Regression analyses of repeated measures data in cognitive research

Area of Science:

• Cognitive Psychology
• Statistics
• Research Methodology

Background:

• Repeated measures designs with nonorthogonal variables are increasingly common in cognitive psychology research.
• Standard regression textbooks often do not cover these designs, leading to potential analytical errors.
• Current analytical approaches may use inappropriate error terms for hypothesis testing.

Purpose of the Study:

• To identify and critique common inappropriate statistical analyses of repeated measures designs with nonorthogonal variables.
• To present a more statistically appropriate method for analyzing such data.
• To illustrate alternative computational procedures for the recommended analysis.

Main Methods:

• Review and critique of two commonly used analytical approaches for repeated measures designs.
• Development and presentation of a statistically sound alternative analysis.
• Illustration of computational procedures for the proposed analysis.

Main Results:

• The study argues that existing common methods incorrectly use error terms for testing independent variable effects.
• A more appropriate analytical framework is proposed.
• Two computational methods for implementing the appropriate analysis are demonstrated.

Conclusions:

• Common analytical practices for nonorthogonal repeated measures designs in cognitive psychology are often inappropriate.
• Correct statistical analysis is crucial for valid hypothesis testing in these designs.
• The presented method offers a statistically sound alternative for researchers.