Growth Modeling Applications in Two-Method Measurement Planned Missing Designs

• In Event: Methodological Considerations and Guidance for Estimating Longitudinal/Growth Models

Abstract

This paper examines linear growth models in the two-method measurement planned missing design framework. Using simulated data for a three occasion linear growth model, we examined the power, bias, and coverage of key parameter estimates. We manipulated the overall sample size, amount of missing data for gold standard measurements, number/ratio of gold standard to efficient indicators, and the response bias loadings. If researchers are primarily interested in estimating the mean structure of some growth process, they can reliably estimate these parameters with smaller sample sizes and large amounts of missing data. If researchers are also interested in the variance/covariance structure, however; more participants and better indicators are needed with smaller amounts of missing data.

Authors

• Joshua Peri, The Ohio State University - Columbus

• Minjung Kim, The Ohio State University

• Jessica Logan, The Ohio State University - Columbus