Correcting Measurement Error in a Conditional Latent Growth Model Using the Single Indicator Factor Approach

• In Event: Marriott Roundtable Session 52

Abstract

Since composite scores (e.g., sum of psychological scale items) are commonly used as time-varying predictors in a conditional latent growth model (LGM), the occasion-specific effects of a time-varying predictor on its corresponding outcome measure could be biased due to the unreliability in its measurement. The measurement error in the composite time-varying predictor may be corrected by single indicator (SI) factor approach. As such, this study compared the parameter and standard error recovery rates of the LGM with a composite time-varying predictor and those of the LGM using the SI factor approach. The results indicated that the parameter estimates from the LGM with the SI factor approach were less biased than those of the LGM with a composite time-varying predictor.

Authors

• Kejin Lee, University of Nebraska - Lincoln

• Daniel Peche Gonzalez, The University of Texas - Austin

• Susan Natasha Beretvas, The University of Texas - Austin