Dr. Kilem Gwet — Inter-Rater Reliability Repository

Two raters can agree 90% of the time and still score a kappa of zero.

AgreeStat Analytics is Dr. Gwet's research and software home for measuring how well raters, coders, and diagnostic instruments actually agree — covering Cohen's Kappa, Gwet's AC1/AC2, Krippendorff's alpha, and intraclass correlation.

Try it — two raters, binary rating

See the paradox for yourself

Edit the 2×2 table below. Watch raw agreement stay high while Cohen's Kappa collapses — and Gwet's AC1 stays honest.

Rater A × Rater B
B: YesB: No
A: Yes
A: No
n = 100 subjects rated
Raw observed agreementPercent of subjects both raters scored the same
Cohen's KappaChance-corrected, sensitive to prevalence
Gwet's AC1Chance-corrected, stable under skewed prevalence

With the default counts, agreement runs high even though most subjects fall in one category. That's exactly the condition under which Kappa is known to misbehave — the puzzle Dr. Gwet's research addresses directly.

What you'll find here

A working repository, not a brochure

Everything on this site exists to help you compute, defend, and publish inter-rater reliability results correctly.

Papers

Free articles

Dr. Gwet's published and unpublished papers on chance-corrected agreement coefficients, made freely available for researchers and reviewers.

Browse the papers →
Code

Computer programs

Download reliability-coefficient routines in R and SAS, built to reproduce the formulas exactly as documented in the Handbook.

Get the code →
Reference

Book excerpts

Read excerpts from the current and forthcoming editions of the Handbook of Inter-Rater Reliability before you buy.

Read excerpts →
AgreeStat360 · Excel / Windows

Statistical analysis your committee can open in Excel

Compute chance-corrected agreement coefficients and intraclass correlation coefficients without leaving a spreadsheet your co-authors already trust.

Walkthrough of AgreeStat360 for Excel/Windows. Prefer a bigger screen? Watch on YouTube ↗

CAC statistics include Cohen's Kappa, Gwet's AC1/AC2, and Krippendorff's alpha. ICC statistics cover several coefficients built on different ANOVA models.

CAC coefficients
ICC coefficients
Cohen's Kappa
One-way random ANOVA ICC
Gwet's AC1 / AC2
Two-way random ANOVA ICC
Krippendorff's alpha
Two-way mixed ANOVA ICC
Handbook of Inter-Rater Reliability, 5th Edition, Volume 1 — categorical ratings Handbook of Inter-Rater Reliability, 5th Edition, Volume 2 — intraclass correlation
Now in two volumes

Handbook of Inter-Rater Reliability, 5th Edition

The reference text behind most of the methods on this site, now split by rating type — Volume 1 for categorical ratings, Volume 2 for quantitative ratings — so you only need the volume that matches your data.

In the browser

No install required

Two free-to-try tools for teams who'd rather work from a browser tab than a spreadsheet.

Cloud

AgreeStat360 / Cloud-Based

Point-and-click inter-rater reliability analysis, right in your browser. Register for a trial at agreestat360.com — any paid license key also works with AgreeStat360 for Excel/Windows.

Free

AgreeTest / Cloud-Based

Test whether the difference between two agreement coefficients is statistically significant — entirely from your browser, no spreadsheet setup needed.

AgreeTest interface preview