Detailed Outline: Mixed Models for Longitudinal Data

Day 1 (June 12): Introduction To Growth Modeling With Mixed Models.
Daniel Bauer

1. Classical methods for repeated measures
   • Within-subjects ANOVA
   • MANOVA
   • Residualized change analysis
   • AR and cross-lag models None capture individual trajectories of change in a theoretically optimal way.
2. Conceptual Orientation to Growth Modeling
   • Trajectories of change
   • Desire to identify functional form of trajectories
   • Desire to summarize individual differences in change
   • Desire to predict individual differences in change
3. Case-wise OLS estimation of individual trajectories
   • Give empirical example
   • Advantages
     • Direct estimation of parameters for individual curves
   • Disadvantage
     • Variance inflation, etc.
4. Growth models as hierarchical linear models
   • Repeated measures nested within individuals
   • Within-person model
   • Between-person model
   • Reduced-form model
   • Highlight assumptions
5. The unconditional linear growth model in PROC MIXED
6. Afternoon Session: Hands On Data Analysis
   • Analysis of an empirical example
     • Repeated measures ANOVA (time?)
     • Residualized change analysis (time?)
     • Case-wise OLS estimation of individual trajectories
     • Fitting unconditional model within PROC MIXED

Day 2 (June 13). Advanced Topics In Growth Modeling With Mixed Models

1. Modeling nonlinear growth (applied to empirical example)
   • Higher-order polynomials (quadratic, cubic, etc)
   • Piecewise linear models
   • Nonlinear models
2. One function fits all? A short conceptual discusssion on growth mixture models.
3. Conditional growth models (w / time invariant predictors)
   • "Slopes as outcomes"
   • Note implication of cross-level interaction with time
     • Can be explored/probed like any other interaction
   • Modeling repeated measures for predictors as well as outcomes
     • Models with time-varying covariates
     • Multivariate Growth Models
       • "Parallel Process" models
     • Three-level models, higher levels of nesting
       • Repeated measures within person within cluster
     • Afternoon Session: Hands On Data Analysis
       • Analysis of an empirical example
         • Evaluation of functional form (linear, nonlinear, how to model nonlinear?)
         • Inclusion of time-invariant predictors
         • Inclusion of time-varying covariate
         • Estimation of multivariate growth model?