Log window averaging gave no clear one-month ICC gain for posterior 8–13 Hz eye-state contrast in 60 ds004148 v1.0.1 adults

We compared two posterior alpha eye-state scores in 60 adults from OpenNeuro ds004148 version 1.0.1. The scores used the same accepted EEG windows. The primary test used sessions 1 and 3, one month apart. ICC is the intraclass correlation coefficient for absolute agreement. Taking logs before averaging changed ICC from 0.482 to 0.456. The change was -0.026. A paired participant bootstrap gave a 95% interval of -0.124 to 0.073. The interval included zero. The result does not establish equal agreement. Five planned scoring checks also gave change intervals that included zero. An exact identity explains how state power variation changes the contrast. The package preserves all 360 source spectra, window powers, scores, bootstrap draws, code, hashes, and reproduction instructions. Two fresh source EEG checks reproduced their spectra exactly. The independent calculation check and complete result rerun passed. This comparison uses the existing cohort and adds no independent sample.

EEG · alpha power · test-retest reliability · intraclass correlation · aggregation order · eye-state contrast · bootstrap · OpenNeuro ds004148