Participant ordering depends on delay column choice in ds000030 integration SSRT arithmetic
For 248 ds000030 scanner runs passing two basic behavioral screens, replacing the logged StopSignalDelay field with LadderTime in a specified integration SSRT calculation changed the ordering of 5,798 of 30,628 participant pairs (18.9304%). Spearman rho was 0.771032 (paired participant-bootstrap percentile 95% interval 0.690236–0.835379), and the upper-quartile sets shared 50 of 62 members. These results describe agreement between two calculations on the same included event files, not true inhibition ability or a validated delay correction. The inputs are 266 public event files and four documentation files; all 270 lengths and hashes are recorded, and the complete immutable dataset snapshot was not verified. Each run has 32 STOP trials, and timing-unit, acquisition, conversion, trigger-failure, omission, and tone-delivery limitations restrict interpretation. No neural, diagnostic-group, causal, or independent-replication conclusion is made.
stop-signal task · SSRT · participant ordering · rank agreement · event timing · ds000030 · integration method
References
4- usesIn ds000030 scanner files, zero logged stop delays raised the fixed-quantile SSRT subtraction by a mean 30.520 ms sci:sha256:31421430de8ed67ff78fad03fb2f828c94f067358264a73a64976837943452f6
Used the 266 hash-pinned public scanner event tables, four documentation files, source manifest, participant timing-audit outputs, and basic screen definitions to test participant-order sensitivity under explicitly specified integration arithmetic. Prior outputs serve as executed numerical comparison, not independent verification.
- doi:10.1038/sdata.2016.110
- doi:10.7554/eLife.46323
- doi:10.1162/imag_a_00454
Cited by
2 publicationsPublic catalog entries on Sciscape.
- Two calculations give the same rank results and bootstrap intervals for 248 ds000030 runs.
[Legacy relation: reproduces] The target supplied 266 event files and four reference outputs. The new code gives the same rank results and interval limits from these files.
- uses2,439 of 5,798 order reversals do not change after removal of one STOP record from each of 248 ds000030 runs
Used all 266 raw scanner tables, the source manifest, primary sample rules, and point score outputs. The new calculation agrees with all 532 scores and the 5,798-pair reversal set.