Same-day ranges for posterior EEG alpha with closed eyes contained 57% of one-month pair differences in 30 test adults

We tested ranges for posterior EEG alpha differences between people in OpenNeuro ds004148 version 1.0.1. The score used closed eyes, 8 to 13 Hz, 280 planned seconds, and a 59-channel mean reference. A fixed split used 30 calibration people and 30 test people. The empirical 90th percentile of same-day calibration pair errors gave a half width of 2.094 dB. This range contained 329 of 435 same-day test differences (75.6%) and 246 of 435 month differences (56.6%). Pairs shared people. A bootstrap resampled whole people and fitted the width again. The 95% interval for month containment was 33.2% to 83.1%. The month-minus-same-day difference was -19.1 percentage points, with interval -38.7 to +1.7 points. Thus the primary comparison did not establish a population decrease. All 100 planned split checks had lower month containment. They reuse one cohort. The package preserves all 360 spectra, code, source hashes, scores, decisions, draws, and instructions. Six full raw records rebuilt all saved arrays exactly. The range predicts recorded differences and gives no population or individual-pair coverage guarantee.

electroencephalography · EEG · alpha power · test-retest reliability · prediction range · pair differences · bootstrap · measurement error