Cz gives a 0.332 dB lower mean log alpha power ratio than the average reference in 24 participants.

In 24 participants from the EEG Motor Movement/Imagery Dataset, version 1.0.0, the mean eyes-closed to eyes-open alpha power ratio at O1, Oz, and O2 was positive for every participant under the supplied, 64-channel average, and Cz references. For [5,55) seconds and [8,13) Hz, mean log ratios were 6.966 dB (95% participant-bootstrap interval 5.383 to 8.621) for the supplied reference, 8.096 dB (6.481 to 9.804) for the average reference, and 7.764 dB (6.129 to 9.467) for Cz. The paired mean difference, Cz minus average, was -0.332 dB (95% interval -0.507 to -0.123); participant ranks were highly correlated (Spearman rho 0.989), and no participant changed direction. The difference remained negative in the planned time and frequency checks, and was -1.118 dB when alpha power was divided by [2,30) Hz power. The records were not filtered and artifacts or participants were not removed. R01 preceded R02, so condition is confounded with recording order. This result does not establish a causal eye-closure effect, a correct reference, or an effect in all 109 participants. Bootstrap intervals describe sampling from a larger similar population and are not measurement-error bounds for this fixed set.

electroencephalography · EEG · alpha power · reference · Cz · average reference · Welch spectrum · bootstrap · eyes closed · PhysioNet