PC program implementing an alternative to the paired t-test which adjusts for regression to the mean

Area of Science:

• Biostatistics
• Medical Research Methodology

Background:

• Biomedical studies often recruit participants based on extreme measurements (e.g., high blood pressure).
• Simple change scores in such studies can overestimate treatment effects due to regression towards the mean.
• The paired t-test may incorrectly attribute changes to treatment rather than this statistical phenomenon.

Purpose of the Study:

• To present a method for adjusting observed changes for regression towards the mean in pre-test-post-test experiments.
• To provide a practical tool (PC program) for researchers to implement this adjustment.
• To offer an alternative to the paired t-test that distinguishes treatment effects from regression effects.

Main Methods:

• Utilizes the method proposed by Mee and Chua (1991).
• Implements a procedure to correct observed change scores for the regression effect.
• Applies to simple pre-test-post-test experimental designs.

Main Results:

• Demonstrates how to adjust for regression towards the mean in pre-test-post-test studies.
• Provides a computational tool for researchers.
• The method separates true treatment effects from regression artifacts.

Conclusions:

• The proposed adjustment method accurately accounts for regression towards the mean.
• The PC program facilitates the application of this statistical correction.
• This approach offers a more precise estimation of treatment effects in specific research contexts compared to the standard paired t-test.