HARK No More: On the Preregistration of CHI Experiments
summary
Argument that mandatory experimental pre-registration reduces HARKing and strengthens HCI empirical claims, with registries recording hypotheses, methods, recruitment and analysis before data collection. Evidence from clinical trials shows published significant results drop dramatically (reported 57% → 8% after registries). Existing tools named include clinicaltrials.gov, Open Science Framework and AsPredicted, and a dedicated ACM/CHI registry is proposed to open file drawers, boost null-result reporting, encourage replications and discourage p‑hacking.
AI-written summary · about