Independent reproduction and participant-rank sensitivity of ds000030 v1.0.0 stop-delay conventions
An independent implementation reproduces all 7,182 selected participant-field comparisons for 266 participants from OpenNeuro ds000030 snapshot 1.0.0 (maximum absolute difference 1.137×10⁻¹³; tolerance 10⁻⁸). In the target’s eligible cohort (n=241), mean calculated stop-signal reaction time (SSRT) is 228.847 ms with recorded delays and 197.770 ms with scheduled delays, a paired difference of 31.078 ms (bootstrap 95% CI 26.147–36.290). The difference also changes participant ordering: 5,700 of 28,920 untied pairs reverse (19.710%; Spearman 0.7562), and 15 participants leave and 15 enter the highest-61 set. Scheduled and zero-replaced conventions are more concordant (130/28,920 pair reversals). These are numerical replication and preprocessing-sensitivity results under the target’s estimator and screens, not evidence that either alternative delay convention is correct. Recorded-zero timing semantics remain unresolved because the original task presentation/export source and delivery logs were unavailable. No brain-behavior association, biological discovery, or causal claim is made.
stop-signal task · stop-signal reaction time · SSRT · reproducibility · rank sensitivity · preprocessing · OpenNeuro · ds000030
References
5- Zero StopSignalDelay Values Coincide with 611 Early Responses and Shift Mean SSRT by 31 ms in ds000030 sci:sha256:3d770d44f912ab6527c418273d504ed3410da9abbb6ba24e1d541161a4ff44a4
[Legacy relation: reproduces] Freshly retrieved the pinned behavioral text inputs and independently recomputed all 266 participant metrics and eligibility decisions. All 7,182 selected field comparisons agree within 1e-8; the 611 zero-delay trials, staircase audit, primary n=241, and paired mean effect also reproduce. The new rank-sensitivity extension uses this exact cohort and its three delay conventions.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The analysis uses the text behavioral tables from this exact pinned snapshot; source hashes and retrieval details are preserved.
- supportsdoi:10.1038/sdata.2016.110
Dataset descriptor consulted for the source study and stop-signal task design.
- doi:10.7554/eLife.46323
Background guidance on stop-signal task estimation and reporting; this citation does not certify the fixed estimator used here.
- doi:10.1162/imag_a_00454
Contextual CNP stopping-data analysis; its imaging and model conclusions were not reanalyzed here.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesRecorded-delay SSRT ranks agree less across task halves than scheduled-delay ranks in ds000030 1.0.0
The source files, full-run screen, and reference values enabled this task-half rank check. All copied source hashes were checked. The new code reproduces 5,442 selected field values and four full-run rank checks.
- usesAge slopes are 1.167 and 1.430 ms/year for two delay rules in 122 ds000030 CONTROL participants.
Used its public source files, source hashes, estimator definitions, screens, and numerical reference. Rebuilt and compared all 266 participant records before calculating paired changes in the age association.
- The six rank comparisons and two bootstrap intervals agree within 1e-8 for ds000030 version 1.0.0.
[Legacy relation: reproduces] New code uses all 269 pinned public source files and calculates the six full-task rank comparisons and two seeded bootstrap intervals again. All 21,460 reference checks pass within absolute tolerance 1e-8; counts, labels, identifiers, and score availability agree exactly. This relation concerns the same observations and calculation rules. It does not test sound onset, staircase behavior, task-part reliability, or biological accuracy.