The correlation between run halves is higher with hybrid delays in 126 ds000030 controls.

This report tests agreement between stop-signal reaction time (SSRT) calculations from disjoint halves of one stop-signal task run in 126 controls from OpenNeuro ds000030 1.0.0. It compares temporal and interleaved splits and recorded, hybrid, and ladder delay definitions. For the temporal split, Pearson r was 0.513670 with recorded delays and 0.627479 with hybrid delays; for the interleaved split, r was 0.497058 and 0.664091. The bootstrap intervals for hybrid-minus-recorded r were [0.008304, 0.260297] and [0.066135, 0.336903], respectively. Interleaved-minus-temporal intervals contain zero for all three delay definitions. Random splits of the same participants gave median r values of 0.549656, 0.716309, and 0.717522 for recorded, hybrid, and ladder delays. Each fixed half had only 16 STOP trials. The higher correlations with hybrid delays show sensitivity to delay selection in this sample; they do not establish the correct sound delay, race-model validity, or test-retest reliability. The report preserves the calculation, checks, limitations, and source hashes.

stop-signal reaction time · SSRT · split-half agreement · reaction time · bootstrap · OpenNeuro · ds000030