| Hyewon Chung
John Jay College, CUNY
Incorporating Student Mobility in Longitudinal Datasets: A Multiple-Membership Random Effects Modeling Approach
Student mobility is widespread in the United States; student mobility was even more widespread among poor and minority students. In scenarios in which there is student mobility, researcher should focus on not only the effects of the students' current school but include the effects of any prior school(s) attended. Conventional hierarchical linear modeling (HLM) can handle pure hierarchical data structures, for example, in which a student is nested within a single school. However, the effects of prior school(s) that mobile students attended on academic achievement cannot be fully understood when using conventional HLM. In the presence of student mobility, multiple membership random effects modeling (MMREM) technique is required to handle multiple-membership in a longitudinal dataset setting. Schools are being held accountable for students' performance and thus an accurate estimation of schools' effects is crucial. In school accountability and school choice where value-added models of school effectiveness are frequently used in the publication of school level performance indicators, the contribution of prior school(s) that mobile students have attended to their academic achievement scores should be modeled. The goal of the current study is to investigate various factors associated with academic achievement using the MMREM approach when modeling student performance across time in a data set with mobile students. The current study will employ Early Childhood Longitudinal Study Kindergarten (ECLS-K) data set. This study will also use simulated data with known variance quantities to investigate whether the MMREM approach produces more accurate variance estimates than conventional HLMs.
Back to Funded Research Grants Page |