Do Specific Combinations of Parent–Child Relationships Predict Moral Values?

ABSTRACT

In this study we examined to what extent different combinations of parent–child relationships explain adolescents’ moral values and whether the influence of school and peers predicts moral values. We also investigated whether anticipated shame and anticipated guilt predict moral values. A convenience sample of 1,120 adolescents, including 120 adolescents in youth custody, was used. Results suggest that combinations of parent–child relationships do not predict moral values, that school and peers are significant predictors of adolescents’ moral values, and anticipated shame and guilt do not consistently and equally predict moral values. Results were highly similar for boys and girls.

Notes

¹ When we use the concept of socialization, we do not suggest that moral norms are mainly the product of learning processes, i.e. we reject the ‘Blank Slate view’ (Pinker 2004) and recognize the importance of biosocial interactions as ‘causes of the causes’ of crime. In line with evolutionary biology, humans’ capacity for morality can be seen as a consequence of the need to cooperate (Krebs 2011; Tomasello 2016).

² They also noted that these institutions have a different impact depending on the “person’s biological maturation and social and cognitive development” (Wikström et al. 2012: 32). Nevertheless, biological factors are not discussed in this paper.

³ Of course we recognize exceptions to this observation, e.g. in troublesome youth groups or gangs.

⁴ The discussion between scholars on this topic is confused by different conceptions of causes or mechanisms (Wikström 2007).

⁵ The percentage of respondents is difficult to compute with this method. We estimate this percentage between 10% and 15%.

⁶ Here, we indicate correlations instead of alphas. Alphas should be based on at least three items. Correlations between .70 and .90 indicate high and positive relationships between variables (Hinkle, Wiersma, and Jurs 2003).

⁷ Age is considered to be a control variable. Socioeconomic status (SES) was not taken into account as a control variable because the adolescents were often unaware of their parents’ educational level or type of job (often used to measure SES).

⁸ Note that criminological studies often use offending as the dependent variable. In the context of the Situational Action Theory, the parental socialization process cannot directly explain offending. The socialization process is part of the developmental part of the SAT. Indeed, socialization agents are indirect causes of crimes.

⁹ The scales used for LCA all are significantly correlated to each other (see Appendix, Table II). Although some correlations were rather high (e.g. .72 between warmth and autonomy support, or −.63 between rejection and autonomy support), multicollinearity was not a problem in our analyses, in which the VIF (variance inflation factor) scores were all well below 10 (the highest was 3.21) and the tolerance statistics all well above 0.2, the commonly accepted rules of thumb.

¹⁰ The distribution into classes for adjudicated and nonadjudicated adolescents (see Appendix, Table IV) indicates a similar percentage of both groups in Class 3 (41.67% and 42.50%, respectively). For Class 1 and Class 2, distributions are different. There are proportionally less adjudicated adolescents belonging to Class 1 (39.17%, vs. 50.30% for nonadjudicated), and more belonging to Class 2 (19.17%, vs. 7.20% for nonadjudicated). Concerning the distribution of nonadjudicated boys and girls: this is almost the same for Class 1, while there are proportionally slightly more girls in Class 2 and more boys in Class 3. Concerning adjudicated adolescents, while the distribution of boys and girls is proportionally nearly the same for Class 3, we notice proportionally twice as many boys (44.44%) as girls (23.33%) in Class 1 and three times more girls (36.67%) than boys (13.33%) in Class 2. Although some differences exist for the repartition, we did not find a specific combination for the subgroup of delinquent.

¹¹ Odds ratio coefficients indicate the probability that membership in one of these three classes will predict the level of morality (low, moderate, or high).

¹² Numbers have been rounded, which explains similar odds ratios for boys and girls, but different effect sizes.

Additional information

Notes on contributors

Julie Grégoire

Julie Grégoire is Doctor of criminology, Scientific Collaborator at Liège University, Faculty of Law, Political Science, and Criminology (Belgium) and member of the Center-Research-Intervention-Youth. She is interested in juvenile delinquency, juveniles in custody, integrated theories, and quantitative and qualitative methods.

Lieven J.R. Pauwels

Lieven J.R. Pauwels is Professor of Criminology at Ghent University, Faculty of Law and Criminology (Belgium). He is Head of the Institute for International Research on Criminal Policy (IRCP). He is interested in theories of crime causation, theoretical integration, the philosophy of science and causation, and quantitative methods.