With source processing fixed, linear age and gender adjustment gave a larger Alzheimer-control theta-to-alpha contrast for the posterior mean than for deviations in ds004504 v1.0.9.

We split posterior EEG band power into the power of a mean signal and the mean power of deviations. The mean signal is the posterior mean minus the 19-channel mean. The split is exact under the same linear spectral rule. Both parts cancel a shared reference signal. We used all 88 processed records and 6,377 windows in OpenNeuro ds004504 version 1.0.9. With linear age and gender adjustment, the Alzheimer-control log10 theta-to-alpha contrast was 0.824 for the mean signal and 0.733 for deviations. The paired difference was 0.091, with a 95% person bootstrap interval of 0.011 to 0.169 from 20,000 draws. The first 20-window check gave 0.036, with an interval of -0.058 to 0.129. The parietal interval also included zero. This is an exploratory study with a local fixed plan and one processed cohort. It does not identify a neural source or test diagnosis in another cohort. We checked all saved matrices, independently repeated all primary bootstrap estimates, and repeated 12 result files with identical hashes. We rebuilt 15 hash-checked source records across all three groups. The other 73 original source records were not rebuilt in this run. The package preserves code, matrices, proof, plan, source hashes, results, figures, and repeat instructions.

electroencephalography · EEG · spectral power · theta-to-alpha ratio · Alzheimer disease · spatial signal · bootstrap