It is essential for educational research which focuses on individual growth or development to conduct longitudinal analysis. This study aims to indicate the necessity and the application possibility of longitudinal studies by investigating longitudina...
It is essential for educational research which focuses on individual growth or development to conduct longitudinal analysis. This study aims to indicate the necessity and the application possibility of longitudinal studies by investigating longitudinal mediation effect.
To testing career development theory, it is appropriate to conduct longitudinal analysis because this theory investigates hypotheses that relate to individual career development over lifetime. Therefore, this study conceptualizes adolescents' career development level as career maturity, and examines whether career maturity was changed during 3 years. Moreover, this study investigates cause of interindividual difference in development of career maturity. From cross-sectional research, gender, parents' education, and parents' income were found to affect career maturity, but their longitudinal effects are unknown and these variables have low possibility of change, it is useful to discover mediator which delivers effect of these variable to outcome variable. Hence, parent-adolescent relationship is adopted as a mediator in this study. For modeling change for not only career maturity but also parent-adolescent relationship, this study utilizes multivariate latent growth modeling.
As a research data, Korea Youth Panel Survey(KYPS) that collects responses of 3,449 adolescents from 2003 to 2005 is used. The followings are corresponding longitudinal nalysis result.
First, career maturity changed positively between 8th grade and 10th grade. Also positive change was found for parent-adolescent relationship. Because the result is based on longitudinal analysis, it enables to make an true inference about change in adolescents' career maturity.
Second, parent-adolescent relationship as a mediator of the gender on career maturity had a significant effect. Because gender is dichotomous variable, multi-group latent growth modeling analysis was conducted first. As a result of multi-group analysis, girls were found to have higher initial status scores than boys on both parent-adolescent relationship and career maturity. Also, girls were found to have higher growth rate scores than boys on both parent-adolescent relationship and career maturity. From these result, gender was introduced as a subject variable in mediational process. For initial status, gender had a significant direct effect on career maturity, also had a significant indirect effect on career maturity through the mediator - parent-adolescent relationship. But for growth rate, gender had a only significant indirect effect on career maturity through mediator. Therefore, in the case of growth rate, causal relationship can be established by introducing parent-adolescent relationship as mediator.
Third, parent-adolescent relationship as a mediator of the parents' education on career maturity had a significant effect. In initial status perspective, parents' education affected initial status of career maturity, consequently direct effect was found to be significant. Also indirect effect through parent-adolescent relationship was found to be significant. But in growth rate, mediation tests could not be conducted because parents' education didn't affect growth rates of parent-adolescent relationship and career maturity.
Fourth, parent-adolescent relationship as a mediator of the parents' income on career maturity had a significant effect. In initial status perspective, although parents' income didn't have significant direct effect on career maturity, had significant indirect effect on career maturity through parent-adolescent relationship. But in growth rate, mediation tests could not be conducted because parents' income didn't affect growth rates of parent-adolescent relationship and career maturity.
From these result, the following issues are discussed.
Cross-sectional effect and longitudinal effect should be treated separately. And, although direct effect is not significant, causal relationship can be established by introducing a mediator in the model. Also, when predictor variable is also changed, utilizing multivariate latent growth modeling is appropriate for testing dynamic relationship between variables.