A comparison of analytical methods for the two group pre- post design

Allis, Vanessa; White, Paul

Abstract

Simulation demonstrates that for a two group randomized design ANCOVA is more powerful than the gain score approach

For ANCOVA with an interaction term the model with pre scores X as a covariate is more powerful than the model with X + Y as the covariate

In ANCOVA (with or without an interaction term) there is no power advantage to be had by reversing the roles of X and Y in the model

In ANCOVA reversing the roles of X+Y and X-Y might have a power advantage but the advantage is less than that observed in the models not using X+Y

The results above are for when the correlation coefficient (

• Presentation Conference Type: Poster
• Conference Name: UWE Student Conference
• Start Date: Apr 10, 2017
• End Date: Apr 10, 2017
• Acceptance Date: Apr 10, 2017
• Publicly Available Date: Jun 7, 2019
• Peer Reviewed: Not Peer Reviewed
• Keywords: comparison, analytical methods, two group pre-post design
• Public URL: https://uwe-repository.worktribe.com/output/889343
• Additional Information: Title of Conference or Conference Proceedings : UWE Student Conference, 2017