Nathan Humphrey's writings
2026-05-30 methodology doi · 10.5281/zenodo.20467200

When a Confound Diagnostic Breaks: Pre-Registered Characterization of Peak-Phase Offset Across a Resolution × Reference-Geography Boundary, Calibrated on the Shark-Attack and Ice-Cream-Search-Interest Pair

by Nathan Humphrey
abstract

The positive association between shark-attack incidence and ice-cream sales is one of the most-cited textbook examples of confounded correlation in observational data. To our knowledge, the pair has never been empirically decomposed in the peer-reviewed literature. We pre-register peak-phase offset analysis as the discriminating diagnostic and characterize its behavior across resolution × reference-geography.

read full paper on zenodo doi.org/10.5281/zenodo.20467200
cite
APA Humphrey, N. (2026). When a Confound Diagnostic Breaks: Pre-Registered Characterization of Peak-Phase Offset Across a Resolution × Reference-Geography Boundary, Calibrated on the Shark-Attack and Ice-Cream-Search-Interest Pair. Zenodo. https://doi.org/10.5281/zenodo.20467200
BibTeX 
@misc{humphrey-2026-confound,
  author    = {Nathan Humphrey},
  title     = {When a Confound Diagnostic Breaks: Pre-Registered Characterization of Peak-Phase Offset Across a Resolution {$\times$} Reference-Geography Boundary, Calibrated on the Shark-Attack and Ice-Cream-Search-Interest Pair},
  year      = {2026},
  month     = {may},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.20467200},
  url       = {https://doi.org/10.5281/zenodo.20467200}
}
in production

The pre-registration discipline characterized here is the one we're building ResolveOSINT around — pre-registered chronology and provenance verification for breaking-news events (coming soon). The same confound-diagnostic framework already underwrites operator selection in the Resolve detector.