Longitudinal measurement models in evaluation research: Examining stability and change

This article provides an introduction to the use of Confirmatory Factor Analysis to test measurement invariance and stability in longitudinal research. The approach is illustrated through examples representing: (a) one construct, two measurement waves; (b) one construct, three waves; (c) two constructs, two waves; and (d) comparison of treatment and control groups in pre-post designs. Basic issues in establishing measurement invariance over time, across treatment groups, and within measurement waves are discussed. Estimates of the stability coefficients that are corrected for measurement error and method variance associated with each specific measured variable are provided. Establishing measurement invariance is a critical requirement for making inferences about treatment effects and changes in constructs overtime. Copyright (C) 1996 Elsevier Science Ltd