Fig. 3

Impact of hearing‑threshold covariate control on feature significance. (A) Scatterplot of –log₁₀(p) values before (x-axis) and after (y-axis) controlling for left-ear pure-tone average (PTA_L) using permutation ANCOVA (n = 110, 210 features). Each point represents a qEEG feature, colored by significance status: green = significant in both models, blue = gained significance, red = lost significance, gray = not significant. Dotted lines indicate the uncorrected p = 0.05 threshold (–log₁₀ ≈ 1.30), and the dashed diagonal represents the identity line. (B) Features that gained (top) or lost (bottom) significance after PTA_L adjustment, ranked by –log₁₀(p). Feature types are color-coded: red = AI, green = coherence, orange = PLV, blue = spectral power. The vertical dashed line marks the p = 0.05 threshold.