Youth Runaway Versus Delinquent Peers: Estimating Changes and Testing the Systematic Relationship Between Two Factors using HGLM(Hierarchical Generalized Linear Models)

Purpose:

Youth runaway has become a serious social problem due to the fact that youth runaway itself is not only a representative type of delinquency, but also plays a significant role as a gate to other delinquent behaviors. To reduce youth runaway and to provide proper interventions, factors of youth runaway should be precisely examined. Many studies in this field have found the strong relationship between youths' delinquent peers and youth runaway. However, most studies were performed with two conflicting 'untested' and 'intuitive' assumptions: one insists increasing delinquent peers cause youth runaway while the other suggests youth runaway increases youths' delinquent peers. Furthermore, only a few longitudinal studies were conducted on this topic. Most research on causal ordering of youth runaway and delinquent peers are drawn from cross-sectional analysis with which not enough to say the accurate relationship between two factors. Therefore, using a 4-wave longitudinal dataset, this study examined the longitudinal causal relationship between the two constructs to determine whether youth runaway influenced youths' delinquent peers significantly, or delinquent peers influenced youth runaway, or both. Then this study suggested needed intervention to prevent youth runaway:if youth runaway increases delinquent peers, other factors predicting youth runaway should be examined and if vice versa, interventions should be focused on reducing deliquent peers or helping youth develope positive relationships.

Method:

Based on the data from Korean Youth Panel, 274 youths who have experienced running-away during 4 years from middle school to high school were analysed. To measure runaway and delinquent peers, the number of runaway during last 1year and the number of friends who had been arrested or punished for delinquency during last 1year were used respectively. HGLM(Hierarchical Generalized Linear Models) were performed to test longitudinal causal relationship and changes. This method is appropriate for examining longitudinal changes in nonlinear models and testing the causal direction by checking which direction is statistically significant. For analysing HGLM, a computer program HLM was used.

Results :

Our research questions are as follows: First, does the number of youth runaway change over time? Is there any difference in individual changes? (How about the number of delinquent peers?). Second, what is the causal directional relationship between youth runaway and delinquent peers?

Results showed that the number of youths' delinquent peer increased significantly over time(with significant individual differences in chage patterns), although the number of youth runaway did not increase(with significant individual differences in chage patterns). Also it showed that there was a significant influence of delinquent peers on youth runaway, rather than vice versa.

Implication :

This findings suggest that we should focus on youths' delinquent peers to prevent youth runaway. The intervention programs for youth runaway problem should focus on helping youth improve positive relationships with peers, or on reducing factors which produce negative relationship with delinquent peers.

Also, applying this same methodology to other reciprocal factors(such as school achievement and self-esteem) is suggested to clarify the exact causal relationship between two factors and to find appropriate interventions.