Blind Analysis
The practice of locking analysis methods before seeing the data, used in high-stakes empirical work to defend against bias toward expected answers.
Blind analysis is the practice of locking analysis methods, parameter sweeps, and decision rules before the analyst is shown the data the analysis will be applied to. The protocol is a structural defense against the bias toward expected answers, named in Klein and Roodman's canonical 2005 Annual Review of Nuclear and Particle Science paper on the practice and developed across the empirical sciences in MacCoun and Perlmutter's 2015 Nature essay. The Event Horizon Telescope's four-team blind imaging protocol is the hackathon-relevant worked example; the structural extension to multi-team replication is the substance of integrity-through-convergence. See also embargo, methodological-diversity, and convergence-event.