Multiplicative correction of subject effect as preprocessing for analysis of variance

Area of Science:

• Neuroscience
• Biostatistics
• Signal Processing

Background:

• Repeated-measures ANOVA assumes additive subject and condition effects.
• Electroencephalography (EEG) and magnetoencephalography (MEG) data may exhibit multiplicative subject effects.
• Standard ANOVA models may not accurately capture these multiplicative effects.

Purpose of the Study:

• To propose and validate a normalization method for EEG/MEG data with multiplicative subject effects.
• To adapt Analysis of Variance (ANOVA) for normalized data exhibiting multiplicative subject effects.
• To assess the performance of the proposed method in simulations.

Main Methods:

• Data normalization by multiplying each subject's response by a subject-specific constant.
• Derivation of ANOVA tables for the normalized data.
• Simulations to evaluate the method's effectiveness.

Main Results:

• The proposed normalization method effectively handles data with multiplicative subject effects.
• ANOVA and multiple comparisons are performed accurately on the normalized data.
• Simulations confirm the method's performance under the multiplicative model.

Conclusions:

• A simple multiplicative normalization method improves ANOVA for specific EEG/MEG data.
• This approach provides a robust statistical framework for analyzing neurophysiological data with non-additive subject effects.
• The method is effective for both ANOVA and multiple comparisons when data adheres to the multiplicative model.