Research Design Considerations for Estimating Latent Time-to-Criterion Growth Models

• In Event: Longitudinal and Structural Equation Modeling

Abstract

Time-to-criterion (T2C) models offer a promising alternative to traditional growth models for policymaking contexts, such as decisions around English Learner education. Instead of estimating an intercept, T2C models estimate the length of time (tau) required to reach a threshold. The latent T2C model was recently shown to be mathematically equivalent to traditional latent growth models, but missingness may challenge the usefulness of T2C in certain circumstances. The present study examined this issue using Monte Carlo simulation of 5-time-point data. Results showed that, in sample sizes less than N = 500, estimates of tau and its predictor exhibited positive bias and lower power when attrition and growth variability were both relatively high, and when the criterion was farther out in time.

Authors

• LizAnne M. Ngo, University of Washington
  Presenting Author

• Micaela Moricet, University of Washington
  Non-Presenting Author

• Elizabeth A. Sanders, University of Washington
  Non-Presenting Author