantly felonies). The sample mostly consists of males (86%), and the largest racial/ethnic group is Black (41%), followed by Hispanic (34%) and White and Other groups combined (25%). The participants were drawn from two study sites: Maricopa County (Phoenix), Arizona (n = 654) and Philadelphia County, Pennsylvania (n = 700; Schubert et al., 2004). Data collection occurred at 11 time points (baseline, 6, 12, 18, 24, 30, 36, 48, 60, 72, and 84 months) during the 7-year field period between March 2003 and March 2010. The Pathways data include a variety of measures that tap into important constructs such as the adolescent’s psychological development, social relationships, routine activities, and experiences with the justice system (Mulvey et al., 2004).

We analyze all 11 waves of data, which meant that study attrition and item nonresponse were a concern. To impute missing values, we relied on BLIMP, which is a software program designed to perform multiple imputation via fully conditional specification (FCS; Keller & Enders, 2018). Multilevel imputation in BLIMP is more appropriate than alternatives such as listwise deletion or single-level multiple imputation because it accommodates analysis models with random intercepts and random slopes (Graham, 2009; Keller & Enders, 2018). Missing values were imputed for all variables used in the analysis. The proportion of missing values among these variables was generally ≤10%. Prior research suggests 20 imputations for missing data in this range (Deryol, Wilcox, & Dolu, 2017; Rubin, 1987; Schafer & Olsen, 1998), and thus, 20 imputed data sets were generated in BLIMP. We relied on the MMI_ANALYZE in SAS 9.4 to obtain the pooled results for the multilevel analyses (Mistler, 2013; SAS Institute, 2011).

Measures

Dependent Variables: Perceptions of Punishment Risk, Perceptions of Punishment Costs, and Perceptions of Rewards From Crime

To tap into different dimensions of perceptual deterrence, the present study employs five dependent variables, which measure respondents’ perceptions of certainty of punishment, personal costs of punishment, social costs of punishment, personal rewards of crime, and social rewards of crime (based on Nagin and Paternoster, 1994). There were a total of 78 items available to construct nine composite measures, whose reliability coefficients range between .68 (for social costs of punishment) and .99 (for personal costs of punishment). These measures were then used to construct the five scales listed above. Below we provide more detail on each of these five scales.¹

Perceptions of certainty of punishment

This study averages the scores of certainty of your punishment and certainty of others’ punishment to construct certainty of punishment.² Certainty of your punishment and certainty of others’ punishment are, respectively, computed as the means of seven items that asked about the perceived likelihood of detection and punishment for several types of offenses that respondents or others had committed (e.g., “How likely is it that kids in your neighborhood would be caught and arrested for fighting?”). Each item is measured on an 11-point Likert-type scale, ranging from 0 (no chance) to 10 (absolutely certain to be caught). Smaller values represent a lower perception of certainty of punishment.

Perceptions of personal costs of punishment

The Pathways data include 18 items that ask whether the respondent experienced personal costs as a result of punishment (e.g., “Has your court sentence kept you from hanging out with your friends as much as you used to?”). The 18 items are answered on a 4-point Likert-type scale ranging between 1 (not at all) and 4 (very much), with higher values indicating greater costs. Two scales are created based on the responses to the 18 items. The first is a variety score that represents the number of items endorsed by the respondent. The second is a weight score that reflects the mean of endorsed items. The perception of personal costs is constructed by averaging the product of variety score and weight score of 18 items. The construct ranges from 0 to 4, and smaller values signify the perception of fewer personal costs of punishment.

Perceptions of social costs of punishment

The measure of social costs of punishment is computed by averaging the scores of six items (e.g., “If the police catch me doing something that breaks the law, how likely is it that I would be suspended from school?”). Each item is measured on a 5-point Likert-type scale, ranging from 1 (very unlikely) to 5 (very likely). Smaller values represent fewer perceived costs of punishment.

Perceptions of personal rewards of crime

The perception of personal rewards is calculated by averaging the scores of seven items which ask about the fun the respondent would have as a result of that act (e.g., “How much ‘thrill’ or ‘rush’ is it to break into a store or home?”). The responses were measured on an 11-point Likert-type scale from 0 (no fun or kick at all) to 10 (a great deal of fun or kick). Smaller values refer to fewer personal rewards of crime.

Perceptions of social rewards of crime

The social rewards of crime were measured by averaging three composite scores, consisting of 15 items, respectively, that ask about what the respondent may gain socially in three types of offense scenarios (stealing, fighting, robbery; e.g., “If I take things, other people my age will respect me more.”). Responses fall between 1 (strongly disagree) and 4 (strongly agree). Smaller values indicate fewer social rewards of crime.

Key Independent Variables: Maturity of Judgment

The present study adopts the Maturity of Judgment (MOJ) scale, which was originally developed by the Pathways researchers. The MOJ scale is computed by combining standardized scores from the Psychosocial Maturity Inventory (PSMI), Future Outlook Inventory (FOI), Weinberger Adjustment Inventory (WAI), and Resistance to Peer Influence (RPI) scale (Cauffman & Steinberg, 2000). The four psychometric scales have shown strong validity and reliability in prior research (Schubert, Mulvey, & Pitzer, 2016). The MOJ scale provides a multifaceted psychosocial maturity score by tapping into the three domains previously discussed: temperance, perspective, and responsibility. The current study followed instructions drawn from the Pathways study website to compute each dimension score.³

Temperance

This study operationalizes temperance by averaging standardized scores of two subscales of the WAI: suppression of aggression and impulse control. Participants were asked how much their behavior in the past 6 months agreed with statements tapping suppression of aggression (e.g. “People who get me angry better watch out.”) and impulse control (e.g., “I say the first thing that comes into my mind without thinking enough about it.”). Participants answered on a 5-point Likert-type scale (1 = false to 5 = true). The suppression of aggression scale comprises seven items, and the impulse control scale is measured with eight items. Each item is reverse-coded so that larger values represent more positive temperance (i.e., greater suppression of aggression and impulse control).

Perspective

Perspective is computed by averaging standardized scores of the total FOI and the consideration of others subscale from the WAI. The total FOI is measured by eight items that asked participants the degree to which each statement reflected their outlook (e.g., “I will keep working at difficult, boring tasks if I know they will help me get ahead later.”). Responses are on a 4-point Likert-type scale (1 = never true to 4 = always true). The consideration of other scales from the WAI asked participants how much their behavior in the past 6 months matched seven statements (e.g., “Doing things to help other people is more important to me than almost anything else.”), and responses were coded on a scale from 1 (false) to 5 (true). Larger scores indicate a more positive perspective (a greater degree of future consideration and planning and a greater consideration for others).

Responsibility

This study measures responsibility by averaging the standardized scores of the total PSMI and the RPI. The total PSMI score is the mean of 30 items in the PSMI (Form D) developed by Greenberger, Josselson, Knerr, and Knerr (1975). Items reflect self-reliance (e.g., “Luck decides most things that happen to me.”), identity (e.g., “I change the way I feel and act so often that I sometimes wonder who the ‘real’ me is.”), and work orientation (e.g., “I hate to admit it, but I give up on my work when things go wrong.”). Participants responded on a 4-point Likert-type scale ranging from 1 (strongly agree) to 4 (strongly disagree). All items are reverse-coded so that larger scores indicate more responsible behavior.

The RPI measure was developed by Steinberg for the Pathways study in 2000 to assess how much adolescents act autonomously in interactions with their peer group. Participants were presented with 10 different sets of scenarios, each representing a unique dimension of peer influence: going along with friends, fitting in with friends, changing your mind, knowingly doing something wrong, hiding true opinion, breaking the law, changing the way you usually act, taking risks, saying things you do not really believe, and going against the crowd. For each set of scenarios, participants were asked to choose one of two scenarios that more closely reflected their behavior (e.g., “Some people go along with their friends just to keep their friends happy.” or “Other people refuse to go along with what their friends want to do, even though they know it will make their friends unhappy.”). For the chosen scenario, participants rated how accurate the statement was on a 4-point Likert-type scale ranging from 1 (it’s really true I’m influenced by my peers) to 4 (it’s really true I prefer to be an individual). The RPI score is the mean across 10 dimensions. Larger scores represent more autonomous behavior against peer influence.

Control Variables

This study controlled for the following covariates: age at interview, prior aggressive offending and income-generating offending, delinquent peer influence, gender, and race/ethnicity. Age is controlled because it is found to have direct influence on psychological traits such as sensation seeking, impulsivity, future orientation, and delay discounting (Steinberg et al., 2008; Steinberg et al., 2009). Prior offending (aggressive and income-generating offending) is controlled because criminal involvement itself can affect the perception of risks, costs, and rewards of crime (Anwar & Loughran, 2011; Pogarsky et al., 2005; Pogarsky & Piquero, 2003). For each type of offense, this study uses the standardized score of a cumulative number of offending incidents the respondent committed up until the focal interview.

Delinquent peer influence may affect perceptual deterrence factors regardless of the respondent’s psychological propensity. The delinquent peer influence measure is computed by averaging the standardized scores of two dimensions of Peer Delinquent Behavior items in the Pathways study: antisocial behavior and antisocial influence (modeled in part on the Rochester Youth Study; Thornberry, Lizotte, Krohn, Farnworth, & Jang, 1994). Antisocial behavior is the mean rating of the prevalence of friends who engage in 12 delinquent behaviors (e.g., “How many of your friends have sold drugs?”). Antisocial influence is the mean rating of the prevalence of friends who encourage the respondent to engage in any of seven delinquent activities (e.g., “How many of your friends have suggested that you should sell drugs?”). Responses are measured on a 5-point Likert-type scale ranging from “1” (none of them) to “5” (all of them). Larger values represent greater antisocial influence from peers.

In addition, this study follows previous literature in controlling for race/ethnicity because socioeconomic conditions that vary across racial/ethnic groups may affect psychosocial development (Loughran et al., 2011; Matthews & Agnew, 2008). Racial/ethnic groups are dummy coded with Black serving as the reference category against which White/Other and Hispanic categories are compared. Finally, gender is used as a control variable (0 = male, 1 = female) to account for any differences in perceptual deterrence between males and females (Steinberg, 2008). Descriptive statistics of variables used in this study are presented in Table 1.

Table 1:

Univariate Descriptives at Baseline

Variable	N	M/%	SD	Minimum	Maximum
Certainty of punishment	1,352	5.41	2.33	0	10
Personal costs of punishment	1,346	1.84	1.42	0	4
Social costs of punishment	1,353	2.73	0.86	1	5
Personal rewards of crime	1,353	2.36	2.42	0	10
Social rewards of crime	1,353	2.00	0.44	1	3.6
Temperance	1,351	0	0.88	−1.93	2.20
Perspective	1,346	0	0.84	−2.62	2.37
Responsibility	1,346	0	0.82	−2.69	1.95
Age	1,354	16.04	1.14	14	19
Delinquent peer influence	1,316	0	0.93	−1.16	3.34
Prior offending (standardized)
Aggressive	1,354	0	1	−0.32	20.26
Income-generating	1,354	0	1	−0.35	9.98
Gender	1,354
Male	1,170	86.41
Female	184	13.59
Ethnicity	1,354
White/Other	339	25.04
Black	561	41.43
Hispanic	454	33.53

Note. N = number of observations; M = mean.

Analytical Strategy

To model both within-individual heterogeneity and between-individual heterogeneity in perceptual deterrence, this study employs a hybrid statistical model that allows random and fixed effects components to be estimated simultaneously (Allison, 2009). Specifically, we estimate a multilevel model with a two-level hierarchical structure where repeated measurements (over time) are nested within persons. Time-varying covariates are decomposed into two components: person-specific means and deviation scores of observations from each person-specific mean (Hox, 2010). Level 1 variables are deviation scores of time-varying covariates for temperance, perspective, responsibility, delinquent peer influence, age, prior aggressive offending, and prior income-generating offending. Person-specific means and time-stable covariates, which correspond to between-person variation, are modeled at Level 2 along with the time-stable covariates gender and race/ethnicity.

There are several features of the hybrid model that make it well suited for the present study. First, the hybrid approach is advantageous as it can control for stable, unmeasured characteristics of the individuals. Second, this approach has an advantage over the fixed effects model in that it can provide estimates for time-stable covariates. Finally, this method can account for random variation in the slope parameters for the time-varying covariates (Allison, 2005). The analysis frees the variance in the coefficients for time-varying covariates.⁴ Then, a likelihood ratio test (D3) is used to compare the full model against a restricted model that does not contain each random slope.⁵

Results

All bivariate correlations between the independent variables were below .7 (Appendix A), and the maximum variance inflation factor (VIF; 1.352) was much lower than the conventional threshold of 4 (Piquero & Weisburd, 2010; Tabachnick & Fidell, 2001). Thus, there was little evidence of multicollinearity between the independent variables that will appear in our models.

The estimates obtained from the hybrid models are reported in Table 2. In each model, the person-specific mean of each time-varying covariate was included in the analysis to help obtain unbiased estimates of the effects of other time-invariant variables. However, the coefficients for the mean variables were not reported because they are likely confounded with the effects of other unobserved variables, making it difficult to provide a substantively meaningful interpretation for them (Allison, 2005).⁶ More importantly, the primary strength of longitudinal data is their ability to offer insight on variability over time, and therefore, we analyze the relationships of interest accordingly (Halaby, 2004).

Table 2:

Hybrid Models (Random Effects Models With Fixed Effects) Estimating the Impact of Maturity of Judgment on Perceptions of Risk

Variable	Model 1Certainty of punishment	Model 2Personal costs of punishment	Model 3Social costs of punishment	Model 4Personal rewards of crime	Model 5Social rewards of crime
Variable	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Intercept	7.573 (1.084)***	0.242 (0.658)	2.678 (0.380)***	1.985 (0.766)**	1.820 (0.206)***
Time-varying covariates^a
Temperance	0.248 (0.043)***	−0.116 (0.032)***	0.031 (0.016)*	−0.433 (0.035)***	−0.104 (0.009)***
Perspective	0.242 (0.038)***	0.062 (0.027)*	0.122 (0.016)***	−0.216 (0.029)***	−0.047 (0.008)***
Responsibility	−0.038 (0.045)	−0.024 (0.028)	−0.042 (0.017)*	−0.141 (0.037)***	−0.099 (0.009)***
Delinquent peer influence	−0.200 (0.031)***	0.125 (0.023)***	0.009 (0.012)	0.264 (0.028)***	0.072 (0.007)***
Aggressive offending	−0.118 (0.022)***	0.074 (0.018)***	−0.024 (0.009)**	0.111 (0.020)***	0.012 (0.005)**
Income-generating offending	−0.158 (0.026)***	0.124 (0.018)***	−0.027 (0.011)*	0.002 (0.024)	0.011 (0.005)*
Age	0.055 (0.011)***	0.062 (0.008)***	0.074 (0.004)***	−0.110 (0.009)***	−0.007 (0.002)**
Time-stable covariates
Gender (Reference: Male)	0.979 (0.068)***	−0.513 (0.066)***	0.137 (0.027)***	−0.555 (0.061)***	−0.104 (0.014)***
Race (Reference: Black)
White/Other	0.508 (0.075)***	−0.449 (0.055)***	0.274 (0.029)***	0.215 (0.065)***	−0.118 (0.015)***
Hispanic	0.087 (0.071)	−0.161 (0.052)**	0.090 (0.028)***	0.027 (0.060)	−0.091 (0.014)***
Overall model fit (D3)	42.445***	21.839***	30.343***	72.299***	75.310***
N_observations (N_individuals)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)

The person-specific mean of each time-varying covariate is also included in the model (not reported).

p ≤ .05. **p ≤ .01. ***p ≤ .001 (two-tailed test).

In Model 1, within-person changes in the perceived certainty of punishment were predicted by two domains (temperance, perspective) of maturity of judgment. Increases in temperance (i.e., better aggression suppression and impulse control) and perspective (i.e., deeper consideration of future and others) led to the perception of higher sanction threats ( $b_{T e m p e r a n c e} = 0.248$ , p ≤ .001; $b_{P e r s p e c t i v e} = 0.242$ , p ≤ .001). However, changes in the responsibility domain (i.e., self-reliance, identity, and work orientation) did not influence changes in perceptions of the certainty of punishment. Within-person changes in perceptions of certainty of punishment were also significantly affected by delinquent peer influence, both offending measures, and age. Increases in delinquent peer influence and prior offending experiences were associated with decreases in the certainty of punishment ( $b_{P e e r} = - 0.200$ , p ≤ .001; $b_{A g g r e s s i v e} = - 0.118$ , p ≤ .001; $b_{I n c o m e} = - 0.158$ , p ≤ .001). That is, having contact with friends who engaged in delinquent activities may have had criminogenic effects by lowering the focal respondent’s perception of detection risks. Prior engagement in offending lowered the perceived likelihood of punishment, which is consistent with previous research (Pogarsky et al., 2005). By contrast, age had a small positive effect on the perceived sanction threats ( $b_{A g e} = 0.055$ , p ≤ .001).

Between-person effects were estimated by testing whether the inclusion of random-effect variance of slope parameter improved the model fit. As Table 3 shows, the inclusion of the random slope for each time-varying covariate significantly increased the model fit (p ≤ .001). The findings indicate that there may be considerable differences among serious juvenile offenders in the rates of change in the certainty of punishment based on maturity of judgment, delinquent peer influence, and age. Other between-person variations were found regarding the respondent’s gender and race/ethnicity (Table 2). Females perceived greater certainty of punishment compared with males ( $b_{G e n d e r} = 0.979$ , p ≤ .001). Compared with Black respondents, Whites and other racial/ethnic groups reported higher sanction risks associated with offending, whereas Hispanics did not show significant difference ( $b_{W h i t e / O t h e r} = 0.508$ , p ≤ .001).

Table 3:

Increase in the Likelihood Ratio Test (D3) Statistics by the Inclusion of Random-Effect Variances

Variable	Model 1Certainty of punishment	Model 2Personal costs of punishment	Model 3Social costs of punishment	Model 4Personal rewards of crime	Model 5Social rewards of crime
Random-effect variances
Temperance	16.026***	10.109***	9.248***	19.439***	14.992***
Perspective	11.065***	8.073***	14.329***	3.616**	10.546***
Responsibility	14.617***	6.041***	8.936***	7.586***	8.351***
Delinquent peer influence	3.966***	2.911**	2.217*	4.585***	5.581***
Age	7.549***	8.504***	11.639***	2.893**	4.755***
Overall model fit (D3)	42.445***	21.839***	30.343***	72.299***	75.310***
N_observations (N_individuals)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)	14,894 (1,354)

p ≤ .05. **p ≤ .01. ***p ≤ .001 (two-tailed test).

In Model 2, within-person changes in the perceived personal costs of punishment were associated with two domains (temperance, perspective) of maturity of judgment ( $b_{T e m p e r a n c e} = - 0.116$ , p ≤ .001; $b_{P e r s p e c t i v e} = 0.062$ , p = .025). However, the effect of temperance seemed counterintuitive because the results indicate that those with positive temperance, which may signify more mature decision-making skills, were found to perceive fewer personal costs of punishment. The influence of delinquent peers was associated with the perception of higher personal costs of punishment ( $b_{P e e r} = 0.125$ , p ≤ .001), which may result from having more frequent direct or indirect experiences with personal costs imposed by punishment. The two prior offending measures had significant positive effects on the personal costs of punishment ( $b_{A g g r e s s i v e} = 0.074$ , p ≤ .001; $b_{I n c o m e} = 0.124$ , p ≤ .001), which may reflect an experiential effect. Similarly, there were significant within-person increases in the personal costs of punishment associated with age ( $b_{A g e} = 0.062$ , p ≤ .001). With respect to between-person variation, the specification of the random-effect variances around slope parameters led to a significant increase in the model fit (p ≤ .001). The significant random-effect variances around the slope parameters identified variability in the pattern of change in three psychological traits, delinquent peer influence, and age among individuals. Females compared with males and non-Blacks compared with Blacks perceived fewer personal costs of punishment ( $b_{G e n d e r} = - 0.513$ , p ≤ .001; $b_{W h i t e / O t h e r} = - 0.449$ , p ≤ .001; $b_{H i s p a n i c} = - 0.161$ , p = .002).⁷

In Model 3, the effects of within-person changes in all three domains (temperance, perspective, and responsibility) of maturity of judgment were statistically significant predictors of perceptions of social costs of punishment ( $b_{T e m p e r a n c e} = 0.031$ , p = .05; $b_{P e r s p e c t i v e} = 0.122$ , p ≤ .001; $b_{R e s p o n s i b i l i t y} = - 0.042$ , p = .013). However, the effect of responsibility was counterintuitive in that the development in responsibility decreased, not increased, the perception of social costs of punishment. Among the control variables, all time-varying factors except for delinquent peer influence showed significant influences on perceptions of the social costs of punishment in expected directions ( $b_{A g g r e s s i v e} = - 0.024$ , p = .006; $b_{I n c o m e} = - 0.027$ , p = .012; $b_{A g e} = 0.074$ , p ≤ .001). Interestingly, the number of prior offenses committed in the past lowered the perceived social costs of punishment. Comparing the results from the previous model that higher perceived personal costs of punishment were associated with a greater number of prior offenses, it may be assumed that juvenile offenders do not fear social costs as substantially as personal costs for offending. Rather, they perceived lower social costs of crime as their criminal career lengthened. Nevertheless, aging showed a significant positive effect on the perception of social costs of punishment. The random effects of the slope coefficients for time-varying covariates were supported by significant increases in model fit (p ≤ .001), indicating that the rates of change in perceived social costs of punishment deriving from three personality traits, delinquent peer influence, and age are different between individuals. Contrary to the findings in Model 2, females compared with males and non-Blacks compared with Blacks reported higher perceived social costs entailed by punishment ( $b_{G e n d e r} = 0.137$ , p ≤ .001; $b_{W h i t e / O t h e r} = 0.274$ , p ≤ .001; $b_{H i s p a n i c} = 0.090$ , p = .001).

In Model 4, within-person changes in the perceived personal rewards of crime (i.e., kick or fun) were significantly related to all three domains of the maturity of judgment and their associations fit expectation (i.e., all three were negative) ( $b_{T e m p e r a n c e} = - 0.433$ , p ≤ .001; $b_{P e r s p e c t i v e} = - 0.216$ , p ≤ .001; $b_{R e s p o n s i b i l i t y} = - 0.141$ , p ≤ .001). In short, the development of temperance, perspective, and responsibility resulted in lower perceptions of personal gratification associated with crime. Stronger delinquent peer influence led to higher expectations of personal rewards of crime ( $b_{P e e r} = 0.264$ , p ≤ .001). Involvement in aggressive offending increased the perception of personal rewards of crime ( $b_{A g g r e s s i v e} = 0.111$ , p ≤ .001), and perceptions of personal rewards of crime gradually decreased with age ( $b_{A g e} = - 0.110$ , p ≤ .001). As for between-person effects, the significant random-effect variances around the slopes for time-varying covariates suggest that substantial between-individual heterogeneity in the rates of change is attributable to three psychosocial variables, delinquent peer influence, and age (p ≤ .001). Females perceived lower personal rewards of crime than males ( $b_{G e n d e r} = - 0.555$ , p ≤ .001). Interestingly, Whites and other racial/ethnic groups, but not Hispanics, generally expected greater personal rewards of crime than Blacks ( $b_{W h i t e / O t h e r} = 0.215$ , p = .001).

In Model 5, within-person changes in the social rewards of crime were influenced by all three domains of the maturity of judgment ( $b_{T e m p e r a n c e} = - 0.104$ , p ≤ .001; $b_{P e r s p e c t i v e} = - 0.047$ , p ≤ .001; $b_{R e s p o n s i b i l i t y} = - 0.099$ , p ≤ .001), such that increases in temperance, perspective, and responsibility were associated with lower perceived social rewards of crime. Delinquent peer influence and aggressive offending had positive effects on the perception of social rewards of crime, and income-generating offending showed a marginally significant positive effect ( $b_{P e e r} = 0.072$ , p ≤ .001; $b_{A g g r e s s i v e} = 0.012$ , p = .007; $b_{I n c o m e} = 0.011$ ; p = .05). The findings indicate that offending may be perceived as a means of gaining social status among juvenile offenders. Also, aging had a direct negative influence on the perception of social rewards of crime ( $b_{A g e} = - 0.007$ , p = .004). As for the between-person effects, the variances around the slope parameters for time-varying covariates were statistically significant (p ≤ .001), indicating there was between-individual heterogeneity in the rates of change caused by psychosocial traits, delinquent peer influence, and age. Females tended to perceive fewer social rewards associated with crime ( $b_{G e n d e r} = - 0.104$ , p ≤ .001). In general, the findings in Model 5 were similar to those in Model 4. However, one difference was found with the effect of race/ethnicity: Non-Blacks perceived fewer social rewards of crime than Blacks ( $b_{W h i t e / O t h e r} = - 0.118$ , p ≤ .001; $b_{H i s p a n i c} = - 0.091$ , p ≤ .001).

Discussion

The findings from this study have several implications for deterrence research and juvenile justice policies. First, the results show that within-person changes in all dimensions of maturity of judgment were associated with within-person changes in perceptual deterrence factors. These findings add to the evidence in support of Steinberg and Cauffman’s (1996) theoretical framework that psychosocial traits—which develop over the life course—affect rational decision-making. At the same time, the results of this study identify limitations in the theoretical premise of the initial deterrence and rational choice frameworks that the capacity to weigh advantages and disadvantages is constant across individuals. Rather, individual-level heterogeneity seems to be important and may explain a nontrivial portion of the between-person variation in deterrability observed at the population level (Jacobs, 2010). Policy initiatives based on the deterrence/rational choice paradigm should be considered in light of these points because, if our findings are replicated and hold in the general population, it suggests general deterrence approaches to crime control may not be equally effective for all citizens.

This study may also provide supportive evidence for juvenile justice policies that view age as a mitigating factor. The findings presented here align with this focus because we show that, even after taking into account psychosocial development and other control variables, there were clear developmental patterns in perceptual deterrence associated with age. Specifically, younger juveniles tended to perceive lower detection risks, lower costs, and greater rewards from crime.⁸ Such findings might justify policies and laws that reduce the culpability of juvenile offenders compared with their adult counterparts. They also underscore the importance of accounting for development in sanction schemes within the juvenile justice system.

The second contribution from this study can be found in several counterintuitive findings. Temperance and responsibility had relationships with changes in personal and social costs of punishment that ran counter to our hypotheses. Increases in temperance lowered the perceived personal costs of punishment and increases in responsibility reduced the perceived social costs of punishment. By contrast, the effect of perspective on perceptual deterrence was consistently significant across five models and had effects that were in the expected directions. With that, this study suggests that programs that target high-risk juveniles to develop good perspective, which is about considering future consequences and other people, may greatly improve their decision-making skills above and beyond the influence of psychosocial development.

Third, this study contributes to the renewed interest in criminal benefits, which is sometimes neglected in perceptual deterrence studies. Out of five perceptual deterrence models, only in the rewards models did all three psychosocial dimensions show significant influences in expected directions. Higher values on temperance, perspective, and responsibility had protective effects against deviance by lowering the perception of criminal rewards. Also, the fit statistics were substantially greater in the reward models than the certainty or cost models. Considering that the cost aspect of crime and punishment has traditionally been the focus of deterrence research, these findings need further attention in that juvenile offenders who are psychosocially mature may differ from immature juveniles in their perceptions of criminal rewards. Future research might explore perceptual differences between high-offending and low-offending groups (or offending and nonoffending groups), focusing on reward expectations.

Fourth, the findings regarding the effects of prior offending likely warrant further attention. This study found that the accumulation of aggressive offending experiences increased the perception of both personal and social rewards of crime, whereas income-generating offending only marginally increased the perceived social rewards. Although this study did not include information on punishment avoidance (i.e., offending experience without detection and punishment; Stafford & Warr, 1993) due to data availability, the findings may be indicating that perceptions of risk are updated in distinct ways across crime types. From another point of view, if those with a higher rate of prior offending are assumed to have had more contact with the criminal justice system, then it appears more contact with the justice system may not be effective in deterring future crime but only aggravate perceived personal costs of punishment.

Fifth, it is noteworthy that gender and race/ethnicity had effects on perceptual deterrence mechanisms. Compared with males, females perceived greater certainty of punishment and social costs of punishment, whereas they reported lower perceived rewards of crime. These differences may partially explain the higher criminal involvement among males. Yet, one notable exception was found in the personal costs of punishment. It turned out that males tended to perceive greater personal costs of punishment than females. On racial/ethnic differences, non-Blacks perceived greater certainty of punishment and greater social costs of punishment compared with Blacks. However, it is also intriguing that Blacks expected higher personal costs of punishment, lower personal rewards, but higher social rewards. One possible explanation is the greater exposure to the criminal justice system among those in Black communities may affect their perceptions by lowering their sense of thrill associated with offending. Moreover, ethnographic research in predominantly Black communities depicts crime and delinquency as the code of the street, which is needed for survival and as a way to “save face” or gain social status (Anderson, 1999; Warr, 2002).

Finally, this study’s findings may speak to other important theories in criminology and psychology. One is a reformulated social control theory arguing that self-control is “one part of a larger constellation of executive functions … localized in the frontal lobes of the brain” (Beaver, Wright, & DeLisi, 2007, p. 1346). This perspective views the capacity for self-control as partly determined by biogenic factors and that levels of self-control can vary until early to mid-20s when brain development is complete (Beaver et al., 2007; Jackson & Beaver, 2013). Another related theory is the dual systems model that explains high rates of risk-taking behavior during adolescence in terms of the disjuncture between an early-maturing socioemotional system (e.g., sensation seeking, reward sensitivity) and a still immature cognitive control system (e.g., self-regulation; Steinberg, 2008). This model posits that the capacity for cognitive control develops linearly with age (Shulman et al., 2016; Steinberg, 2008; Steinberg et al., 2008).

It may seem that maturity of judgment, self-control, and self-regulation are overlapping concepts due to some common components (e.g., impulse control, future consideration). However, this study used the maturity of judgment scale to follow the original theoretical work of Steinberg and Cauffman (1996) as it was deemed to be more closely aligned with the prior literature on deterrence and decision making. Furthermore, the responsibility domain (self-reliance, identity, work orientation) is a concept that is not explicitly described by self-control theory or the dual systems model. Regardless, this study’s key findings that higher scores on some of the maturity of judgment domains (temperance, perspective) are associated with better decision-making skills have implications for these other perspectives on delinquent and criminal behavior.

The results in this study should be considered in light of a few limitations. First, this study uses a sample of serious juvenile offenders, so generalizability may be limited. Relatedly, as the sample of this study comprises high-risk juveniles who were already involved in the justice system, it is possible that the variations of their psychosocial traits and perceptual deterrence were restricted. The study sample might present with, on average, perceptions of punishment costs that are inordinately high (or low) compared with the general population. Unfortunately, these data do not permit comparisons between the Pathways sample and a general population sample, so we invite future work to explore this possibility.

The use of high-risk juvenile offenders, however, can be justified for the purpose of this study, which aims to target the most policy-relevant group. In particular, the finding that those with greater increases in maturity of judgment tend to perceive higher risks and costs and lower rewards from crime may indicate differential responsiveness to rational updating based on maturity level. Moreover, regression analyses of the baseline data show that scores on three domains of maturity of judgment generally predict scores on five deterrence-oriented perceptions in an expected manner (Appendix D). That is, even among juveniles who are high risk, baseline differences in maturity of judgment exist.

Second, we computed composite scores of psychosocial dimensions rather than using separate scales. Although the effect of each psychological trait on perceptual deterrence is not specified, this approach helped to generate more parsimonious models. Future research could delve into the influences of specific psychosocial traits. This might prove helpful to the design of a rehabilitation program because it could allow for more narrowly targeted interventions, which is an important part of contemporary treatment paradigms (Andrews & Bonta, 2010).

Last, the scope of the control measures included in the analyses is somewhat limited. In addition to three key independent variables, this study controls for six covariates. Inclusion of other theoretically relevant variables might shore up the key inferences made here, but there were some limitations in data coverage across the waves of data that restricted our ability to include more covariates. For example, the number of valid cases when adding items for parental attachment and school attachment was fewer than 500 (out of 1,354) after around 3 years.⁹

Conclusion

Deterrence theory’s intuitive message that the infliction of punishment increases the (perceived) costs of crime and makes the act of offending less appealing to rational criminals has been a driving force behind many policy decisions (Pratt et al., 2006). This argument disregards perceptual differences within and between individuals, so recent deterrence studies have revisited the theory with an eye toward explaining differences in perceptions. The present study joins this line of theoretical extension and tests whether within- and between-individual perceptual differences are explained by differences in levels of psychosocial maturity. The results indicate that more mature judgment ability, represented by higher scores on three psychosocial traits, was generally associated with the perception of greater risks, heavier costs of punishment, and fewer rewards of crime. Importantly, these developmental trajectories vary from individual to individual.

Despite its limitations, the present study endorses a renewed direction in deterrence research that integrates two contrasting frameworks in criminology: theories of crime and theories of criminality. As psychosocial factors related to the maturity of judgment significantly influence the perceptions of risks, costs, and benefits associated with offending, the same crime prevention tactics may be differentially processed depending on the would-be offender’s ability to perceive the risks, costs, and benefits of crime. Therefore, crime prevention strategies should manipulate situations that might be mediated by the psychosocial capabilities of those at risk. As Pogarsky, Roche, and Pickett (2018 ) noted, “changing the framing of how an individual perceives the crime decision may increase deterrence” (p. 394). Our findings suggest that perceptions are affected by one’s level of psychosocial maturity, which may one day lend itself to positive rehabilitative interventions.

Footnotes

Appendix

Appendix D

Coefficients (Standard Errors) of the Ordinary Least Square Regression Models Using Baseline Data

Variable	Model 1Certainty of punishment	Model 2Personal costs of punishment	Model 3Social costs of punishment	Model 4Personal rewards of crime	Model 5Social rewards of crime
Variable	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Intercept	8.106 (0.876)***	0.870 (0.531)	3.148 (0.336)***	2.689 (0.823)**	1.817 (0.158)***
Temperance	0.265 (0.080)***	−0.088 (0.048)	−0.010 (0.031)	−0.713 (0.075)***	−0.124 (0.014)***
Perspective	0.269 (0.077)***	0.191 (0.046)***	0.084 (0.029)**	−0.460 (0.072)***	−0.051 (0.014)***
Responsibility	−0.309 (0.081)***	0.029 (0.049)	−0.074 (0.031)*	−0.027 (0.076)	−0.074 (0.015)***
Delinquent peer influence	−0.196 (0.077)*	0.278 (0.047)***	0.002 (0.029)	0.584 (0.072)***	0.076 (0.014)***
Aggressive offending	−0.099 (0.063)	0.008 (0.038)	0.003 (0.024)	0.130 (0.059)*	0.018 (0.011)
Income-generating offending	−0.285 (0.064)***	0.216 (0.039)***	−0.041 (0.025)	−0.135 (0.060)*	0.002 (0.012)
Age	−0.191 (0.054)***	0.080 (0.033)*	−0.037 (0.021)	−0.045 (0.051)	0.017 (0.010)
Gender (Reference: Male)	0.769 (0.177)***	−0.373 (0.107)***	0.058 (0.068)	−0.055 (0.166)	−0.111 (0.032)***
Race (Reference: Black)
White/Other	0.702 (0.156)***	−0.607 (0.094)***	0.301 (0.060)***	0.788 (0.146)***	−0.124 (0.028)***
Hispanic	0.329 (0.146)*	−0.356 (0.089)***	0.317 (0.056)***	0.630 (0.137)***	−0.109 (0.026)***
VIF range	1.022-1.357	1.021-1.369	1.021-1.369	1.021-1.369	1.021-1.369
R² (Adjusted R²)	.118 (.112)	.131 (.125)	.052 (.045)	.284 (.278)	.218 (.212)
F-statistic	17.449***	19.680***	7.199***	51.539***	36.331***
N _ind	1,310	1,311	1,311	1,311	1,311

Note. VIF = variance inflation factor.

p ≤ .05. **p ≤ .01. ***p ≤ .001 (two-tailed test).

Authors’ Note:

Data for the study were provided by the Pathways to Desistance study downloaded from the National Archive of Criminal Justice Data at Inter-university Consortium for Political and Social Research. The original investigators were sponsored by Arizona Governor’s Justice Commission (JBISA012244400), John D. and Catherine T. MacArthur Foundation, Pennsylvania Commission on Crime and Delinquency (2001-J05-011944, 2002-J04-13032, 2003-J04-14560, 2004-J04-15849, 2005-J04-17071, 2006-J04-18272), Robert Wood Johnson Foundation (043357), Centers for Disease Control and Prevention, National Institute on Drug Abuse (R01 DA 019697 05), National Institute of Justice (1999-IJ-CX-0053, 2008-IJ-CX-0023), Office of Juvenile Justice and Delinquency Prevention (2000-MU-MU-0007, 2005-JK-FX-K001, 2007-MU-FX-0002), William Penn Foundation, and William T. Grant Foundation (99-2009-099).

Notes

ORCID iD

Heejin Lee

Heejin Lee is a doctoral student in the School of Criminal Justice at the University of Cincinnati. Her research interests focus on developmental and life-course criminology, cognitive criminology, and criminological theory testing.

Christopher J. Sullivan is a professor and graduate program director in the School of Criminal Justice at the University of Cincinnati. His research interests include developmental and life-course criminology, juvenile delinquency and prevention policy, and research methodology and analytic methods.

J. C. Barnes is an associate professor in the School of Criminal Justice at the University of Cincinnati. His primary research area focuses on the interplay between genetic and environmental factors in the development of human behavior. He has also recently worked on projects in applied statistical analysis and offender decision making.

References

Akers

R. L.

(1985). Deviant behavior: A social learning approach. Belmont, CA: Wadsworth Publishing.

Allison

P. D.

(2005). Fixed effects regression methods for longitudinal data using SAS®. Cary, NC: SAS Institute.

Allison

P. D.

(2009). Fixed effects regression models. Thousand Oaks, CA: SAGE.

Anderson

(1999). Code of the street: Decency, violence, and the moral life of the inner city. New York, NY: W.W. Norton.

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). New Providence, NJ: Anderson.

Anwar

Loughran

T. A.

(2011). Testing a Bayesian learning theory of deterrence among serious juvenile offenders. Criminology, 49, 667-698.

Baker

Piquero

A. R.

(2010). Assessing the perceived benefits—Criminal offending relationship. Journal of Criminal Justice, 38, 981-987.

Beaver

K. M.

Wright

J. P.

DeLisi

(2007). Self-control as an executive function: Reformulating Gottfredson and Hirschi’s parental socialization thesis. Criminal Justice and Behavior, 34, 1345-1361.

Beccaria

(2009). On crimes and punishments ( Thomas

Parzen

, Trans.). In Thomas

(Ed.), On crimes and punishments and other writings (pp. 1-8). Toronto, Ontario, Canada: University of Toronto Press. (Original work published 1764).

10.

Cauffman

Steinberg

(2000). (Im)maturity of judgment in adolescence: Why adolescents may be less culpable than adults. Behavioral Sciences & the Law, 18, 741-760.

11.

Clarke

R. V.

Cornish

D. B.

(1985). Modeling offenders’ decisions: A framework for research and policy. Crime and Justice, 6, 147-185.

12.

Cloward

Ohlin

(1960). Delinquency and opportunity: A theory of delinquent gangs. Glencoe, IL: The Free Press.

13.

Cullen

F. T.

Pratt

T. C.

Miceli

S. L.

Moon

M. M.

(2002). Dangerous liaison? Rational choice theory as the basis for correctional. In Piquero

A. R.

Tibbetts

S. G.

(Eds.), Rational choice and criminal behavior: Recent research and future challenges (pp. 279-296). New York, NY: Routledge.

14.

Decker

Wright

Logie

(1993). Perceptual deterrence among active residential burglars: A research note. Criminology, 31, 135-147.

15.

Deryol

Wilcox

Dolu

(2017). School-based violent victimization in Turkey: An examination of the cross-national generality of lifestyle-routine activities and self-control theories. Victims & Offenders, 12, 913-938.

16.

Fagan

Piquero

A. R.

(2007). Rational choice and developmental influences on recidivism among adolescent felony offenders. Journal of Empirical Legal Studies, 4, 715-748.

17.

Glueck

(1968). Delinquents and nondelinquents in perspective. Cambridge, MA: Harvard University Press.

18.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

19.

Graham

J. W.

(2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology, 60, 549-576.

20.

Greenberger

Josselson

Knerr

(1975). The measurement and structure of psychosocial maturity. Journal of Youth and Adolescence, 4, 127-143.

21.

Halaby

C. N.

(2004). Panel models in sociological research: Theory into practice. Annual Review of Sociology, 30, 507-544.

22.

Hox

J. J.

(2010). Multilevel analysis: Techniques and applications (2nd ed.). New York, NY: Routledge.

23.

Jackson

D. B.

Beaver

K. M.

(2013). The influence of neuropsychological deficits in early childhood on low self-control and misconduct through early adolescence. Journal of Criminal Justice, 41, 243-251.

24.

Jacobs

B. A.

(2010). Deterrence and deterrability. Criminology, 48, 417-441.

25.

Keller

B. T.

Enders

C. K.

(2018). Blimp user’s manual (Version 1.1). Los Angeles, CA. Retrieved from http://www.appliedmissingdata.com/blimpuserguide-5.pdf

26.

Loughran

T. A.

Paternoster

Piquero

A. R.

Pogarsky

(2011). On ambiguity in perceptions of risk: Implications for criminal decision making and deterrence. Criminology, 49, 1029-1061.

27.

Loughran

T. A.

Piquero

A. R.

Fagan

Mulvey

E. P.

(2012). Differential deterrence: Studying heterogeneity and changes in perceptual deterrence among serious youthful offenders. Crime & Delinquency, 58, 3-27.

28.

Loughran

T. A.

Pogarsky

Piquero

A. R.

Paternoster

(2012). Re-examining the functional form of the certainty effect in deterrence theory. Justice Quarterly, 29, 712-741.

29.

Matthews

S. K.

Agnew

(2008). Extending deterrence theory: Do delinquent peers condition the relationship between perceptions of getting caught and offending? Journal of Research in Crime & Delinquency, 45, 91-118.

30.

Mistler

S. A.

(2013). A SAS macro for applying multiple imputation to multilevel data. In Proceedings from the SAS Global Forum 2013. San Francisco, CA: Contributed Paper (Statistics and Data Analysis). Retrieved from http://support.sas.com/resources/papers/proceedings13/438-2013.pdf

31.

Moffitt

T. E.

(1990). The neuropsychology of juvenile delinquency: A critical review. Crime and Justice, 12, 99-169.

32.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701.

33.

Mulvey

E. P.

(2016). Research on pathways to desistance [Maricopa County, AZ and Philadelphia County, PA]: Subject measures, 2000-2010. Inter-university Consortium for Political and Social Research [distributor]. doi:10.3886/ICPSR29961.v2

34.

Mulvey

E. P.

Steinberg

Fagan

Cauffman

Piquero

A. R.

Chassin

. . . Losoya

S. H.

(2004). Theory and research on desistance from antisocial activity among serious adolescent offenders. Youth Violence and Juvenile Justice, 2, 213-236.

35.

Nagin

D. S.

(1998). Criminal deterrence research at the outset of the twenty-first century. Crime and Justice, 23, 1-42.

36.

Nagin

D. S.

Paternoster

(1993). Enduring individual differences and rational choice theories of crime. Law and Society Review, 27, 467-496.

37.

Nagin

D. S.

Paternoster

(1994). Personal capital and social control: The deterrence implications of a theory of individual differences in criminal offending. Criminology, 32, 581-606.

38.

Parker

Grasmick

H. G.

(1979). Linking actual and perceived certainty of punishment. Criminology, 17, 366-379.

39.

Paternoster

(2010). How much do we really know about criminal deterrence? Journal of Criminal Law and Criminology, 100, 765-824.

40.

Paus

(2005). Mapping brain maturation and cognitive development during adolescence. Trends in Cognitive Sciences, 9, 60-68.

41.

Pauwels

Weerman

Bruinsma

Bernasco

(2011). Perceived sanction risk, individual propensity and adolescent offending: Assessing key findings from the deterrence literature in a Dutch sample. European Journal of Criminology, 8, 386-400.

42.

Piquero

A. R.

Weisburd

(Eds.). (2010). Handbook of quantitative criminology. New York, NY: Springer.

43.

Pogarsky

(2002). Identifying “deterrable” offenders: Implications for research on deterrence. Justice Quarterly, 19, 431-452.

44.

Pogarsky

Kim

Paternoster

(2005). Perceptual change in the National Youth Survey: Lessons for deterrence theory and offender decision-making. Justice Quarterly, 22, 1-29.

45.

Pogarsky

Piquero

A. R.

(2003). Can punishment encourage offending? Investigating the “resetting” effect. Journal of Research in Crime & Delinquency, 40, 95-120.

46.

Pogarsky

Roche

S. P.

Pickett

J. T.

(2018). Offender decision-making in criminology: Contributions from behavioral economics. Annual Review of Criminology, 1, 379-400.

47.

Pratt

T. C.

Cullen

F. T.

Blevins

K. R.

Daigle

L. E.

Madensen

T. D.

(2006). The empirical status of deterrence theory: A meta-analysis. In Cullen

F. T.

Wright

Blevins

(Eds.), Taking stock: The status of criminological theory (pp. 367-396). New Brunswick, NJ: Transaction Publishers.

48.

Rowe

D. C.

Osgood

D. W.

Nicewander

(1990). A latent trait approach to unifying criminal careers. Criminology, 28, 237-270.

49.

Rubin

D. B.

(1987). Multiple imputation for nonresponse in surveys. New York, NY: John Wiley.

50.

SAS Institute. (2011). SAS/STAT® 9.3 user’s guide. Cary, NC: Author.

51.

Schafer

J. L.

Olsen

M. K.

(1998). Multiple imputation for multivariate missing-data problems: A data analyst’s perspective. Multivariate Behavioral Research, 33, 545-571.

52.

Schubert

C. A.

Mulvey

E. P.

Pitzer

(2016). Differentiating serious adolescent offenders who exit the justice system from those who do not. Criminology, 54, 56-85.

53.

Schubert

C. A.

Mulvey

E. P.

Steinberg

Cauffman

Losoya

S. H.

Hecker

. . . Knight

G. P.

(2004). Operational lessons from the pathways to desistance project. Youth Violence and Juvenile Justice, 2, 237-255.

54.

Schulz

(2014). Individual differences in the deterrence process: Which individuals learn (most) from their offending experiences? Journal of Quantitative Criminology, 30, 215-236.

55.

Shulman

E. P.

Smith

A. R.

Silva

Icenogle

Duell

Chein

Steinberg

(2016). The dual systems model: Review, reappraisal, and reaffirmation. Developmental Cognitive Neuroscience, 17, 103-117.

56.

Stafford

M. C.

Warr

(1993). A reconceptualization of general and specific deterrence. Journal of Research in Crime & Delinquency, 30, 123-135.

57.

Steinberg

(2008). A social neuroscience perspective on adolescent risk-taking. Developmental Review, 28, 78-106.

58.

Steinberg

Albert

Cauffman

Banich

Graham

Woolard

(2008). Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: Evidence for a dual systems model. Developmental Psychology, 44, 1764-1778.

59.

Steinberg

Cauffman

(1996). Maturity of judgment in adolescence: Psychosocial factors in adolescent decision making. Law and Human Behavior, 20, 249-272.

60.

Steinberg

Graham

O’Brien

Woolard

Cauffman

Banich

(2009). Age differences in future orientation and delay discounting. Child Development, 80, 28-44.

61.

Tabachnick

B. G.

Fidell

L. S.

(2001). Using multivariate statistics (5th ed.). New York, NY: Pearson.

62.

Thornberry

T. P.

Lizotte

A. J.

Krohn

M. D.

Farnworth

Jang

S. J.

(1994). Delinquent peers, beliefs, and delinquent behavior: A longitudinal test of interactional theory. Criminology, 32, 47-83.

63.

Warr

(2002). Companions in crime: The social aspects of criminal conduct. Cambridge, UK: Cambridge University Press.

64.

Zimring

F. E.

Hawkins

(1968). Deterrence and marginal groups. Journal of Research in Crime and Delinquency, 5, 100-114.