The codebook clarified the question.
A BMJ CONSORT screener, in three drafts.
v1 · apr 10
study_type (one of 11 categories)
study_design (one of 14 categories)
Readers argued about the taxonomy, not the trial.
↓
v2 · apr 17
manuscript_type (one of 4)
randomisation_evidence
A health-economics paper cited a trial it wasn’t reporting.
↓
v3 · apr 17
analyzing_data (yes / no)
patient_level_health_outcomes (yes / no)
The last draft was the first time we knew what we were looking for.
@karlrohe