Multilevel Models for Studying Organizational Transition Effects.

Over the past 15 years a body of research has investigated organizational transitions in adolescent development. Researchers studying school transitions exclusively ask whether significant transition affects students' social and psychological development. To test the presence of transition effects, researchers have typically observed mean differences on certain observed outcomes between the students who experienced school transition and those who did not. This paper shows that much more could be learned about organizational transitions if investigators change their operational definition of transition effects from "mean difference" to "an effect being in a transition group." The new definition changes the research question from "whether" to "how much," and accounts for the uncertainties of transition effects as well as the outcome under investigation by multilevel modeling. The paper demonstrates multilevel modeling by analyzing longitudinal data from NELS:88 to identify the effects of between-school transitions on student self-esteem. The final sample for this paper was composed of 483 students in 100 middle schools and 113 high schools, who were observed at the 8th and the 10th grades. The analyses show the estimates of transition effects and identify the components of transition effects, the effects of individual background variables, past and present organizational characteristics, and the interaction effects between past and present organizations. Multilevel modeling also allows investigation of the interaction effects between individual backgrounds and school characteristics. Twelve tables and 3 figures are included. (Contains 19 references). (LMI)