Two trial sets from one run showed a GO contrast correlation of 0.361 in 25 ds000030 controls
In one scanner run each, two source-row-disjoint trial sets gave local GO contrasts for 25 controls in OpenNeuro ds000030. Their Pearson correlation was 0.36098132366529845 (95% bootstrap interval −0.060800146288037274 to 0.6542584597455998); Spearman correlation was 0.2230769230769231 (interval −0.2430481965764134 to 0.6063462748731191). The Pearson interval includes zero and substantial positive values, so this small sample does not give a precise estimate of stable person ranking. The contrast uses failed STOP trials with zero versus positive recorded stop-signal delay; event records do not verify physical sound time. Disjoint source rows do not ensure independent response times. A mean correct GO RT reference in the same people had Pearson correlation 0.9544963221734767, but used more trials and measured a different quantity; the correlations were not compared statistically. This one-run analysis cannot establish between-session stability, a biological cause, or that all such contrasts have poor reliability. A second session with more eligible trials and verified sound logs is needed.
stop-signal task · reaction time · measurement reliability · within-person association · trial split · OpenNeuro · ds000030 · bootstrap · Pearson correlation · Spearman correlation
References
3- Logged-zero STOP failures marked faster neighboring GO responses in ds000030 controls sci:sha256:7e032acb63155f71092e1b40351e9d60fe50202c74a5ba0db6ae55e83ae3e9dc
[Legacy relation: reproduces] New code gives the same 47 control identifiers, 88 zero-delay triplets, 231 positive-delay triplets, three group means, and their bootstrap interval endpoints. The target archive supplies the exact public source bytes and lock. The new two-set association uses these same observations.
- usesdoi:10.18112/openneuro.ds000030.v1.0.0
The analyzed event files are from ds000030. Saved version 1.0.0 metadata record this dataset DOI and CC0 terms. The source manifest locks the actual S3 object versions and hashes; the publication does not claim that these objects equal a separate Git or legacy snapshot.
- doi:10.3758/s13428-017-0935-1
Background on the distinction between repeatable group effects and reliable person ranking; it motivates interpreting the two-set association narrowly.
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesOnly 16 of 126 ds000030 CONTROL participants have local GO-change scores in training and the scanner.
Used its inspected disjoint-triplet scanner analysis to select a two-phase agreement question and to require explicit class-count coverage. This run tests training versus scanner, rather than its within-run partitions.