A new way for handling mobility in longitudinal data

Area of Science:

• Social Sciences
• Educational Research
• Statistical Modeling

Background:

• Applied researchers in social sciences frequently encounter multilevel data with non-exclusive clustering.
• This data structure poses a statistical challenge for traditional analysis methods.
• Accurate modeling is crucial for understanding complex relationships in such data.

Purpose of the Study:

• To introduce and demonstrate the utility of a multiple membership growth curve model (MM-GCM).
• To provide a flexible statistical tool for analyzing longitudinal multilevel data with non-exclusive clustering.
• To address the limitations of existing methods when dealing with mobile units.

Main Methods:

• Development and application of a multiple membership growth curve model (MM-GCM).
• Utilized a real longitudinal educational dataset featuring students who changed schools.
• Compared MM-GCM parameter estimates against 'final school'-GCM and 'delete'-GCM approaches.
• Conducted a simulation study to assess the impact of ignoring student mobility.

Main Results:

• The MM-GCM offers greater flexibility in modeling growth curve intercepts.
• Ignoring student mobility in analyses (e.g., 'final school'-GCM, 'delete'-GCM) leads to significant bias.
• Bias is particularly pronounced in cluster-level coefficients and variance components.

Conclusions:

• The MM-GCM is a robust method for analyzing longitudinal multilevel data with student mobility.
• Ignoring student mobility can severely distort statistical findings.
• Applied researchers should adopt advanced models like MM-GCM for accurate analysis of complex data structures.