Triangulation Research
Measurement validity for study designs and diagnostic criteria.
Drug Target Evidence
Observational and Mendelian randomization evidence for 41 drug mechanisms, set against Phase III outcomes. A preregistered rule classifies 24 of 32 correctly, and the observational evidence adds nothing beyond MR status.
Cross-Design Evidence 1 paperFederated Analysis
Across 42 million COVID-19 records, pooled site-level estimates measure a different quantity from the patient-level effect. Grouping, not sample size, decides whether they detect anything.
Ecological Inference 1 paperCriterion Overlap
Two criteria applied to one population yield a 2×2 table, and one kappa can come from nested or from crossing diagnoses. The table can be recovered from the figures a study prints: seven of nine eligible comparisons identify a single table.
Diagnostic Boundaries 1 paperAlzheimer's Criteria
The two 2024 criteria place the same cognitively unimpaired, biomarker-positive person on opposite sides of the disease boundary, and neither follows from the biomarker profile both accept.
Diagnostic Boundaries 1 paperMS Criteria
Six classifications of multiple sclerosis sort six different things. Two diagnostic rules applied to one cohort of 309 patients moved the disability rate inside the diagnosed group from 46% to 27%, with no patient's course changed.
Diagnostic Boundaries 1 paperPsychiatric Classification
The DSM, the Research Domain Criteria, the Hierarchical Taxonomy of Psychopathology and network theory each presuppose a different answer to when two presentations count as the same disorder.
Diagnostic Boundaries 1 paper