With valid times fixed, 150–1,000 ms invalid trial scores allowed both signs of the adjusted visual minus auditory group contrast in 144 ds004574 v1.0.0 records

We tested a completed cue score in 144 selected people from OpenNeuro ds004574 v1.0.0: 96 with Parkinson's disease and 48 controls. Valid recorded times stayed fixed. Each invalid novel trial could receive a score from 150 to 1,000 ms. The score was the mean of all 24 visual scores minus the mean of all 24 auditory scores. Ordinary least squares adjusted the PD minus Control contrast for age and gender. The exact coefficient range was -98.84 to +51.46 ms. One score shared by all invalid trials gave -29.56 to -17.82 ms. Two scores shared by cue across both groups gave -76.39 to +29.02 ms. Thus one common penalty preserved a sign that two bounded cue scores did not preserve. Zero first became possible at a maximum change of 162.57 ms from personal valid cue means. An independent parser, QR weights, direct regression fits, and linear programs checked the bounds and distance. All 36 analysis files matched in offline and fresh-source repeats. The package preserves 152 source files, hashes, code, a local exploratory plan, proof, trial assignments, and instructions. These are bounds for fixed records and an assumed numerical score. They are not confidence limits or actual response times after errors.

Parkinson disease · response time · missing data · sensitivity analysis · linear regression · OpenNeuro