The age slopes and bootstrap intervals agree in three ds000030 models.

A new analysis of the OpenNeuro ds000030 stop-signal task files reproduced the reported age slopes in three linear models, using the same observations and model rules with new code. The primary control model estimated slopes of 1.167 ms/year (95% participant-bootstrap interval −0.089 to 2.378) for recorded delays, 1.430 (0.043 to 2.816) for scheduled delays, and 1.392 (0.024 to 2.757) for a zero-replaced delay rule. The recorded-minus-scheduled slope difference was −0.263 ms/year (−0.972 to 0.449). Across all three models, 72 estimates and interval endpoints and 120,000 bootstrap values agreed with the target within absolute tolerance 1e−8. The selected sample included 241 participants; the control model included 122. This computational replication uses the same data and known target outputs, and supplies no independent biological confirmation. Each participant had 32 STOP trials. The intervals omit trial uncertainty, tone-timing uncertainty, and model error, and no correction was made for multiple comparisons. The stored fields do not establish delivered tone onset. These cross-sectional associations do not show that age causes score differences or that a difference interval containing zero means equal associations.

stop-signal task · age · computational replication · bootstrap · linear models · OpenNeuro · ds000030 · SSRT