Multirater Kappa

Purpose-built for studies with multiple raters, this tool computes Fleiss' kappa to quantify agreement beyond chance. It accepts a rating matrix, category labels, and optional weighting to support ordinal data. The core calculation aggregates category frequencies across subjects to derive P_bar (overall agreement) and P_e (chance agreement), then forms kappa as (P_bar - P_e) / (1 - P_e). It supports missing data through exclusion rules or specific handling strategies, and can produce standard errors and 95% confidence intervals via normal approximation or bootstrapping. Outputs include the overall kappa, P_bar, P_e, standard error, confidence intervals, and a per-category agreement profile to identify skew or bias in category usage. The tool also offers a compatibility mode for two raters to compute Cohen's kappa, ensuring seamless interoperability with common study designs. Intended to support research planning, method reporting, and results interpretation in psychology, education, healthcare, and social sciences. It emphasizes reproducibility by exposing explicit input structures and calculation steps, and provides an auditable trail suitable for peer review.