Statistical Audit
Statistical errors in published research are rarely calculation mistakes — they are design and configuration mistakes. Wrong test for the data structure. Multiple comparisons without correction. Pseudoreplication treated as independent data. Statistical Audit is a collection of checks and guards that surfaces these risks before export.
What is Statistical Audit?
Statistical Audit is Licklider's pre-export diagnostic system. It aims to evaluate figures across several dimensions — including normality, effect size, confidence intervals, multiple comparison compliance, replication type, missing data disclosure, outlier criteria, and power justification. Not all dimensions are fully implemented; see individual pages for current coverage.
Beyond scoring, Statistical Audit operates through several guard layers that can block or warn at specific decision points: test configuration, experimental design, confounding, regression diagnostics, and robustness. Guards are tiered: some guards can affect export status, others surface risks with explanations, and others provide best-practice suggestions. Implementation depth varies by guard; individual pages document current product behavior and known limitations.