Content Validity Index (CVI)

Verified

Reliability (legacy hub)

Independently verified. Every statistic this test reports has been re-derived against an independent reference — never the library the pipeline itself calls — the rendered output was read back in a browser, and the result is locked with a committed regression suite.

Run this test straight away on a free built-in teaching dataset — no data of your own needed — or bring your own. Either opens the guided workspace: variable setup, assumption diagnostics, results with effect sizes and confidence intervals, figures, and APA-ready reporting.

Loading teaching datasets…

Or use your own dataset

Loading your datasets…

Content Validity Index — the proportion of experts who rate an item (or the whole scale) as relevant.

The item-level CVI (I-CVI) and scale-level CVI (S-CVI) summarize expert agreement on relevance; I-CVI ≥ .78 and S-CVI/Ave ≥ .90 are the usual thresholds.

Worked example

Is the scale's content valid by expert consensus?

Ten experts rated each item's relevance (relevant / not); I-CVI and S-CVI summarize agreement.

Result

18 of 20 items had I-CVI ≥ .78 and the S-CVI/Ave was .94.

How you'd report it (APA)

Content validity was supported, S-CVI/Ave = .94 (18 of 20 items at I-CVI ≥ .78).

Try it yourself: Load this ready-made sample and follow the run above.

When to use it

  • Expert relevance ratings, item and scale level
    A panel rates each item relevant / not; you want I-CVI per item and S-CVI for the whole scale. e.g. 10 experts, 20 items.
  • Deciding which items to keep
    I-CVI ≥ .78 and S-CVI/Ave ≥ .90 give clear retention thresholds.

When NOT to — use instead

  • Experts used a graded relevance scale
    If ratings are 1–5 rather than relevant/not, a rescaled index fits better. → Aiken's V
  • You want the 'essential' judgement
    Lawshe's essential/not format is the CVR. → Content Validity Ratio

Assumptions (and what to do if they fail)

Chance agreement with few expertshigh

Check: With a small panel, a high I-CVI can arise by chance; report a chance-corrected agreement (modified kappa) where possible.

If violated: An item passes on luck; the modified-kappa adjustment is exactly what guards against it.

Consistent dichotomisation of relevancemedium

Check: A 4-point relevance scale must be collapsed to relevant (3–4) vs not (1–2) the same way for every expert.

If violated: Different cut rules across experts make the proportions non-comparable.

Adequate, qualified panelmedium

Check: 5–10 content experts is the usual range; the S-CVI benchmark assumes a real panel.

If violated: Too few or ill-matched experts make content validity a formality rather than evidence.

Ready to run a Content Validity Index (CVI) on your own data?

Guided setup, automatic assumption checks, effect sizes, figures and an APA write-up.

Run this test →