New concepts of multiple tests and their use for evaluating high-dimensional EEG data

Area of Science:

• Neuroscience
• Statistics

Background:

• Traditional methods for multiple comparisons include Family-Wise Error Rate (FWER) and False Discovery Rate (FDR) control.
• Novel type I error control criteria have recently been proposed.

Purpose of the Study:

• To introduce and evaluate new type I error control criteria for multiple comparisons.
• To investigate the power of statistical tests under these new criteria using simulations.
• To apply these methods to real-world data, specifically EEG coherence during memory encoding.

Main Methods:

• Simulations were conducted to assess the power of test procedures across various factors.
• Factors examined included the number and correlation of endpoints and the proportion of false hypotheses.
• Electroencephalography (EEG) coherence data was analyzed using different multiple testing procedures.

Main Results:

• The study identified how statistical power is influenced by endpoint characteristics and hypothesis proportions.
• Significantly higher EEG coherence was observed for subsequently recalled nouns compared to those not recalled.
• This suggests a neural correlate of successful memory encoding detectable with EEG.

Conclusions:

• The newly proposed type I error control methods offer alternative approaches to multiple comparisons.
• EEG coherence analysis can differentiate between subsequently recalled and not recalled items.
• These findings contribute to understanding the neural basis of memory encoding and highlight the utility of advanced statistical methods.