The Contributions of Propensity,Delinquent Peers,Low Parental Supervision,and Empathy to the Emergence of Antisocial Behavior in Childhood and Adolescence: Testing Developmental Path Models Combining Psychological- and Sociological-Criminological Approaches

Abstract

Psychological- and sociological-criminological research refers to, for example, cumulative risk factor models (e.g., Lösel & Bender, 2003) and Situational Action Theory (SAT; e.g., Wikström, 2006). The German longitudinal study “Chances and Risks in the Life Course” (research project A2, Collaborative Research Center 882; e.g., Reinecke, Stemmler, & Wittenberg, 2016) focuses upon the development of antisocial behavior from a psychological and sociological point of view. Two-wave panel data of two cohorts (children and adolescents) were utilized to test the power of developmental path models investigating the development of antisocial behavior. Individual risk seems to have both direct and indirect influences on antisocial behavior, supporting the ideas of risk factor models; antisocial behavior might be the outcome of the interaction between propensity and criminogenic exposure, so there is evidence for SAT. Additionally, empathy seems to be related to both propensity and low parental supervision. Implications for the study of antisocial behavior in childhood and adolescence are discussed in line with developmental criminology.

Keywords

Propensity criminogenic exposure individual risk empathy antisocial behavior

Introduction

The emergence of antisocial behavior is elucidated by different approaches from the psychological-criminological and the sociological-criminological perspective which often overlap and to some extent complement one another. On one hand, from the psychological-criminological point of view, amongst others, the model of cumulating risk in the development of persistent antisocial behavior of Lösel and Bender (2003) and Moffitt’s theory concerning adolescence-limited and life-course persistent antisocial behavior (Moffitt, 1993; see also Moffitt, 2018) are important approaches. Generally, biopsychosocial risk models correspond with Moffitt’s theory of life-course persistent antisocial behavior. On the other hand, from the sociological-criminological perspective, amongst others, the Situational Action Theory (SAT), developed by Wikström (2006, 2009), is crucial. Relating to the psychological-criminological and sociological-criminological approaches, certain principal similarities emerge. According to the biopsychosocial risk model ( Lösel & Bender, 2003), the cumulation of risks can support the development of a persistent antisocial lifestyle. Numerous distinct risks can contribute to the explanation of antisocial behavior in childhood and adolescence (e.g. Corrado, 2012; Farrington, Ttofi, & Piquero, 2016; Stemmler, Wallner, & Link, 2018; Wallner, Lösel, Stemmler, & Corrado, 2018), for example, individual and environmental risks such as low self-control, poor child rearing, or deviant peers. Similarly, according to Wikström, Oberwittler, Treiber, and Hardie (2012), SAT “[] aims to integrate individual and environmental perspectives on crime causation by proposing that acts of crime (which are defined as moral rules stated in law) are the result of a perception-choice process guided by the interaction between a person’s propensity to commit crime and their exposure to criminogenic settings [. . .]” (ibid., p. vii). Correspondingly, two core propositions of SAT are crime propensity that is influenced by morality and the capability to exercise self-control and criminogenic exposure (e.g., delinquent peers; e.g., Wikström, 2009). In addition, concerning preceding factors that influence an individual’s action while not being a direct cause, Wikström et al. (2012) stated that “[t]he main argument advanced is that the problem of the causes of the causes (of action) is best analysed in terms of processes of (social and personal) emergence, and processes of (social and self) selection” (ibid., p. 30). For example, Schepers (2017) studied social disadvantages that are not causes of criminal behavior but rather causes of the causes and that affect the emergence of crime propensity and crimi-nogenic exposure. Both the psychological-criminological and sociological-criminological approaches focus on the importance of similar individual and environmental characteristics, even though the approaches differ with respect to basic premises, general key constructs, and assumptions.

Deficits in executive functioning and morality representing criminal propensity in SAT are also considered as important characteristics of psychopathy (cf. Fox, Jennings, & Farrington, 2015). Specifically, psychopathic traits “[. . .] can be considered as adaptive mechanisms underlying criminality in SAT, especially in certain high criminogenic environments [. . .]” (ibid., p. 280). Thus, an integration of the psychopathy construct into the SAT framework seems to be promising. According to McCuish and colleagues (McCuish, Corrado, Hart, & DeLisi, 2015), a situational action theory perspective can be used to help clarify the utility of psychopathy symptoms in explaining persistent violence. In this context, affective and emotional deficits “[. . .] were specified as being conceptually related to low levels of deterrence” (ibid., p. 10). Generally, focusing on violence as situational action, Wikström and Treiber (2009) suggested that psychopathic violence can be described by a “lack of strong emotions” (ibid., p. 87). According to SAT, they emphasized the person’s emotional involvement pointing out that “[. . .] the choice to intentionally harm someone [. . .] depends substantially upon a person’s emotional involvement in the violent course of action and whether or not he/she is capable of inhibiting, or compensating for, that emotional impetus and exhibiting self-control” (Wikström & Treiber, 2009, pp. 87-88).

Regarding the psychopathy construct, especially callous-unemotional (CU) traits may play an important role for severe and/or persistent antisocial behavior (e.g., Frick, Ray, Thornton, & Kahn, 2014; Wallner, 2016) and, hence, are stressed in many criminological studies on different antisocial outcomes (e.g., cf. Byrd, Loeber, & Pardini, 2012; Frick & White, 2008; Frick, Stickle, Dandreaux, Farrell, & Kimonis, 2005; Pardini & Fite, 2010). Concerning clinical samples, CU traits are crucial relating to conduct disorder according to DSM-5 (American Psychiatric Association, 2013) that incorporates a specifier “with limited prosocial emotions” to the diagnosis of conduct disorder. The DSM-5 CU subtype of conduct disorder focuses on individuals at heightened risk for severe antisocial behavior (ibid.; cf. also Frick et al., 2014; Kimonis et al., 2015; Moffitt et al., 2008). Psychopathic traits refer to CU traits, impulsivity, and narcissism (cf. Cooke & Michie, 2001; Frick & Hare, 2001). For instance, Bergstrøm and Farrington (2018) accentuated the combination of CU, daring-impulsive, and grandiose-manipulative traits, and the relevance for later life outcomes showing, for example, that high levels of CU and (at the same time) daring-impulsive traits go along with high risk in childhood and poorer life outcomes in adulthood. Similarly, Andershed et al. (2018) focused on CU, impulsive, and grandiosity (narcissism) traits in the prediction of different antisocial outcomes in early adolescence and pointed out the importance of other psychopathic traits than CU traits only. Relating to age, CU traits and impulsivity seem to predict conduct disorder symptoms better in adolescence than in childhood (Fanti, Kyranides, Lordos, Colins, & Andershed, 2018). Moreover, especially focusing on the role of parenting across pathways to conduct disorder, Frick (2012) noted studies suggesting that “[. . .] different aspects of parenting play a role in the development or maintenance of conduct problems, depending on whether the child shows significant levels of CU traits [. . .]” (ibid., p. 382). Silva and Stattin (2016), for instance, examined the moderating role of parenting on the relation between psychopathy and antisocial behavior in adolescence. Overall, the psychopathy concept seems to be useful for explaining the causal mechanisms which are related to chronic, serious, and violent offending (see Corrado, DeLisi, Hart, & McCuish, 2015).

The Present Study

Based on the literature summarized above, the following core research questions were derived, combining findings from psychological-criminological and sociological-criminological research and, additionally, considering different age cohorts. First, individual risk in childhood and adolescence has both direct and indirect influences on later antisocial behavior, supporting cumulative risk factor models. Second, antisocial behavior seems to be the outcome of the previous interaction between propensity and criminogenic exposure in childhood and adolescence, providing evidence for SAT. Propensity and exposure are used simultaneously as mediators; this can also be well explained by the “causes of the causes” discussion that was held within SAT. Against this background, developmental path models for children and adolescents are tested, integrating the partially overlapping and complementing approaches outlined above, including individuals’ empathy, and investigating the development of antisocial behavior in two age cohorts. The research questions are illustrated in detail utilizing the underlying path model (see Fig. 1 ). The approach taken in the current work is a strictly confirmatory approach (Jöreskog, 1993).

Fig. 1

Underlying path model illustrating the hypotheses (indirect effects are not displayed).

Methods

Research Sample

Our research project affiliates with the Collaborative Research Center (“Sonderforschungsbereich”, SFB) 882 “From Heterogeneities to Inequalities”, which was established at Bielefeld University and was funded by the German Research Foundation (DFG) until 2016. Research project A2 (i.e., the longitudinal study “Chances and Risks in the Life Course” [CURL]; “The Emergence and Development of Deviant and Delinquent Behavior over the Life Course and its Significance for Processes of Social Inequality”; e.g., Reinecke et al., 2016; Wallner, Weiss, Reinecke, & Stemmler, 2019), focuses on the emergence and development of deviant and delinquent behavior. The study enables the investigation of the associations between different self-reported deviant and delinquent outcomes (e.g., dark figures) and various important precursors (i.e., psychosocial and sociological variables). The psychological-criminological and sociological-criminological study utilizes a cohort-sequential design that allows for the investigation of children’s and adolescents’ development over time. Overall, this study is based on two other studies from psychology and sociology (i.e., the ENDPS [“Erlangen-Nuremberg Development and Prevention Study”]; e.g., see Lösel & Stemmler, 2012) and the CrimoC (“Crime in the Modern City”) study (e.g., see Boers & Reinecke, 2007).

Our longitudinal data refer to school-based surveys (group-testing sessions for several questionnaires) at the two German cities of Nuremberg and Dortmund. At large, Nuremberg and Dortmund are comparable cities concerning general aspects of population. Importantly, the Nuremberg sample is composed of students from lower-track schools; the Dortmund sample comprises different school types. Two cohorts have been assessed, namely a younger cohort that is a (late) childhood cohort, and an older cohort that is an adolescent cohort. In the research presented here, two assessment points (time 1: t1; time 2: t2) have been considered. Hence, each longitudinal dataset (i.e., each two-wave panel) consisted of male and female participants with available data at the first (t1 : 5th vs. 9th grade) and second assessment points (t2, i.e., one year later: 6th vs. 10th grade), that is N = 1,023 children (younger cohort) and N =871 adolescents (older cohort). Considering drop-out issues for both two-wave panels, in each case the average antisocial behavior score (see measures section) for the whole sample at t1 (younger cohort: M = 1.14, SD = 0.27, N = 1,319; older cohort: M = 1.13, SD = 0.25, N = 1,413) was nearly identical to the average antisocial behavior score of the remaining sample at t1 (younger cohort: M = 1.13, SD = 0.26, N = 1,013; older cohort: M = 1.12, SD =0.24, N = 867), comprising all participants who did not drop out of the study after t1 but also took part at least in the second wave of our study (i.e., the two-wave or three-wave panel of the younger/older cohort). Additional information on panel mortality can be obtained from Weiss and Link (2019). Further participant characteristics (i.e., descriptive statistics) of these two-wave panels are summarized in Table 1. The results of the path analysis and the multiple group analysis of the path model using listwise deletion of cases are based on smaller samples (see results section).

Table 1

Characteristics of Participants (Two-Wave Panels; t1-t2)

Category	Group 1:Younger cohort	Group 2:Older cohort
Age: years; M (SD)	10.81 (0.66)	15.10 (0.83)
Gender: male; %(n)	47.0 (481)	44.5 (388)
Migration background: yes; %(n)	57.1 (584)	55.6 (484)
School type: lower track school; %(n)	41.8 (428)	45.8 (399)

Notes. Descriptive statistics refer to t1 characteristics (both cohorts). Younger cohort (childhood): N = 1,023; older cohort (adolescence): N = 871.

Measures

Low parental supervision –poor monitoring. Parenting deficits at t1 were assessed using a modified self-report version of the Alabama Parenting Questionnaire (cf. Essau, Sasagawa, & Frick, 2006; Shelton, Frick, & Wootton, 1996; modified German parent version: cf. Lösel et al., 2003). This questionnaire enables the assessment of parenting practices (in contrast to parenting attitudes; see Shelton et al., 1996). The subscale used is a five-item self-report measure for children and youths which has been derived from the parent version of the questionnaire (cf. Meinert, Kaiser, & Guzy, 2014). Overall, the items from the subscale Poor Monitoring/Low Supervision were applied and were answered in a five-point rating format ranging from never through always. Item examples include “My parents get so busy that they forget where I am and what I’m doing” and “I go out in the evening past the time I am supposed to be home”. Cronbach’s alpha is satisfactory (i.e., α= 0.75, 5th grade; α= 0.71, 9th grade; cf. Arnis, 2015).

Criminogenic exposure –delinquent peers. Additionally, in order to operationalize the deviant and delinquent behavior of peers (t1), we employed a measure according to the CrimoC study (e.g., see Boers & Reinecke, 2007) and PADS+ (Peterborough Adolescent and Young Adult Development Study; e.g., see Wikström et al., 2012). According to the PADS+ scale Peer Crime Involvement and CrimoC delinquency items, the measure refers to the delinquency (e.g., burglary) and deviance (drug use) of peers (cf. Meinert et al., 2014). This measure encompasses seven items covering the frequencies of committing these specific delinquent and deviant acts in a five-point rating format ranging from never to very often. Participants had to answer the question of how often it happened that some of their friends committed the mentioned acts. Item examples are the following: “Does it often happen that some of your friends steal a bike?”, “Does it often happen that some of your friends use drugs?”. According to Arnis (2015), the Cronbach’s alpha coefficients were acceptable (i.e., α= 0.86 [5th grade] and α= 0.85 [9th grade]; cf. Arnis, 2015).

Propensity –self-control, morality. To operationalize a core construct of SAT (e.g., Wikström, 2006), we formed an index of propensity comprising self-control and morality (t1). Items concerning self-control were applied utilizing the German version of the Grasmick Scale (Grasmick, Tittle, Bursik Jr., & Arneklev, 1993; German version: Eifler & Seipel, 2001). The ten items from the five-point subscales Risk Behavior, Impulsivity, Temper, and Simple Tasks range from strongly disagree to strongly agree (cf. Meinert et al., 2014). Item examples are “Sometimes I will take a risk just for the fun of it” and “I never think about what will happen to me in the future”. Due to the fact that high values originally capture low levels of self-control, items were recoded. Cronbach’s alphas were satisfactory: α= 0.71 (5th grade) and α= 0.75 (9th grade; cf. Arnis, 2015). Other items refer to morality (i.e., the evaluation of deviant behavior of peers). A scale by Wikström et al. (2012) was administered comprising 16 items which are answered on a five-point Likert scale ranging from not wrong at all through very wrong (cf. Meinert et al., 2014). Introductorily, students were asked the following: “How serious do you think it is, when someone at your age does the following?”. Subsequently, the following items (examples) should be evaluated: “Get drunk with friends on a Friday evening” and “Use a weapon or force to get money or things from another young person”. The Cronbach’s alpha coefficients were quite good: α = 0.94 (5th grade) and α = 0.89 (9th grade; Arnis, 2015). Overall, we built an index of propensity by summing up the z-standardized scale scores of self-control and morality. Correspondingly, in the current work, high propensity scores refer to high levels of self-control and morality.

Empathy. A measure of empathy was derived from a personality questionnaire called FEPAA (“Fragebogen zur Erfassung von Empathie, Prosozialität, Aggressionsbereitschaft und aggressivem Verhalten”; Lukesch, 2006; cf. Meindl, 1998), which is designed to measure empathy, prosocial behavior, disposition for aggression, and aggressive behavior. In the current work, eight dichotomous items from the subscale Empathy were used (t1), considering a low level of empathy as a single aspect of CU traits. At large, four cases (i.e., vignettes) were presented (cf. Meinert et al., 2014). Students had to answer questions on certain everyday situations assessing the thoughts and feelings of peers, for example: “The following questions ask about how you assess certain everyday situations. Please mark with a cross how you assess the thoughts and feelings of peers and how you would act: ‘Felix has a new mobile. He shows it to his friend Lukas. Lukas would like to try it out. When Lukas takes the mobile, he stumbles. The mobile falls to the floor and is scratched. How does Felix feel when he sees that his mobile is scratched? How does Lukas feel?”’. Internal consistency (Cronbach’s alpha) of this scale was relatively low: α= 0.27 for 5th grade and α= 0.28 for 9th grade (Arnis, 2015). However, we considered the scale useful because of its high face validity.

Antisocial behavior –physical aggression and delinquency/destroying things. Social behavior problems one year later (t2) were measured through a German adaptation of the Social Behavior Questionnaire (SBQ; Tremblay, Vitaro, Gagnon, Piché, & Royer, 1992; German parent version: Lösel, Beelmann, & Stemmler, 2002; here: modified self-report version, cf. Meinert et al., 2014). Within this study we used the secondary scale Conduct Disorder (i.e., a broad composite score that contains the two primary scales Physical Aggression [item example: “I kick, bite, and hit others”] and Delinquency/Destroying Things [item example: “I destroy things belonging to others”]). As the label “conduct disorder” can be deceptive in a non-clinical sample, we commonly use the label Physical Aggression and Delinquency/Destroying Things or, in the context of the current work, the shorter label Antisocial Behavior for this aggregate scale. Thus, this scale is predictive of conduct disorder, however, it is not clinical in terms of providing a clinical diagnosis. The 12 items are answered in a three-point rating format ranging from never/not true to almost always/true most of the time. This secondary scale showed acceptable internal consistency (Cronbach’s alpha) in the present samples (i.e., α = 0.84 for 6th grade and α = 0.87 for 10th grade).

Results

Data were modeled utilizing path models using the software program Mplus version 8.2 (Muthén & Muthén, 1998–2018). Model fit was evaluated using Model Chi-Square, Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR) and Comparative Fit Index (CFI). We estimated both models using the relatively strict option listwise deletion of cases. Though not excessive, missing data had to be considered. In order to decrease the dropout rate for the path analyses, we again estimated both models using the Mplus default Full Information Maximum Likelihood (FIML). These additionally conducted, alternative analyses provided a general consolidation of our results, however, these analyses as well as cross-sectional analyses (t1) are not presented here (the respective analyses may be requested from the corresponding author). In addition, the results of the multiple group analysis (also using listwise deletion of cases) of the path model are presented. Further information on multiple group analysis can be obtained from Reinecke (2014).

Developmental Path Models

Developmental path model for the younger cohort –childhood. First, a model for the younger cohort (N = 911) was estimated (see Fig. 2 ): Relating to our longitudinal results, antisocial behavior (i.e., physical aggression and delinquency/destroying things) at t2 appears to be the outcome of the criminogenic exposure (i.e., delinquent peers) at t1. Analyses revealed that high propensity (i.e., high self-control and high moral standards) seems to reduce the risk for later antisocial behavior (β= –0.23, p < 0.001; this section refers to the standardized coefficients that are also displayed in Fig. 2 ). In addition, the association between having delinquent peers and propensity reached statistical significance (β= –0.29, p < 0.001). There is a positive effect of low parental supervision (i.e., poor monitoring) on antisocial behavior one year later (i.e., β= 0.18, p < 0.001), and low parental supervision seems to increase the risk for socializing with delinquent peers and subsequently seems to increase the risk of antisocial behavior. The indirect (mediated) effects of low parental supervision on antisocial behavior have been requested. The corresponding betas are 0.06 (indirect effect via delinquent peers) and 0.10 (indirect effect via propensity), both attaining statistical significance (p < 0.001). The indirect effect of empathy on antisocial behavior via propensity is lower (i.e., β= –0.01, p < 0.05). The regression relationship between empathy and delinquent peers is not significant and not considered here. Further, a negative regression relationship between propensity and low parental supervision (β= –0.42, p < 0.001) and a negative correlational relationship between empathy and low parental supervision (β= –0.16, p < 0.001) are observed. Results indicate that there is a regression relationship between propensity and empathy in the younger sample (β= 0.06, p < 0.05).

The corresponding model fit is very good. The RMSEA value of 0.016 (90%-CI = [0.000; 0.070]) indicates a close fit. Hence, the specified path model fitted the data very well. Generally, RMSEA is bounded below by zero and a value of about 0.05 or less indicates a close fit (Browne & Cudeck, 1992); according to Hooper, Coughlan, and Mullen (2008), “[. . .] in a well-fitting model the lower limit is close to 0 while the upper limit should be less than 0.08” (p. 54). Relating to the current work, the other fit statistics for the younger cohort’s model are the following: χ² = 2.483 (df = 2), p = 0.2890; SRMR = 0.012; CFI = 0.999.

Fig. 2

Developmental path model for the younger cohort –childhood.

Developmental path model for the older cohort –adolescence. A path model for the older cohort (N = 781) was estimated (see Fig. 3 ). Relating to this path model, antisocial behavior (i.e., physical aggression and delinquency/destroying things) at t2 seems to be the outcome of the criminogenic exposure (i.e., delinquent peers) at t1. Moreover, a negative effect of propensity (i.e., high self-control and high moral standards) on antisocial behavior was observed (β= –0.22, p < 0.001; in this section we refer to the standardized coefficients that are also displayed in Fig. 3 ), suggesting that high propensity reduces the risk for antisocial behavior one year later. There is a negative association between having delinquent peers and propensity (β= –0.39, p < 0.001). Results also suggest that low parental supervision (i.e., poor monitoring) has direct and indirect effects on antisocial behavior, and low parental supervision might increase the risk for having delinquent peers and the risk of antisocial behavior one year later. The regression relationship between antisocial behavior and low parental supervision (i.e., β= 0.08) also reached the level of significance (p < 0.05). The indirect (mediated) effects of low parental supervision on antisocial behavior are statistically significant (p < 0.001; β= 0.06 [indirect effect via delinquent peers] and β= 0.11 [indirect effect via propensity]). The indirect effects of empathy on antisocial behavior are lower (i.e., both β= –0.02, each p < 0.01; indirect effect via delinquent peers; indirect effect via propensity). In addition, low parental supervision is associated with both propensity (negative regression relationship; β= –0.50, p < 0.001) and empathy (negative correlational relationship; β= –0.14, p < 0.001). In the older cohort, again, propensity seems to be associated with empathy (β= 0.10, p < 0.01), indicating that high levels of self-control and morality are regressed on high empathy. In the older cohort, a negative regression relationship between delinquent peers and empathy is observed (β= –0.14, p < 0.001).

Fig. 3

Developmental path model for the older cohort – youth.

Overall, the model fit was assessed. Concerning the model for the older cohort, the RMSEA value of 0.000 (90%-CI = [0.000; 0.053]) indicates a very good fit (see above). Altogether, the fit statistics for the older cohort’s model are: χ²= 0.036 (df = 1), p = 0.8501; SRMR = 0.002; CFI = 1.000.

Multiple Group Analysis

The following table (Table 2) lists the results of the multiple group analysis of the path model, starting from the basic model (maximum restrictions) with 14 parameters to variant 6 (no restrictions) with 28 parameters. Variant 1 only frees the variances and covariances of the exogenous variables (three parameters: Ψ₁₁, Ψ₂₂, Ψ₂₁), variant 2 additionally frees the residual variances of the other variables including the residual covariance between propensity and criminogenic exposure (additionally four parameters: Ψ₃₃, Ψ₄₄, Ψ₅₅, Ψ₄₃). Variant 3 additionally frees the path between empathy and criminogenic exposure across the groups (β₁₅), while variant 4a additionally frees the path between low parental supervision and antisocial behavior (β₃₄). Alternatively, in variant 4b the path between criminogenic exposure and antisocial behavior (β₃₁) is freed. Both paths prove to be significantly different, independent of each other, so that in variant 5 both are freed simultaneously (β₁₅, β₃₁, β₃₄). If the remaining four paths are freed in variant 6 across both groups, then the result of the χ² difference test is not significant. Variant 5 can therefore be accepted as the most parsimonious model.

Table 2

Results of the Multiple Group Analysis of the Path Model, Starting from the Basic Model (Maximum Restrictions) with 14 Parameters to Variant 6 (no Restrictions) with 28 Parameters

Model	Group	χ²	df	χ²_Diff	df _Diff	RMSEA	p-value	SRMR
Basic model	1	109.98
Matrices Ψ, B	2	82.53
constrained to be equal	Σ	192.51	16	–	–	0.114	0.000	0.152
Variant 1	1	88.18
Ψ₁₁, Ψ₂₂, Ψ₂₁	2	60.13
freed	Σ	148.30	13	44.21	3	0.111	0.000	0.123
Variant 2	1	23.57
Matrix Ψ	2	27.56
completely freed	Σ	51.14	9	97.16	4	0.074	0.019	0.051
Variant 3	1	19.79
β₁₅	2	15.51
freed	Σ	35.30	8	15.84	1	0.064	0.013	0.043
Variant 4a	1	10.52
β₁₅, β₃₄	2	9.33
freed	Σ	19.85	7	15.45	1	0.047	0.549	0.030
Variant 4b	1	11.59
β₁₅, β₃₁	2	11.26
freed	Σ	22.84	7	12.46	1	0.052	0.409	0.033
Variant 5	1	7.24
β₁₅, β₃₁, β₃₄	2	7.42
freed	Σ	14.66	6	20.64	2	0.041	0.662	0.030
Variant 6	1	2.48
Matrix B	2	0.04
completely freed	Σ	2.52	2	12.14	4	0.017	0.778	0.009

Notes. Group 1 = younger cohort. Group 2 = older cohort. Matrix Psi: The elements are variances/covariances/residual covariances of the variables. Matrix B: The elements are path (regression) coefficients of the model.

Table 3 shows the result of the multiple group analysis of variant 5 in relation to the path coefficients (unstandardized coefficients). Three of the seven coefficients across the two groups are freed: the effect of delinquent peers on antisocial behavior (stronger for the younger cohort), the effect of low parental supervision on antisocial behavior (stronger for the younger cohort), and the effect of empathy on delinquent peers (not significant for the younger cohort, negative effect for the older cohort). While the relationships between empathy and antisocial behavior are completely mediated by propensity and exposure, a direct effect from low parental supervision on antisocial behavior can be specified that is significant in both groups. Table 3 also shows the result of the decomposition of the effect with the unstandardized coefficients (total and indirect effects of low parental supervision on antisocial behavior). The indirect effect is stronger in the younger cohort than in the older cohort.

Table 3

Result of the Multiple Group Analysis of Variant 5 in Relation to the Path Coefficients (Unstandardized Coefficients) and Result of the Effect Decomposition with the Unstandardized Coefficients (Total and Indirect Effects of Low Parental Supervision on Antisocial Behavior)

	Group 1		Group 2
	Est.	S.E.	Est.	S.E.
Multiple group analysis in relation to the path coefficients
Antisocial behavior ON
Delinquent peers	0.091	0.015	0.047	0.012
Low parental supervision	0.060	0.011	0.020	0.010
Propensity	–0.031	0.004	–0.031	0.004
Delinquent peers ON
Low parental supervision	0.260	0.017	0.260	0.017
Empathy	0.011	0.092	–0.680	0.153
Propensity ON
Low parental supervision	–0.981	0.046	–0.981	0.046
Empathy	0.822	0.235	0.822	0.235
Effect decomposition
Total	0.114	0.010	0.063	0.009
Total indirect	0.054	0.005	0.043	0.004

Notes. Group 1 = younger cohort. Group 2 = older cohort.

Table 4 comprises the results of the multiple group analysis of variant 5 with the means and the intercepts, respectively (model and parameters). Three of five values vary across both groups (low parental supervision, delinquent peers, propensity), two values are invariant across both groups (empathy and antisocial behavior). Low parental supervision is more pronounced in the older cohort than the younger cohort, as expected, the values for delinquent peers and propensity also are higher in the older cohort. The values of empathy and antisocial behavior do not differ in the two groups.

Table 4

Result of the Multiple Group Analysis of Variant 5 with the Means/Intercepts: Model and Parameters

Model	Group	χ²	df	χ²_Diff	df _Diff	RMSEA	p-value	SRMR
Variant 5	1	9.68
(means/intercepts)	2	9.86
β₁₅, β₃₁, β₃₄ freed	Σ	19.53	8	–	–	0.041	0.699	0.033
Parameters	Group 1		Group 2
	Est.	S.E.	Est.	S.E.
Low parental supervision	1.576	0.022	2.259	0.029
Empathy	0.788	0.003	0.788	0.003
Delinquent peers	0.808	0.082	1.483	0.131
Propensity	0.937	0.214	1.598	0.230
Antisocial behavior	0.939	0.018	0.939	0.018

Notes. Group 1 = younger cohort. Group 2 = older cohort.

Discussion

Overall, our findings are in agreement with international criminological research. First, the results are in accordance with psychological-criminological research suggesting that there are associations bet-ween individual risk and antisocial behavior. Ind-ividual risk has direct and indirect influences on later antisocial behavior supporting risk factor models (e.g., Lösel & Bender, 2003) and Moffitt’s theory (cf. Moffitt, 1993; see also Moffitt, 2018). Hence, these findings are consistent with prior findings on developmental criminology issues (e.g., cf. Corrado, 2012; Farrington, 2005; Farrington, Piquero, & Jennings, 2013; Wallner et al., 2018). In general, the psychological-criminological literature mentioned above suggests that parenting deficits (e.g., poor mon-itoring) are associated with delinquent peer contacts and antisocial behavior. Referring to our study, low parental supervision might increase the risk for socializing with delinquent peers and subsequently might increase the risk of antisocial behavior: results are in accordance with findings from the literature. Overall, low parental supervision seems to be less relevant for antisocial behavior for the adolescent sample. The results for the younger sample reveal a clear positive effect of low parental supervision on antisocial behavior one year later, confirming the assumption that parenting deficits (e.g., low parental supervision) might be related to antisocial outcomes in childhood (e.g., cf. Corrado, 2002; Lösel & Bender, 2003). Concerning the result of the multiple group analysis in relation to the path coefficients, the effect of low parental supervision on antisocial behavior is stronger for the younger cohort than for the older cohort. Related to the result of the multiple group analysis with means/intercepts, low parental supervision is more pronounced in the older cohort than in the younger cohort. As expected, the level of parental control appears to be lower in adolescence than in childhood, where there is a stronger association between low parental supervision and antisocial behavior. Relatively strong negative, cross-sectional relations between propensity (i.e., high self-cont-rol and high moral standards) and low parental sup-ervision and clear negative, cross-sectional relationships between empathy and low parental supervision might indicate the importance of contextual parenting deficits for both propensity and empathy (in both cohorts). Correspondingly, the single individual characteristics (i.e., low self-control, low moral standards, and low empathy in childhood and adolescence) are associated with adverse parenting practices based on poor monitoring by parents. Furthermore, there are rather modest regression relationships between propensity and empathy in both cohorts. Since the literature suggests that high levels of CU traits are associated with severe antisocial behavior (e.g., cf. Frick et al., 2014), we considered an empathy measure as an additional predictor in our path models for both cohorts, capturing a single aspect of CU traits (see below). At large, the chosen risks seem to be meaningful predictors of antisocial behavior one year later, so that the results confirm the findings of previous empirical studies on the relations between psycho-social risks and antisocial behavior problems. Second, in line with our sociological-criminological assumptions, there seems to be evidence for SAT (e.g., Wikström, 2009). Referring to the described results, antisocial behavior appears to be the outcome of propensity and criminogenic exposure (i.e., delinquent peers). With regard to the “causes of the causes” discussion held within SAT (Wikström et al., 2012), delinquent peers and propensity are used simultaneously as mediators. In this context, a direct effect of low parental supervision on antisocial behavior (see above) that is significant in both cohorts is to be specified, while the relationships between empathy and antisocial behavior are completely mediated by propensity and delinquent peers. The additional model with means/intercepts (multiple group analysis with means/intercepts) shows interesting results. Although, as expected, the intercepts of propensity and exposure are higher in the older cohort, the path coefficients on antisocial behavior do not necessarily have to be stronger. The result of the multiple group analysis in relation to the path coefficients suggests that the effect of delinquent peers on antisocial behavior is stronger for the younger cohort. Relating to the development of antisocial behavior, this finding possibly suggests that having delinquent peers might be more critical in childhood, whereas delinquent peer contacts might be more normative in adolescence. Delinquent peers are a risk factor mainly related to adolescence (e.g., cf. Lösel & Bender, 2003; Moffitt, 2018). While the effect of empathy on delinquent peers expectedly is negative for the older cohort, the regression relationship between delinquent peers and empathy was not statistically significant in the younger cohort. Concerning the multiple group analysis with means/intercepts, however, the values of empathy and antisocial behavior do not differ in the two age cohorts. Empirical literature suggests that CU traits and impulsivity are predictors of conduct disorder symptoms in childhood and, especially, in adolescence (e.g., cf. Fanti et al., 2018). With regard to both samples of the present study, empathy is positively related to propensity, which is in turn negatively related to antisocial behavior one year later. Hence, our analyses provided evidence for a relatively close (longitudinal) connection between these variables. According to SAT and, specifically, according to violence as situational action, Wikström and Treiber (2009) emphasized the importance of both the “emotional involvement in the violent course of action” and the capability of “inhibiting, or compensating for, that emotional impetus and exhibiting self-control” (pp. 87-88). The considered single psychopathic feature seems to be a crucial individual characteristic in the emergence of antisocial behavior, especially in adolescence. Overall, the path model showed a very good fit (younger and older cohort). Both the considered psychological-criminological approach and the sociological-criminological approach, that to some extent complement one another, consider important individual and environmental characteristics relating to the emergence of antisocial behavior, and our models based on these crucial characteristics fit the data closely. However, developmental models contributing to the explanation of antisocial behavior in childhood and adolescence have to be refined continuously in the context of longitudinal studies. Therefore, continuative developmental research especially on age-specific causes of antisocial behavior is needed that might further extend our findings and admit even more explicit results with regard to the partly differing underlying relationships.

Limitations

Some limitations of the current study have to be mentioned. First, the specific sample composition has to be considered when interpreting results. The Nuremberg sample only comprises students from lo-wer-track schools, while the Dortmund sample is composed of a broader range of school types, so that conclusions relating to the population are not possible on grounds of these unweighted data (cf. Wallner & Weiss, 2019). A further issue that needs to be mentioned is the attrition rate in our study sample. Often the most problematic cases drop out over time. Indeed, we account our data relatively robust regarding the issue of dropout (see methods section), however, we cannot completely remove this issue from consideration. Also, the analyses of the current work do not include sociodemographic variables (e.g., gender, migration background, socio-economic status, school type). Relating to other empirical work of our research project, for example, Uysal, Stemmler, and Weiss (2019) studied antisocial behavior and violence among boys with a migration background. Especially gender differences in the associations be-tween antisocial outcomes and different predictive variables should be further examined as well; therefore, future developmental studies should also test gender-specific models. Indeed, literature suggests lower prevalence rates for girls relating to different antisocial outcomes (e.g., violent offending) than for boys, however, concerning the importance of risk factors a more general perspective might be largely legitimated (cf. Moffitt, Caspi, Rutter, & Silva, 2001). Because of constraints concerning available indicators for the relevant predictor variables in our study, more complex models including latent variables could not be implemented. Another important issue pertains to the selection of variables providing just a small portion of antecedents of antisocial behavior, so there were issues that we could not consider in our study. Additional psychological, sociological, and biological risk factors (e.g., cf. Farrington et al., 2016; Stemmler et al., 2018) as well as protective factors (e.g., see Lösel & Farrington, 2012) should be incorporated, assisting in further analyses on antisocial development and elucidating the underlying developmental patterns. Several individual risks may be additionally important, such as sensation-seeking, low educational achievement, negative emotions (e.g., based on strain), inappropriate self-esteem (e.g., self-derogation or egocentricity) or others. Criminogenic exposure may also be represented through social disorganization in one’s neighborhood, antisocial parents, and other aspects. In accordance with the aims of the current work, attention should particularly be paid to a joint examination of additional meaningful psychological and sociological variables. The specific impact of risks could be further investigated in more detail, predicting antisocial behavior differently in childhood and adolescence. Additional limitations referring to the utilized measures capturing different antisocial facets may be noted: Importantly, the risk variable delinquent peers and the outcome antisocial behavior partly allude to similar aspects being related to antisocial problem behavior, so that, to a certain extent, identical information is incorporated in both measures. Furthermore, antisocial behavior (i.e., physical aggression and delinquency/destroying things) has been considered as outcome measure in the current analyses, though other outcome measures would have been alternatively supposable (e.g., general offending, violence, deviant behavior, externalizing behavior, bullying).

Implications

A strength of the present research is the use of a longitudinal sample. The current longitudinal study augments our understanding of the concurrence of different predictors of later antisocial behavior, exa-mining two different age cohorts over time. Therefore, the results of our analyses should be useful for future research on specific risk combinations related to childhood and youth age in the development of antisocial behavior. In a nutshell, studies on antisocial behavior in childhood and adolescence should first take into consideration both individual and contextual characteristics and, hence, combine psychological-criminological and sociological-criminological constructs to increase the explained variable. Studies should then integrate direct and indirect effects as well as reciprocal effects over time, investigating developmental risk patterns and antisocial outcomes for different ages within the framework of developmental criminology. With that in mind, future ana-lyses might at least incorporate data concerning the early and/or middle childhood and early adolescence period, facilitating more extensive and precise findings concerning the emergence of antisocial behavior earlier in life (e.g., Corrado, 2002) or the termination of antisocial behavior (e.g., focusing on the specific transitions from adolescence to adulthood; e.g., see Sampson & Laub, 2016). Regarding developmental and life-course criminology issues (e.g., cf. Boers, Lösel, & Remschmidt, 2009; Farrington et al., 2013), additional longitudinal analyses relating to longer prediction periods might be fruitful.

Author notes

This article considers findings from the longitudi-nal study “Chances and Risks in the Life Course” (CURL; “The Emergence and Development of Deviant and Delinquent Behavior over the Life Course and its Significance for Processes of Social Inequality”; e.g., Reinecke et al., 2013; Reinecke, Stemmler, & Wittenberg, 2016; Wallner et al., 2019). The study was part of the Collaborative Res-earch Center (“Sonderforschungsbereich”, SFB) 882 “From Heterogeneities to Inequalities”, which was established at Bielefeld University, Germany, and which was funded by the German Research Foundation (“Deutsche Forschungsgemeinschaft”, DFG) until 2016. We wish to thank Maria Arnis, Nihad El-Kayed, Eva Link, Julia Meinert, Andreas Pöge, Debbie Schepers, Burcu Uysal, Maren Weiss, and Jochen Wittenberg for their valuable contributions to the research project. We also gratefully acknowledge the cooperation of the schools and students participating in the CURL study.

Bio Sketches

Dr. Susanne Wallner, Ph.D. in Psychology since 2007 and Research Associate at the University of Erlangen-Nuremberg, Germany. Research on developmental psychopathology, criminology, and risk/needs assessment.

Mark Stemmler, Ph.D., Professor of Psychology since 2007, and, since 2011, Professor in Psychological Assessment at the University of Erlangen-Nuremberg, Germany. Research on quantitative methods, developmental psychopathology, prevention, and criminal psychology.

Dr. Jost Reinecke, Professor of Quantitative Methods of Empirical Social Research at Bielefeld University, Germany. Research on quantitative methods, statistical analysis of structural equation models, multiple imputation, longitudinal adolescents’ criminology, and measurements of social cohesion.

References

American Psychiatric Association (2013). Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. Arlington, VA: American Psychiatric Association.

Andershed,

, Colins,

O. F.

, Salekin,

R. T.

, Lordos,

, Kyranides,

M. N.

, Fanti,

K. A.

(2018). Callous-unemotional traits only versus the multidimensional psychopathy construct as predictors of various antisocial outcomes during early adolescence. Journal of Psychopathology and Behavioral Assessment, 40, 16–25. doi: 10.1007/s10862-018-9659-5

Arnis,

(2015). Die Entstehung und Entwicklung devianten und delinquenten Verhaltens im Lebensverlauf und ihre Bedeutung fur soziale Ungleichheitsprozesse: Fragebogendokumentation der Schülerbefragung in Dortmund und Nürnberg. Band 2: Skalendokumentation. Erste Erhebungswelle, 2012. [The emergence and development of deviant and delinquent behavior over the life course and its significance for processes of social inequality: Questionnaire documentation of the student survey in Dortmund and Nuremberg. Volume 2: Scale documentation. First survey wave, 2012.] (SFB 882 Technical Report Series, 18). Bielefeld: DFG Research Center (SFB) 882 From Heterogeneities to Inequalities.

Bergstrøm,

, Farrington,

D. P.

(2018). Grandiose-manipulative, callous-unemotional, and daring-impulsive: The prediction of psychopathic traits in adolescence and their outcomes in adulthood. Journal of Psychopathology and Behavioral Assessment, 40, 149–158. doi: 10.1007/s10862-018-9674-6

Boers,

, Lösel,

, Remschmidt,

(2009). Entwicklungskriminologie und kriminologische Lebenslaufforschung (Sonderheft). [Developmental criminology and criminological life course research (special issue).]Monatsschrift fur Kriminologie und Strafrechtsreform, 2 (3).

Boers,

, Reinecke,

(2007). Delinquenz im Jugendalter. Erkenntnisse einer Münsteraner Längsschnittstudie. [Delinquency during adolescence. Findings from a longitudinal study in Münster.] Munster: Waxmann.

Browne,

M. W.

, Cudeck,

(1992). Alternative ways of assessing model fit. Sociological Methods & Research, 21, 230–258. doi: 10.1177/0049124192021002005

Byrd,

A. L.

, Loeber,

, Pardini,

D. A.

(2012). Understanding desisting and persisting forms of delinquency: The unique contributions of disruptive behavior disorders and interpersonal callousness. Journal of Child Psychology and Psychiatry, 53, 371–380. doi: 10.1111/j.1469-7610.2011.02504.x

Cooke,

D. J.

, Michie,

(2001). Refining the construct of psychopathy: Towards a hierarchical model. Psychological Assessment, 13, 171–188. doi: 10.1037111040-3590.13.2.171

10.

Corrado,

R. R.

(2002). An introduction to the risk/needs case management instrument for children at risk for violence. In R. R. Corrado, R. Roesch, S. D. Hart, & J. K. Gierowski (Eds.), Multi-problem violent youth: A foundation for comparative research on needs, interventions and outcomes (pp. 295–301). Amsterdam: IOS Press.

11.

Corrado,

R. R.

(2012). Comprehensive risk management instruments for serious and violent young offenders: Challenges and advantages. In T. Bliesener, A. Beelmann, & M. Stemmler (Eds.), Antisocial behavior and crime: Contributions of developmental and evaluation research to prevention and intervention (pp. 239–260). Göttingen: Hogrefe.

12.

Corrado,

R. R.

, DeLisi,

, Hart,

S. D.

, McCuish,

E. C.

(2015). Can the causal mechanisms underlying chronic, serious, and violent offending trajectories be elucidated using the psychopathy construct?Journal of Criminal Justice, 43, 251–261. doi: 10.1016/j.jcrimjus.2015.04.006

13.

Eifler,

, Seipel,

(2001). Eine deutsche Version einer Skala zur Erfassung von Self-Control. [A German version of a scale for assessing self-control.] (Bielefelder Arbeiten zur Sozialpsychologie, 199). Bielefeld: Universität Bielefeld, Fakultät für Soziologie.

14.

Essau,

C. A.

, Sasagawa,

, Frick,

P. J.

(2006). Psychometric properties of the Alabama Parenting Questionnaire. Journal of Child and Family Studies, 15, 597–616. doi: 10.1007/s10826-006-9036-y

15.

Fanti,

K. A.

, Kyranides,

M. N.

, Lordos,

, Colins,

O. F.

, Andershed,

(2018). Unique and interactive associations of callous-unemotional traits, impulsivity and grandiosity with child and adolescent conduct disorder symptoms. Journal of Psychopathology and Behavioral Assessment, 40, 40–49. doi: 10.1007/s10862-018-9655-9

16.

Farrington,

D. P.

(2005). Childhood origins of antisocial behavior. Clinical Psychology and Psychotherapy, 12, 177–190. doi: 10.1002/cpp.448

17.

Farrington,

D. P.

, Piquero,

A. R.

, Jennings,

W. G.

(2013). Offending from childhood to late middle age: Recent results from the Cambridge Study in Delinquent Development. New York: Springer.

18.

Farrington,

D. P.

, Ttofi,

M. M.

, Piquero,

A. R.

(2016). Risk, promotive, and protective factors in youth offending: Results from the Cambridge Study in Delinquent Development. Journal of Criminal Justice, 45, 63–70. doi: 10.1016/j.jcrimjus.2016.02.014

19.

Fox,

B. H.

, Jennings,

W. G.

, Farrington,

D. P.

(2015). Bringing psychopathy into developmental and life-course criminology theories and research. Journal of Criminal Justice, 43, 274–289. doi: 10.1016/j.jcrimjus.2015.06.003

20.

Frick,

P. J.

(2012). Developmental pathways to conduct disorder: Implications for future directions in research, assessment, and treatment. Journal of Clinical Child & Adolesc-ent Psychology, 41, 378–389. doi: 10.1080/15374416.2012.664815

21.

Frick,

P. J.

, Hare,

(2001). Antisocial Process Screening Device (APSD). Toronto, ON: Multi-Health Systems.

22.

Frick,

P. J.

, Ray,

J. V.

, Thornton,

L. C.

, Kahn,

R. E.

(2014). Can callous-unemotional traits enhance the understanding, diagnosis, and treatment of serious conduct problems in children and adolescents? A comprehensive review. Psychological Bulletin, 140, 1–57. doi: 10.1037/a0033076

23.

Frick,

P. J.

, Stickle,

T. R.

, Dandreaux,

D. M.

, Farrell,

J. M.

, Kimonis,

E. R.

(2005). Callous-unemotional traits in predicting the severity and stability of conduct problems and delinquency. Journal of Abnormal Child Psychology, 33, 471–487. doi: 10.1007/s10648-005-5728-9

24.

Frick,

P. J.

, White,

S. F.

(2008). Research review: The importance of callous-unemotional traits for developmental models of aggressive and antisocial behavior. Journal of Child Psychology and Psychiatry, 49, 359–375. doi: 10.1111/j.1469-7610.2007.01862.x

25.

Grasmick,

H. G.

, Tittle,

C. H.

, Bursik

R.J.

Jr , Arneklev,

B. J.

(1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research in Crime and Delinquency, 30, 5–29. doi: 10.1177/0022427893030001002

26.

Hooper,

, Coughlan,

, Mullen,

(2008). Structural equation modelling: Guidelines for determining model fit. Electronic Journal of Business Research Methods, 6, 53–60.

27.

Jöreskog,

K. G.

(1993). Testing structural equation models. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 294–316). Newbury Park, California: Sage.

28.

Kimonis,

E. R.

, Fanti,

K. A.

, Frick,

P. J.

, Moffitt,

T. E.

, Essau,

, Bijttebier,

, Marsee,

M. A.

(2015). Using self-reported callous-unemotional traits to cross-nationally assess the DSM- ‘With Limited Prosocial Emotions’ specifier. Journal of Child Psychology and Psychiatry, 56, 1249–1261. doi: 10.1111/jcpp.12357

29.

Lösel,

, Beelmann,

, Jaursch,

, Scherer,

, Stemmler,

, Wallner

(2003). Skalen zur Messung elterlichen Erziehungsverhaltens bei Vorschul- und Grundschulkindern. Die deutschen Versionen der Parenting Sense of Competence Scale (PSOC), der Parenting Scale (PARS) und des Alabama Parenting Questionnaire (APQ). [Scales for assessing parenting behavior among preschool and primary school children. The German versions of the Parenting Sense of Competence Scale (PSOC), the Parenting Scale (PARS), and the Alabama Parenting Questionnaire (APQ).] Universität Erlangen-Nürnberg: Institut für Psychologie.

30.

Lösel,

, Beelmann,

, Stemmler,

(2002). Skalen zur Messung sozialen Problemverhaltens bei Vorschul- und Grundschulkindern. Die deutschenVersionen des Eyberg Child Behavior Inventory (ECBI) und des Social Behavior Questionnaire (SBQ). [Scales for assessing social problem behavior in preschool and primary school children. The German versions of the Eyberg Child Behavior Inventory (ECBI) and the Social Behavior Questionnaire (SBQ).] Universität Erlangen-Nürnberg: Institut für Psychologie.

31.

Lösel,

, Bender,

(2003). Protective factors and resilience. In D. P. Farrington & J. W. Coid (Eds.),Early prevention of adult antisocial behavior (pp. 130–204). Cambridge: Cambridge University Press.

32.

Lösel,

, Farrington,

D. P.

(2012). Direct protective and buffering protective factors in the development of youth violence. American Journal of Preventive Medicine, 43 (2S1), 8–23. doi: 10.1016/j.amepre.2012.04.029

33.

Lösel,

, Stemmler,

(2012). Preventing child behavior problems in the Erlangen-Nuremberg Development and Prevention Study: Results from preschool to secondary school age. International Journal of Conflict and Violence, 6, 214–224. doi: 10.4119/UNIBI/ijcv.278

34.

Lukesch,

(2006). FEPAA. Fragebogen zur Erfassung von Empathie, Prosozialitat, Aggressionsbereitschaft und aggressivem Verhalten. [FEPAA. Questionnaire to assess empathy, prosociality, disposition for aggression, and aggressive behavior.] Göttingen: Hogrefe.

35.

McCuish,

E. C.

, Corrado,

R. R.

, Hart,

S. D.

, DeLisi,

(2015). The role of symptoms of psychopathy in persistent violence over the criminal career into full adulthood. Journal of Criminal Justice, 43, 345–356. doi: 10.1016/j.jcrimjus.2015.04.008

36.

Meindl,

(1998). Entwicklung und Validierung eines Fragebogens zur Erfassung von Empathie und seine Anwendung im Rahmen eines Empathietrainings. [Development and validation of a questionnaire to assess empathy and its application in empathy training.] Universität Regensburg: Unveröffentlichte Diplomarbeit.

37.

Meinert,

, Kaiser,

, Guzy,

(2014). Die Entstehung und Entwicklung devianten und delinquenten Verhaltens im Lebensverlauf und ihre Bedeutung fur soziale Ungleichheitsprozesse: Fragebogendokumentation der Schülerbefragung in Dortmund und Nürnberg. Band 1: Itemdokumentation. Erste Erhebungswelle, 2012. (SFB 882 Technical Report Series, 7). [The emergence and development of deviant and delinquent behavior over the life course and its significance for processes of social inequality: Questionnaire documentation of the student survey in Dortmund and Nuremberg. Volume 1: Item documentation. First survey wave, 2012.] Bielefeld: DFG Research Center (SFB) 882 From Heterogeneities to Inequalities.

38.

Moffitt,

T. E.

(1993). Adolescence-limited and life-course-per-sistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701. doi: 10.1037/0033-295X.100.4.674

39.

Moffitt,

T. E.

(2018). Male antisocial behaviour in adolescence and beyond. Nature Human Behaviour, 2, 177–186. doi: 10.1038/s41562-018-0309-4

40.

Moffitt,

T. E.

, Caspi,

, Rutter,

, Silva,

P. A.

(2001). Sex differences in antisocial behaviour: Conduct disorder, delinquency, and violence in the Dunedin Longitudinal Study. Cambridge: Cambridge University Press.

41.

Moffitt,

T. E.

et al. (2008). Research review: DSM-V conduct disorder: research needs for an evidence base. Journal of Child Psychology and Psychiatry, 49, 3–33. doi: 10.1111/j.1469-7610.2007.01823.x

42.

Muthén,

L. K.

, Muthén,

B. O.

(1998–2018). Mplus user’s guide. 8th edition. LA, CA: Muthén & Muthén.

43.

Pardini,

D. A.

, Fite,

P. J.

(2010). Symptoms of conduct disorder, oppositional defiant disorder, attention-deficit/hyperactivity disorder, and callous-unemotional traits as unique predictors of psychosocial maladjustment in boys: Advancing an evidence base for DSM-V. Journal of the American Academy of Child and Adolescent Psychiatry, 49, 1134–1144. doi: 10.1016/j.jaac.2010.07.010

44.

Reinecke,

(2014). Strukturgleichungsmodelle in den Sozialwissenschaften. [Structural equation models in the social sciences.]München: OldenbourgWissenschaftsverlag GmbH.

45.

Reinecke,

, Stemmler,

, Sünkel,

, Schepers,

, Weiss,

, Arnis,

, & Wittenberg,

, (2013). The development of deviant and delinquent behavior over the life course in the context of processes of social inequalities. (SFB 882 Working Paper Series, 17). Bielefeld: DFG Research Center (SFB) 882 From Heterogeneities to Inequalities..

46.

Reinecke,

, Stemmler,

, Wittenberg,

(Eds.) (2016). Devianz und Delinquenz im Kindes- und Jugendalter. Ungleichheitsdimensionen und Risikofaktoren. [Deviance and delinquency in childhood and adolescence. Dimensions of inequality and risk factors.] Wiesbaden: Springer VS.

47.

Sampson,

R. J.

, Laub,

J. H.

(2016). Turning points and the future of life-course criminology: Reflections on the 1986 criminal careers report. Journal of Research in Crime and Delinquency, 53, 321–335. doi: 10.1177/0022427815616992

48.

Schepers,

(2017). Causes of the causes of juvenile delinquency: Social disadvantages in the context of Situational Action Theory. European Journal of Criminology, 14, 143–159. doi: 10.1177/1477370816649622

49.

Shelton,

K. K.

, Frick,

P. J.

, Wootton,

(1996). Assessment of parenting practices in families of elementary school-age children. Journal of Clinical Child Psychology, 25, 317–329. doi: 10.1207/s15374424jccp2503_8

50.

Silva,

T. C.

, Stattin,

(2016). The moderating role of parenting on the relationship between psychopathy and antisocial behavior in adolescence. Development and Psychopathology, 28, 505–515. doi: 10.1017/S0954579415001121

51.

Stemmler,

, Wallner,

, Link,

(2018). Risikofaktoren für die Entwicklung dissozialen Verhaltens in der Kindheit und Jugend. [Risk factors for the development of dissocial behavior in childhood and adolescence.] In D. Hermann & A. Pöge (Eds.), Kriminalsoziologie: Handbuch für Wissenschaft und Praxis (pp. 247–261). Baden-Baden: Nomos Verlag.

52.

Tremblay,

, Vitaro,

, Gagnon,

, Piché,

, Royer,

(1992). A prosocial scale for the Preschool Behaviour Questionnaire: Concurrent and predictive correlates. International Journal of Behavioral Development, 15, 227–245. doi: 10.1177/016502549201500204

53.

Uysal,

, Stemmler,

, Weiss,

(2019). Antisocial behaviour and violent delinquency among boys with a migration background: A German panel study. International Journal of Developmental Science, 13, 97–108. doi: 10.3233/DEV-190272

54.

Wallner,

(2016). Psychopathische Merkmale und abweichendes Verhalten. [Psychopathic characteristics and deviant behavior.] In J. Reinecke, M. Stemmler, & J. Wittenberg (Eds.), Devianz und Delinquenz im Kindes- und Jugendalter. Ungleichheitsdimensionen und Risikofaktoren (pp. 217–229). Wiesbaden: Springer VS.

55.

Wallner,

, Lösel,

, Stemmler,

, Corrado,

R. R.

(2018). The validity of the Cracow Instrument in the prediction of antisocial development in preschool children: A five-year longitudinal community based study. International Journal of Forensic Mental Health, 17, 181–194. doi: 10.1080/14999013.2018.1460423

56.

Wallner,

, Weiss,

(2019). Prävalenz und Verlauf von Devianz und Delinquenz. [Prevalence and course of deviance and delinquency.] In S. Wallner, M. Weiss, J. Reinecke, & M. Stemmler (Eds.), Die Entwicklung von Devianz und Delinquenz in Kindheit und Jugend. Neue Ansätze der kriminologischen Forschung (pp. 39–61). Wiesbaden: Springer VS.

57.

Wallner,

, Weiss,

, Reinecke,

, Stemmler,

(Eds.) (2019). Devianz und Delinquenz in Kindheit und Jugend. Neue Ansätze der kriminologischen Forschung. [Deviance and delinquency in childhood and adolescence. New approaches in criminological research.] Wiesbaden: Springer VS.

58.

Weiss,

, Link,

(2019). Analysen zur Panelmortalität. [Analyses on panel mortality.] In S.Wallner, M.Weiss, J. Reinecke, & M. Stemmler (Eds.), Die Entwicklung von Devianz und Delinquenz in Kindheit und Jugend. Neue Ansätze der kriminologischen Forschung (pp. 63–82). Wiesbaden: Springer VS.

59.

Wikström,

P.-O. H.

(2006). Individuals, settings, and acts of crime: Situational mechanisms and the explanation of crime. In P.-O. H.Wikström & R. J. Sampson (Eds.), The explanation of crime (pp. 61–107). Cambridge: Cambridge University Press.

60.

Wikström,

P.-O. H.

(2009). Crime propensity, criminogenic exposure and crime involvement in early to mid-adolescence. Monatsschrift fur Kriminologie und Strafrechtsreform, 92, 253–266. doi: 10.1515/mks-2009-922-312

61.

Wikström,

P.-O. H.

, Oberwittler,

, Treiber,

, Hardie,

(2012). Breaking rules: The social and situational dynamics of young people’s urban crime. Oxford: Oxford University Press.

62.

Wikström,

P.-O. H.

, Treiber,

K. H.

(2009). Violence as situational action. International Journal of Conflict and Violence, 3, 75–96. doi: 10.4119/UNIBI/ijcv.49