A selected delay coefficient gives lower mean test rank agreement in four splits of 102 ds000030 controls.

Across 500 paired partitions of 102 OpenNeuro ds000030 control participants, selecting a common stop-signal delay coefficient on 51 participants increased mean training rank agreement in all four task splits, but gave a negative mean test change in each split compared with coefficient 1. Mean test changes ranged from -0.001923620 to -0.033355023. Some test directions were positive; these partition quantiles are not confidence intervals and do not support a population confidence claim. Each half had 16 STOP trials, and all partitions came from one cohort. The analysis used a 1,001-value coefficient grid and fixed full-run screens. The result does not establish physical sound timing or explain the lower test agreement. A separate session or cohort with actual sound-time records remains necessary.

stop-signal task · stop-signal reaction time · rank agreement · Spearman correlation · delay coefficient · split-half reliability · participant partition · OpenNeuro ds000030