What problem does it solve? Research findings often fail because biases like self-selection, leading questions, and confirmation bias distort results, and vague advice like "stay objective" does nothing to prevent them. This Skill provides a canonical reference that pairs every named bias with a concrete control, so studies can be audited before fielding and findings can be trusted before they gate decisions. ## Core Features & Use Cases - Bias-to-control mapping: Every named bias across five lifecycle stages (sampling, instrument design, moderation, analysis, study-level confounds) is paired with a concrete design change that removes or bounds it. - Two audit modes: A fast triage path covering the top 5 study-killers for low-stakes work, and a full audit path for findings that gate decisions or external claims. - Internal vs external validity separation: Forces explicit statements about whether a claim is valid within the study, generalizable beyond it, or both. - Use Case: Before shipping a redesign based on "5 of 6 beta users said the new flow is easier," run the bias audit to catch self-selection, social desirability, survivorship, and overgeneralization, then attach concrete fixes like recruiting lapsed users and reporting prevalence as X of Y. ## Quick Start Audit my planned usability study for biases and tell me which threats actually endanger my conclusion and what concrete controls to add.
