Materials
Calculators and utilities we built for our own work and kept: agreement statistics, rating-scale scorers, transcript preparation, study planning. Every one runs entirely in your browser—no account, and nothing you paste is ever uploaded.
Compute and plan inter-rater agreement: every coefficient side by side, with confidence intervals and sample sizes.
Percent agreement, Cohen's κ, weighted κ, Scott's π, Fleiss' κ, Gwet's AC1 and Krippendorff's α—computed together, so you can see where they disagree.
Open toolHow many items does a reliability substudy need? Plans N for a target kappa or ICC confidence-interval width, and shows the precision curve.
Open toolConvert Dice to IoU and back with the exact identity, or enter two mask areas for both metrics, a to-scale Venn diagram, and the kappa-on-pixels demonstration.
Open toolAll six Shrout & Fleiss forms with exact F-distribution intervals, both naming conventions, and a decision aid that picks the form your design actually calls for.
Open toolTurns your study details into a paste-ready methods and results paragraph, and audits it live against the nine reporting elements the GRRAS checklist asks of one.
Open toolPairwise F1 under all four SemEval-2013 matching schemes, boundary IoU with a threshold slider, and a token-level bridge—marked directly on the text.
Open toolAlpha with an exact Feldt interval, corrected item-total correlations, alpha-if-deleted, and a reverse-keying check that catches the most common data-entry error.
Open toolSEM by all four published definitions, MDC90 and MDC95, the group-level threshold, and an honest verdict when the definitions disagree about your change.
Open toolHow much a positive screen actually tells you, with verified PHQ-9, GAD-7, PHQ-2 and GAD-2 presets and a 1,000-person icon array.
Open toolEstimate the unglamorous parts of a transcript study before you commit to them: time, cost and turnaround.
How long will an hour of audio actually take? Estimates typing time, word count, cost and turnaround across DIY, professional and AI-plus-review workflows.
Open toolItems × passes × time × wage, plus QA and setup—to a total, a cost per label, and a timeline that assumes a realistic working day.
Open toolSearch 187 categories from 25 published instruments by construct, edit the entries, and export a citable codebook as YAML, Markdown or CSV.
Open toolWER, CER, MER and word information lost, with a colour-coded diff and normalisation toggles for case, punctuation, speaker labels and fillers.
Open toolA decision aid over 25 curated instruments: filters by construct, rater, purpose and unit of analysis, and explains every near miss.
Open toolPrepare sensitive transcripts for sharing and analysis, without the text ever leaving your browser.
Score the standard clinical instruments, with severity bands, subscale readouts and the scoring quirks each scale is known for.
Nine items to a 0–27 total with severity band, the ≥10 screening threshold, and an explicit item-9 risk flag.
Open toolSeven items to a 0–21 total with severity band and the ≥10 clinical-significance threshold.
Open toolFourteen items to a 0–56 total, with psychic and somatic anxiety split out and both competing cutoff conventions shown side by side.
Open toolSeventeen items on mixed 0–4 and 0–2 ranges, with severity band, remission flag, and a view of the scale's uneven item weighting.
Open toolTen items to a 0–60 total with Snaith bands, remission threshold, response calculation and an item-10 risk flag.
Open toolEleven items with the four double-weighted ones handled correctly, plus a subtotal that exposes the scale's signature scoring quirk.
Open toolEighteen items on either numbering (0–6 or the original 1–7), with Shafer's five factor subscores and an approximate PANSS conversion.
Open toolTranslate depression totals between the two clinician scales using the published equipercentile tables—both directions, with bands, remission and change.
Open toolExplore the category systems behind an annotation scheme, and take them away as a label set you can code with.