The Development of Romantic Partner Involvement and Tolerance of Deviance among Juvenile Offenders: Relevance for Offending Behaviors in Early Adulthood

ABSTRACT

This study sought to elucidate heterogeneity in developmental patterns of romantic relationship characteristics and examine the relevance of patterns for predicting offending in adulthood. Group-based trajectory modeling was used to elucidate trajectory groups describing developmental heterogeneity of odds of having a romantic partner and romantic partners’ tolerance of deviance. Poisson regression was used to estimate the effects of group assignments for predicting offending. Three- and four-group models best fit the romantic partner characteristics data. Consistently high odds of having a romantic partner and consistently seeking partners with high tolerance for deviance was associated with greater offending frequency in adulthood.

Notes

¹ These two acts were: how would your partner react if he/she knew you were involved in illegal activity?; how would your partner react if he/she knew you used drugs?

² The ordinal measure of participants’ romantic partners’ tolerance for engagement in deviance was recoded in several ways to best incorporate it in the analysis. The original variable was coded ordinally and ranged from 1–4. Preliminary analyses revealed that the GBTM method was identifying values below the floor category of 1 as possible values, thus, impacting analyses. Because of this, the variable was recoded to set 0 as the floor category. Examination of the distribution of the pooled responses for this variable indicated that it was highly skewed to the left. Because GBTM analyses allow for zero-inflated poisson, logit, and censored normal probability distributions to be used in modeling, the variable was reverse coded so that it resembled a traditional poisson distribution. The final coding of the variable ranged from 0–3 (0 = Would get very upset at me; 1 = Would be bothered and would talk to me about it; 2 = Would be bothered, but would not say anything to me about it; 3 = Would not care at all).

³ Dummy variables corresponding to assignment to the High romantic partner odds trajectory group and Low romantic partners’ tolerance of deviance trajectory group were omitted from analyses as reference groups.

⁴ Sensitivity analyses demonstrated that individuals assigned to the Moderate tolerance group engaged in significantly less frequent offending at age 23 than individuals assigned to the High tolerance group when included in the full model.

⁵ Alternative models examined in sensitivity analyses were composed of four, five, and six group solutions. All models utilized had a solution with a group which had consistently high odds of having a romantic partner throughout the entire study period as the reference group omitted from the poisson regression models. All other trajectory groups were transformed into dummy variables corresponding to group membership vs. all other participants in the sample in a way analogous to the way this was done in the primary analyses. Results remained robust despite these alternative examinations, as having consistently high odds of having a romantic partner predicted heightened offending at age 23 compared to individuals assigned to any other group in the model.

Additional information

Notes on contributors

Thomas W. Wojciechowski

Thomas W. Wojciechowski received his MA sociology in 2016 and terminal doctoral degree in sociology in 2018 from the University of Florida. He earned his BA Sociology with a Criminal Justice Focus from Central Michigan University in 2012. His research takes a life-course approach and focuses mainly on substance abuse, mental health, stress processes, violence, and prevention science.