“No Remorse, No Repent”: Linking Lack of Remorse to Criminal Offending in a Sample of Serious Adolescent Offenders

Abstract

Linking emotions to offender decision-making has only recently become of theoretical interest to criminologists, but empirical work in this area has not kept pace nor has such research examined the role of emotions to offending in offender-based samples. Recently, Warr outlined regret as one such emotion that may be useful in thinking about offending. Specifically, he argued that regret may be related to discontinuity in offending, or conversely that a lack of remorse may be related to continuity in offending. This paper uses data from a sample of serious adolescent offenders followed for seven years to investigate this hypothesis. Results provide support for Warr’s hypothesis that remorse-resistant adolescents incur a higher number of re-arrests, while remorse-prone adolescents incur fewer re-arrests, even after controlling for other relevant risk factors.

Keywords:

• remorse
• regret
• offending
• adolescent offenders

Notes

¹ Sherman’s (2003) Presidential Address to the American Society of Criminology advanced a new paradigm of sorts that would focus on “the explicit recognition and management of the effects of emotions in causing the behavior of both officials and offenders” (p. 2).

² Although not as explicit, Gottfredson and Hirschi’s (1990) general theory of crime also incorporates some emotions into the decision-making process (see also Piquero, Gomez-Smith, & Langton, 2004), while psychologically-based theories have focused on hostile attribution bias, which certainly incorporates emotions in the decision-making process (see Lochman & Dodge, 1994).

³ Warr (in press, p. 5) further notes that “Genuine regret reveals a change of heart, from offender to desister.” Some psychologists have further considered the inter-relationships between shame, guilt, and remorse, noting that guilt may be a more useful “umbrella term, subsuming ‘remorse’” (Tangney, Stuewig, & Hafez, 2011a, p. 706), while at the same time taking more care to distinguish between shame and guilt. For Tangney and her colleagues (2011b, p. 712), there are two ways that individuals may feel bad about their transgressions, one (guilt) is more adaptive than the other (shame). It would not be a stretch to expect Warr to hold to the idea that regret/remorse is more adaptive than it would be dysfunctional.

⁴ For example, King’s (2013) qualitative study of desistance narratives among twenty individuals under probation supervision in England revealed a strong sentiment for the impact of the offender’s actions on victims and significant others; as one offender put it: “I felt very remorseful for the people that I caused injuries to” (p. 155).

⁵ In the restorative justice literature, Ahmed, Harris, Braithwaite, and Braithwaite (2001) have highlighted the role of shame acknowledgement as an inhibitor of subsequent offending. The overlap between remorse and shame acknowledgement is an important one, and will be further discussed in the discussion section regarding policy implications.

⁶ Specifically, the four and three factor model of the PCL:YV was associated with the most chronic and serious offending trajectory, as were the lifestyle and antisocial factors, but the interpersonal and affective factors did not predict membership in the high-rate chronic group (Corrado et al., 2015, p. 357).

⁷ Pardini and Byrd (2013) recently provided an overview of the psychopathy literature. Of relevance to the current study, these authors noted that the affective features of psychopathy are commonly referred to as CU traits, and remorse is an affective component within both CU and psychopathy (the latter which is more expansive than CU). The affective component includes “experiencing affective discomfort following wrongdoing (i.e. guilt) and having empathic concern for the well-being of others [citations omitted], which are core characteristics of CU traits” (Pardini & Byrd, 2013, p. 64). They also pointed out that there was a solid base of longitudinal research documenting the stability of psychopathic features in children and adolescents (Frick, Cornell, Barry, Bodin, & Dane, 2003; Obradović, Pardini, Long, & Loeber, 2007).

⁸ Specifically, the remorse item was: “When he does something wrong, does he feel bad about it?”.

⁹ For example, “I am concerned about the feelings of others” and “I feel bad or guilty when I do something wrong”. It is also worth noting that aside from the conceptual overlap between remorse, regret, psychopathy, and CU traits, basic definitions also point to quite a bit of common ground. According to Webster’s Dictionary, shame is defined as a feeling of guilt, regret, or sadness that you have because you know you have done something wrong; guilt is defined as a bad feeling caused by knowing or thinking that you have done something bad or wrong; regret is defined as to feel sad or sorry about (something that you did or did not do); remorse is defined as a feeling of being sorry for doing something bad or wrong in the past/a feeling of guilt. In short, there is quite a bit of both conceptual and definitional overlap among these individual characteristics (see Tangney, Miller, Flicker, & Barlow, 1996).

¹⁰ Specifically, “desire to put matters right” was measured with an item “Felt concerned to put matters right and put it behind me.” As well, Murphy and Harris incorporated several measures associated with shame acknowledgement and shame displacement in their analysis. Within the shame acknowledgement construct in particular, two items (“Felt what I’d done was wrong” and “Felt regret”) would be consistent with Warr’s notion of regret/remorse. Yet, while more shame acknowledgement was associated with a higher “desire to put matters right”, it was not related to re-offending (p. 909).

¹¹ As Tangney et al. (2014, p. 799, emphasis added) observe: “When people feel guilt about a specific behavior, they experience tension, remorse, and regret. Research has shown that this sense of tension and regret typically motivates reparative action—confessing, apologizing, or somehow repairing the damage done [citations omitted]” and this is especially the case when the feelings arise over the “bad things done” (Stuewig et al., 2015, p. 218).

¹² Stuewig and McCloskey (2005) examined the link between shame- and guilt-proneness to depression and delinquency and found that shame-proneness in childhood was significantly associated with higher depression in a two-year follow-up, whereas guilt-proneness was linked to significantly less delinquency and (marginally significantly) to less depression. They also found that harsh parenting in childhood was associated with higher shame-proneness but lower guilt-proneness.

¹³ An anonymous reviewer correctly pointed out that these are single items of a broader measure.

¹⁴ The one-item measure of remorse could be viewed as weak from a psychometric perspective.

¹⁵ Specifically, the committing offenses included: 41% for violent crimes against persons (e.g. murder, rape, robbery, assault), 26% for property crimes (e.g. arson, burglary, receiving stolen property), 10% for weapons, 4% for sex crimes, and 4% for other crimes (e.g. conspiracy, intimidation of a witness).

¹⁶ Schubert et al. (2004, pp. 247–250) provided a detailed comparison of the enrolled cases compared to the non-enrolled cases that went through the courts during the enrollment period. Because of space constraints, it is noted here that comparisons between the adjudicated-but-not-enrolled cases versus the enrolled cases revealed only slight differences. For example, the enrolled group was slightly younger (15.9 vs. 16.1), had a slightly higher number of prior petitions (2.1 vs. 1.5), were younger at first prior petition (13.9 vs. 14.2), were less likely to be male (86% vs. 91%, due to the fact that very few females met all of the criteria for inclusion during the enrollment period in both locales), were more likely to be white (25% vs. 20%, though no other race/ethnicity differences emerged), and were less likely to be adjudicated with later case dismissal at the time of disposition at adjudication (1% vs. 2%). In all, based on the many similarities and few substantive differences across groups, as well as qualitative information gleaned from the Pathways to Desistance Working Group with project staff as well as personnel in both locales, there is a high degree of confidence that the enrolled group is representative of the types of cases that were being processed during the enrollment period and that met the study eligibility criteria.

¹⁷ An anonymous reviewer correctly pointed out that there is variation among social scientists in how they interpret and describe the strength of correlations, especially with respect to the terms “moderate” or “strong”. For example, Hinkle, Wiersema, and Jurs (2003) suggest that correlations between .3 and .5 are low positive while correlations between .5 and .7 are moderate. On the other hand, Bergström, Forth, and Farrington (2015) recommend that correlations between .2 and .39 reflect moderate stability/strength, while correlations of .4 to .59 reflect high stability/strength. Although correlations between earlier and later observations ebb over time (i.e. 6 to 84 months) as expected in longitudinal analysis, correlations in adjacent time periods are much stronger, ranging between .43 and .53.

¹⁸ There is some debate over whether emotions like remorse, guilt, and shame are (relatively) stable over the life course or if there is significant heterogeneity, especially in the adolescent as opposed to adult years. Unfortunately, there are very few offender-based longitudinal studies of remorse in particular, with frequent assessments obtained during the mid-adolescence to early-adulthood time period, so stability or variability remains an open question. There is some literature on the extent to which there is stability in psychopathy in longitudinal studies—but these studies do not converge on a single answer, i.e. stability or variability, in large part because they are contingent on the type of sample and its age range. Several studies indicate that total psychopathy scores are quite stable during childhood and adolescence (Frick et al., 2003; Lynam, Loeber, & Stouthamer-Loeber, 2008; Salihovic, Kerr, Ozdemir, & Pakalniskiene, 2012), while other studies indicate that mean levels of psychopathic traits generally decline from late adolescence and into early adulthood (Hawes, Mulvey et al., 2014). Other studies indicate differences within the subscales of the most common psychopathy measures, the Psychopathy Checklist (PCL), with studies indicating that the affective subscale (which includes remorse) is more stable than other subscales (see Blonigen, Hicks, Krueger, Patrick, & Iacono, 2006; Brandt et al., 1997). Very few studies have prospectively examined the stability of psychopathy scores in the transition from adolescence to adulthood and those that do yield mixed evidence (Lynam, Caspi, Moffitt, Loeber, & Stouthamer-Loeber, 2007; Hawes et al., 2014). One early study following 889 prisoners between ages 16 and 70 revealed very good stability in Factor 1 of the PCL-R, which measures interpersonal and affective traits—the latter of which includes items related to remorse (Harpur & Hare, 1994). Cauffman, Skeem, Dmietrieva, and Cavanagh’s (2015) recent study examined stability in the PCL:YV and PCL:R in two samples of offenders, 202 adolescents (ages 14–17) and 134 adults (ages 26–29) in three assessments at one month, twelve months, and twenty-four months after a baseline assessment. Their findings indicated that, at an absolute level, adolescents’ scores were moderately stable over a two-year period, but that total scores dropped an average of 2 points between ages 14–17; relative to adults, adolescents’ scores were less stable than adults’ scores; and that psychosocial maturity was related to decreases in psychopathy scores among adolescents but not adults. As well, there was variability in adolescent subscale scores, with more stability observed in the interpersonal and antisocial scales than the affective and lifestyle scales. In a recent review of the research on the stability of psychopathy, Bergström et al. (2015) observed that the existing knowledge base, as well as their own forty-year analysis using data from the Cambridge Study in Delinquent Development, consistently documents moderate to high stability at the rank-order level with respect to psychopathic trait measures, with very little changes within persons: “…people keep their relative position on psychopathic traits within the sample over time (Roberts, Caspi, & Moffitt, 2001)” (p. 17). The findings of moderate to strong stability across different studies, with different samples, and with different measures is, according to Bergström and her colleagues “quite remarkable” (p. 17).

¹⁹ In the pooled negative binomial regression, the standard errors were cluster-corrected, which relaxes the usual requirement that the observations be independent (Stata 11.2). As well, regression coefficients from count-based models are expressed in the form of Incidence Rate Ratios (IRRs). That is, the parameter estimates may be interpreted as the log of the ratio of expected counts, or a “ratio of ratios” and counts by definition can also be referred to as a rate. A rate is the number/frequency of events per time or space, and the rate at which these events occur can be considered an incidence rate (see Long, 1997; Long & Freese, 2014).

²⁰ GEE is a form of population-averaged model, where the longitudinal data are treated in a cross-sectional manner and where the main interest lies in the population-averaged response (in this case, the lack of remorse).

²¹ The author wishes to thank David McDowall for comments and suggestions regarding the methodological issues.

²² As Tangney et al. (2007, p. 345) note: “Moral emotions represent a key element of our human moral apparatus, influencing the link between moral standards and moral behavior.” An anonymous reviewer correctly noted that there remains much to be done with respect to not only creating measures of regret but also in testing how such measures are similar to or distinct from other moral emotions, especially guilt. Further, the reviewer raised the intriguing possibility that perhaps individuals need to feel guilt which then leads into regret and remorse—or perhaps that when coupled with shame regret and remorse may be maladaptive. Further, s/he noted that there may be different types of regret (i.e. “I regret that I committed the crime because I got caught” vs. “I regret that I harmed that person”). These are all excellent points and opportunities for future work.

²³ The evidence in favor of empathy-based interventions among sexual and violent offenders remains mixed (see Day, Casey, & Grace, 2010).

Additional information

Notes on contributors

Alex R. Piquero

Alex R. Piquero is Ashbel Smith Professor of Criminology and Associate Dean for Graduate Programs in the School of Economic, Political and Policy Sciences at the University of Texas at Dallas; adjunct professor at the Key Center for Ethics, Law, Justice, and Governance, Griffith University Australia; and faculty affiliate with the Center for Violence and Injury Prevention, George Warren Brown School of Social Work, Washington University in St. Louis. He has received several research, teaching, and mentoring awards and is a fellow of both the American Society of Criminology and the Academy of Criminal Justice Sciences. In 2014, he received the University of Texas System Regents’ Outstanding Teaching Award.