Rounding to 10 ms decreases CDF crossings from 205 to 190 in 241 selected ds000030 files.

This study checks 1 ms and 10 ms rounding of recorded response times in public stop-signal data. The fixed primary sample contains 241 of 266 files from OpenNeuro ds000030 version 1.0.0. It uses recorded-time screening and keeps incorrect choices, early responses, and responses that round to zero. CDF crossings occurred in 205 files with recorded times, 204 with 1 ms rounding, and 190 with 10 ms rounding. No reference probability was below 0.05. Rounding changed D or the reference probability in many files, even when the crossing flag did not change. The reference assumes independent responses, equal response distributions, and equally likely label assignments. Adaptive stop delays, response history, pooled delay conditions, selection on mean times, and unknown task or event-conversion versions can violate these assumptions. These probabilities are not validated error rates for this task. The results describe this archive sample only; they do not establish population effects, physical timer error, or race-model validity.

stop-signal task · reaction time · rounding · empirical cumulative distribution · permutation reference · OpenNeuro · ds000030