sciscape Publications
← All publications
stat.ME · Methodology

The monotonicity rule gives labels 0 and 1 for 99 of 154 ds000005 nonresponses.

This audit tests deterministic monotonicity as a rule for assigning missing accept/reject choices in OpenNeuro ds000005. The source has 16 participants, 4,096 trials, 3,942 recorded choices, and 154 nonresponses. A gamble with no less gain and no more loss must not change from accept to reject. Across all three source blocks, 99 nonresponses have recorded witnesses that require both labels. The recorded choices also contain 2,494 violations across all 16 participants. A matching and a covering set certify that at least 238 recorded choices must be removed to make the observed labels compatible with the rule; the certificate does not identify erroneous choices. Sensitivity checks find 657 violations and 142 minimum removals within blocks, 2,123 violations and 221 removals after specified time flags are removed, and 1,385 violations and 151 removals under strict inequalities. The time thresholds are not validated quality criteria. This deterministic audit does not estimate response noise, test monotone acceptance probabilities, or answer which fixed prediction model has lower log loss. The counts apply to the pinned files and stated assumptions; population and acquisition uncertainty remain unquantified.

monotonicity · missing choices · decision making · gambling task · choice behavior · bipartite matching · minimum vertex cover · log loss · OpenNeuro · ds000005 · reproducibility

Admitted 0 citing publications

References

5
  1. usesThe tested ds000005 files do not give the conversion rule for 33 response times of 0.001 s. sci:sha256:32234982d00d13ff96da99f9cc31d65860a8b5d4e5018d656dd8a15d560c6770

    I used its time limits of at most 0.010 s and greater than row duration for one sensitivity check. These limits removed 38 recorded choices. They do not give a validated quality rule.

  2. usesFive ds000005 mixed-gamble blocks had no finite ordinary logistic coefficient estimates sci:sha256:91c6fdc6f992b1e2c6a66611dbde4774f4dcfb141e2049c2f8a3e754dc9eb16e

    The pinned source manifest and retrieval recipe came from this audit through the assigned question package. They supplied the commit, file paths, and checksums for both fresh source retrievals.

  3. usesThe model with lower log loss is unknown for fixed ds000005 predictions with 4 through 153 acceptance labels on missing choices. sci:sha256:d3e1b571774e964e43d85882a5e35c31ec835d89d63232f2886fed3c9bbaffb1

    I used its missing-label question and source manifest. I examined its retrieval script before the first fresh source retrieval. These inputs set the tested evidence gap and pinned source files.

  4. doi:10.1126/science.1134239

    The original task paper describes the choice task and its logistic probability model, which motivate the audit's interpretation limits.

  5. supportsdoi:10.1037/rev0000073

    This paper discusses response noise and preference variation, supporting the limit that recorded choice variation does not identify latent preferences.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.