Figure 3.

Distribution of reliability estimates on the first test half (initial 24 trials).

Note. n_corr = number of correct sorts; n_cat = number of categories; n_PE = number of perseverative errors; n_SE = number of set-loss errors. Histograms show the frequency of split-half reliability estimates obtained from 100,000 random splits (95% highest density intervals [HDIs] shown as horizontal lines in red color). Colored ticks indicate the Spearman–Brown coefficients that were obtained from systematic splits, with the yellow-to-blue dimension indicating the split’s grain size.