An autoregressive latent change score model for randomized pretest, posttest, follow-up designs

Area of Science:

• Psychometrics
• Developmental Psychology
• Biostatistics

Background:

• Randomized pretest, posttest, follow-up designs are crucial for evaluating treatment efficacy.
• Existing statistical methods like ANOVA and ANCOVA have limitations in analyzing such designs, including reduced statistical power and the use of residualized change scores.
• Latent change score models (LCSMs) offer an alternative but do not simultaneously address all limitations of previous methods.

Purpose of the Study:

• To develop a novel autoregressive latent change score model (AR-LCSM) that overcomes the limitations of existing analytical approaches for randomized pretest, posttest, follow-up designs.
• To ensure the proposed model preserves the interpretability of change scores without requiring residualization.
• To compare the performance of the new AR-LCSM against established statistical alternatives.

Main Methods:

• Development of an autoregressive latent change score model (AR-LCSM).
• A simulation study was conducted to compare the performance of the AR-LCSM with competing statistical models.
• Application of the AR-LCSM to real-world data from children at high risk for autism spectrum disorders.

Main Results:

• The proposed autoregressive latent change score model (AR-LCSM) demonstrated performance comparable to or better than existing methods in a simulation study.
• The AR-LCSM successfully preserves the change score interpretation without the need for residualization.
• The model's utility was further validated through its application to a dataset of children at high risk for autism spectrum disorders.

Conclusions:

• The autoregressive latent change score model (AR-LCSM) provides a powerful and flexible statistical tool for analyzing longitudinal data from randomized pretest, posttest, follow-up designs.
• This novel approach addresses key limitations of traditional methods, offering improved statistical power and clearer interpretation of treatment effects.
• The AR-LCSM is a valuable advancement for researchers in psychology, developmental science, and other fields employing longitudinal study designs.