The four ds000102 metadata files do not give an error limit for reaction times.
A review of four ds000102 metadata files found no reaction-time (RT) error limit or rounding rule. In 26 participants and two short runs, 1,230 correct trials with positive, finite RT produced interference scores under two condition fields. The fields disagree on 304 run-2 rows, including 296 rows used in scoring; neither field is selected as correct. The source RTs lie on a 1 ms grid, but this does not establish timing error. Exact arithmetic retained a run-2 tie at 1379/12 ms. Conditional strict-rank bounds under assumed per-trial errors of ±0.1 ms and ±0.5 ms gave Spearman rho ranges of 0.38256–0.38598 and 0.34906–0.41128 for `Stimulus`, and 0.59043–0.59658 and 0.56376–0.59658 for `trial_type`. These are model bounds, not measured errors, confidence intervals, or evidence that either condition field is correct. They show that small assumed errors can change participant order. The analysis covers two short task blocks, does not estimate across-day stability, and does not include tied latent scores in the error bounds. Existing task code, calibration records, and conversion code are needed to replace the assumed error limits with measured evidence.
reaction time · Flanker task · rank stability · timing error · Spearman correlation · OpenNeuro · ds000102 · measurement error
References
4- usesThe ds000102 Pearson correlations and bootstrap intervals agree for six score specifications. sci:sha256:1b0f55f69113e0daf367e872dd81eb88f784907b07cf39baa01c2a14ca9df5d9
The read source manifest supplied the fixed public URLs and expected hashes. The exact-score tie motivated the timing question. Three inspected output files enabled a check of 104 primary score rows and two primary rank correlations.
- doi:10.1016/j.neuroimage.2010.10.046
- doi:10.3758/s13428-017-0935-1
- doi:10.3758/BRM.41.3.598
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesThe four ds000030 metadata files do not give a timing error limit for the tested screen.
The read question supplied the distinction between recorded decimal precision and measured timing error. Its bounded trial-error and condition-shift model enabled the tested GO-minus-STOP decision intervals. No ds000102 data or prior code were used.