Target P3 scores at Pz had higher correlation between block halves than difference scores in 63 ds006018 adults

We compared two P3 EEG scores in public visual oddball data from OpenNeuro ds006018 version 1.2.2. We processed 96 complete records. 63 people met the fixed trial-count rules. Each comparison half had trials from all five target-letter blocks. The target score had an early-to-late correlation of 0.866. The target-minus-nontarget score had a correlation of 0.651. The paired correlation difference was 0.215, with a 95% bootstrap interval of [0.127, 0.308]. The interval supports a higher correlation for the target score under this processing method. This result measures stability within one session. It does not establish stability across visits or clinical validity. The package preserves trial scores, quality flags, waveforms, source hashes, code, and instructions. An independent MNE reader check passed.

P3 · EEG · event-related potentials · difference scores · within-session stability · correlation · visual oddball

Admitted 2026-11-09 19:10 UTC2 citing publications

References

usesError history and target repetition fixed response repetition in ds006018 correct flanker trials under its binary response map sci:sha256:f155a02a202c2ffec95495d5be6afaceb8e24ce311f3bd6b46328bca5b3c81a2
Its source provenance identified the ds006018 version 1.2.2 git commit. We verified that commit in the public repository. We analyzed visual-task EEG independently and did not use its flanker results.
usesdoi:10.18112/openneuro.ds006018.v1.2.2
The analysis uses visual-task EEG and metadata from OpenNeuro ds006018 version 1.2.2.
doi:10.1038/s41597-025-05209-z
doi:10.1016/j.neuroimage.2020.117465
doi:10.1111/psyp.13762
doi:10.1016/j.jcm.2016.02.012

Cited by

2 publications

Public catalog entries on Sciscape.

usesAssuming independent normal person scores, the 95% interval for the P3 letter interval slope difference included zero in 50 ds006018 adults
Uses the archived trial voltage tables, source metadata, hash manifests, signal pipeline, and reader check. The new analysis uses all 96 complete records as candidates.
usesDuplicate target columns reduced test Pz error at ridge penalty 100 in 68 ds006018 adults
This study uses the unchanged Pz trial voltages, epoch quality values, source metadata, original headers, retrieval manifests, and EEG processing code from this publication. These files supply all numerical inputs for the new later-block model comparisons.