Common average reference reduced the posterior-minus-frontal Alzheimer-control theta-to-alpha contrast in 88 ds004504 v1.0.9 records under a linear age and gender model with source processing fixed

We tested whether EEG reference choice changes a spatial Alzheimer-control contrast. We used all 88 ds004504 v1.0.9 records and 6,377 saved windows. The source processing was fixed. The score was posterior-minus-frontal log10(theta power / alpha power). The model used common linear age and recorded gender effects. Mean subtraction across 19 channels reduced the adjusted contrast from 0.210 to 0.090. The paired change was -0.120 log10 units, with a 95% person-bootstrap interval of [-0.201, -0.039]. Both individual region contrasts increased. The frontal increase was larger. Early-window and parietal-only intervals included zero. The package preserves the local plan, code, matrices, person scores, bootstrap draws, figures, and source hashes. A separate regression and power calculation passed. Three source records reproduced the saved matrices exactly. A clean rerun reproduced all numeric outputs. The other 85 signal extractions are inherited. The result concerns one scalp measurement in one cohort. It does not identify a neural source or estimate clinical performance.

electroencephalography · EEG reference · theta-to-alpha ratio · Alzheimer disease · spatial contrast · bootstrap