The paired bootstrap interval for one-month alpha contrast ICC change included zero with a fixed local spectral reference in 60 ds004148 v1.0.1 adults

We tested a fixed local spectral reference for the one-month agreement of posterior alpha eye-state contrast. We used all 60 adults and 360 archived records from OpenNeuro ds004148 version 1.0.1. The reference used the 5 to 7 Hz and 17 to 19 Hz bands to estimate a line in log frequency and log density. The alpha band was 8 to 13 Hz. The raw contrast ICC(A,1) was 0.482. The local contrast ICC(A,1) was 0.501. The paired difference was 0.019, with a 95% participant bootstrap interval of -0.099 to 0.135 from 20,000 samples. The data did not establish higher agreement. The interval does not prove equal agreement. All three planned sensitivity intervals also included zero. Algebra shows that the local measure cancels a positive scale in each state and is zero for an exact power law. These properties do not identify a biological background. Both between-person and residual variance estimates decreased. Independent numerical checks, two complete source record rebuilds, and an offline rerun passed. The package preserves spectra, source hashes, code, all result tables, bootstrap samples, figures, and reproduction steps. This is an analysis of the source cohort. Fixed eye-state order, unequal prior eye-movement processing, and the local fit limit the claim. The plan was saved before the new outcome calculation; it was not externally registered.

EEG · alpha band power · test-retest reliability · intraclass correlation · spectral reference · bootstrap · OpenNeuro