S-CVI/UA (Scale-level Content Validity Index/Universal Agreement) measures the proportion of items on which all experts unanimously agreed the item is essential. It is a more stringent validity criterion than S-CVI/Ave, requiring 100% expert consensus rather than just a high average.
S-CVI/UA is particularly useful when:
- You need a high-stakes instrument (certification exams, safety-critical assessments)
- Expert consensus is critically important
- You want to complement S-CVI/Ave with a stricter measure
- You're developing instruments for sensitive domains
S-CVI/UA Formula
S-CVI/UA = Items with I-CVI of 1.0 / Total number of items
In Plain English
S-CVI/UA is simply the percentage of items that were rated "essential" by every single expert on your panel, with no exceptions.
Example Calculation
A 10-item scale with expert ratings (5 experts per item):
Items with unanimous agreement (I-CVI = 1.0): 1, 2, 4, 7, 8
Items WITHOUT unanimous agreement (I-CVI < 1.0): 3, 5, 6, 9, 10
S-CVI/UA = 5 unanimous items / 10 total items = 0.50 or 50%
Validity Threshold
S-CVI/UA ≥ 0.80: Excellent validity ✓
At least 80% of items achieved perfect expert consensus. This is considered excellent.
S-CVI/UA 0.60–0.79: Acceptable, but not excellent
Some items lack unanimous agreement. Consider revising items with lower I-CVI values.
S-CVI/UA < 0.60: Needs improvement
More than 40% of items lack unanimous agreement. Revision needed.
S-CVI/Ave vs S-CVI/UA
These two metrics measure different things and are often reported together:
| Metric | Definition | Threshold | Focus | Stringency |
|---|---|---|---|---|
| S-CVI/Ave | Average of all I-CVI values | ≥ 0.90 | Overall item agreement | Moderate |
| S-CVI/UA | % of items with I-CVI = 1.0 | ≥ 0.80 | Unanimous agreement | Strict |
What They Reveal Together
Scenario 1: S-CVI/Ave = 0.92, S-CVI/UA = 0.85
✓ Excellent! Experts generally agreed items were essential, AND 85% of items had perfect consensus. This is ideal.
Scenario 2: S-CVI/Ave = 0.90, S-CVI/UA = 0.50
⚠ Good average agreement, but experts disagreed on half the items. Those low-consensus items should be revised.
Scenario 3: S-CVI/Ave = 0.88, S-CVI/UA = 0.90
✓ Surprising but positive: Nearly all items had unanimous agreement, but a few low-agreement items dragged down the average. Consider removing those few items for a higher S-CVI/Ave.
When to Use S-CVI/UA vs S-CVI/Ave
Use S-CVI/Ave When:
- You want the most commonly accepted validity measure
- Your scale can tolerate some items with less-than-perfect consensus
- You're following modern research standards
- You need a 0.90 threshold that's easier to achieve
Use S-CVI/UA When:
- You need exceptional, high-stakes validity (certification, safety)
- Consensus is paramount
- You want to identify which items have universal support
- You're supplementing S-CVI/Ave for added rigor
Practical Interpretation
If S-CVI/UA = 0.90
✓ Excellent! 90% of your items have perfect expert consensus. Your scale is ready for empirical testing. This is a strong indicator of content validity.
If S-CVI/UA = 0.75
⚠ Acceptable but not excellent. 75% of items have perfect consensus, but 25% need refinement. Review items with I-CVI < 1.0 and consider revisions.
If S-CVI/UA = 0.45
✗ Too many items lack unanimous agreement. This suggests either (a) your items need better wording, (b) your expert panel wasn't well-selected, or (c) your scale needs fundamental revision.
The "Agreement Paradox"
Important insight: An item can have high I-CVI (0.85) but not achieve I-CVI of 1.0 for S-CVI/UA purposes. This means:
- 85% of experts said "essential"
- But 15% said "useful but not essential" or "not necessary"
- It counts toward S-CVI/Ave but not S-CVI/UA
This is why some scales have lower S-CVI/UA but still-acceptable S-CVI/Ave — they have items with strong but not perfect consensus.
Combining Both Metrics
Best Practice: Report both S-CVI/Ave (≥ 0.90) AND S-CVI/UA (≥ 0.80) to show that your scale has:
- Strong average expert agreement
- Substantial items with perfect consensus
References
Davis, L. L. (1992). Instrument review: Getting the most from a panel of experts. Applied Nursing Research, 5(4), 194-197.
Shi, J., Mo, X., & Sun, Z. (2012). Content validity index in scale development. Zhong Nan Da Xue Xue Bao Yi Xue Ban, 37(2), 152-155.
Lynn, M. R. (1986). Determination and quantification of content validity. Nursing Research, 35(6), 382-385.