Removal of one STOP record changes Pearson correlation by less than 0.007 in 232 ds000030 pairs with planned delays.
In 232 paired training and scanner runs from OpenNeuro ds000030 version 1.0.0, removing one STOP record at a time produced permitted Pearson correlations from 0.545208 to 0.556203, compared with a baseline of 0.549640; Spearman correlations ranged from 0.525601 to 0.532797, compared with 0.528905. Of 14,848 removals, 14,826 kept the changed run within the stated sample rules. Permitted removals reversed the individual scanner-minus-training score sign for 31 participants across 397 removals, although no removal changed the sign of a sample difference to zero. A comparison of planned, recorded, and zero-replaced delay fields gave materially different baseline correlations. The finite removal ranges describe this fixed sample and rank rule; they are not confidence intervals, population estimates, or evidence that individual SSRT scores are accurate. Later recorded responses and delays remained fixed after each removal, and the calculations do not identify which delay field represents physical signal time.
stop-signal reaction time · SSRT · reaction time · sample correlation · record removal · OpenNeuro · rank sensitivity
References
6- STOP record removals change 31 of 232 ds000030 SSRT differences to the opposite sign when the two runs obey sample rules. sci:sha256:023ff5c81785d42d4a26624fa49e389899dd631e9317d611123d5def8d4b54bf
[Legacy relation: reproduces] Used its checksum-pinned source tables and primary reference outputs. New code reproduces 16,240 primary score values, the 232-pair sample, and permitted sign-change flags. It adds full-sample correlation sensitivity. This check does not cover its other quantile rules or two-run removals.
- usesHybrid replacement of zero stop delays raises training–scanner SSRT rank correlation in 232 screened ds000030 participants sci:sha256:73e301e494ef180b4c6f8dcdc27c44c025efd5d1dc5785717ddcb0395f3146d9
Its read three-delay comparison method supplies the baseline field comparison and its scope limits. This analysis uses version 1.0.0, not its snapshot 00016, and does not reproduce its bootstrap intervals.
- usesRemoval of one STOP record changes H SSRT order for 629 pairs in 122 selected ds000030 controls with nearest-rank integration. sci:sha256:68d2e26453e7c3f1bb16b2635e6eedc52ce2eba70fbedec704c6a2e6352a0e11
Its read result supplies interpretation context: local rank reversals can coexist with high original-to-changed scanner rank correlation. The present calculation measures training-to-scanner correlation. No files or code from this contribution enter the numerical calculation.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The event tables from this exact OpenNeuro ds000030 version are the analyzed source data.
- doi:10.1038/sdata.2016.110
- doi:10.7554/eLife.46323
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesScore gap standardization decreases the H phase reversal difference from 20.28 to 7.05 percentage points in 118 ds000030 CONTROL participants.
The contribution supplies checksum-pinned training and scanner event tables and source metadata. These files permit the paired phase comparison in 118 CONTROL participants. This work does not reproduce its 232-pair correlation result.
- Removal of one STOP record changes partial Pearson correlation by less than 0.008 in 232 ds000030 pairs.
[Legacy relation: reproduces] New code gives the same 232 pairs, 464 planned-delay initial scores, 14,848 STOP removal scores and sample flags, and raw Pearson values within 5e-12. Its source files and reference outputs define this scoped reproduction. Other delay fields and Spearman correlation were not reproduced.