Fig. 2

Diagnosis predicted few EEG variables with both high replication rate and effect size. a. Histograms display 𝞰²partial values associated with prediction of EEG dependent variables by various independent variables (vertical axis of panels) in different age groups (horizontal axis of panels). Colours indicate different categories of EEG variables. All y axes are truncated at 100. Vertical dashed lines indicate the threshold of 𝞰²partial = 0.035. b. Histograms display the replication rate of the ability of various independent variables (vertical axis of panels) to predict EEG dependent variables with 𝞰²partial > 0.035. Replication rate was obtained by bootstrapping random half splits of the data and asking what proportion of splits yielded 𝞰²partial > 0.035 in both halves. Colours indicate different categories of EEG variables. Data are collapsed across age groups for visualisation. Vertical dashed lines indicate the threshold of replication rate = 0.64. c. The scatter plot displays the relationship between 𝞰²partial and p-value for diagnosis coefficients. Lines indicate the threshold of p =.05 and 𝞰²partial = 0.035. Colours indicate age groups. d. The scatter plot displays the relationship between 𝞰²partial and replication rate. Black dots indicate values for age and sex coefficients. Red dots indicate values for diagnosis, diagnosis*age, diagnosis*sex, and diagnosis*age*sex coefficients. The horizontal dashed line indicates the threshold of replication rate = 0.64. Diagnosis was able to predict 13 variables with replication rate > 0.64 and 𝞰²partial > 0.035. The trend exhibited in the plot suggests that lowering the 𝞰²partial threshold would have resulted in very low replication rates