Consensus & statistics · Administrator
Use range consensus on rating scales
Group ordered ratings into ranges and distinguish any-range, high-range, and combined inclusion/exclusion rules.
Screenshots from the original Surveylet guide illustrate the controls below. Layouts and labels may vary with your survey configuration. Select an image or “View full size” to open it in a new tab.
What a range means
Range consensus combines adjacent choices on an ordered scale. On a nine-point scale with three ranges, ratings 1–3 form the low range, 4–6 the middle range, and 7–9 the high range. On a five-point scale, three ranges use 1–2, 3, and 4–5.
Set the Likert Scale flag and verify the stored order of the choices. Range Label changes the text shown for a range; it does not change which choices belong to it. Use the actual supported scale allocation and test non-standard scale lengths.
Choose the intended rule
| Method family | Interpretation |
|---|---|
| Range Consensus 2, 3, 4, or 5 | Can establish agreement in a range meeting the configured threshold. |
| Range Consensus X Hi | Focuses on the highest range rather than agreement in any range. |
| Range Consensus 3:1570 | Uses a high-range threshold together with a low-range limit of less than 15%. |
| Range Consensus 3:1570 7015 | Adds a corresponding consensus-out rule; other outcomes remain undetermined. |
| Patient- or group-specific variants | Apply additional stakeholder criteria. Confirm the variant and group mapping against the protocol. |
Similar names can represent different decisions. Do not select a method only because its name includes a familiar percentage. Confirm both inclusion and exclusion behavior.
A worked inclusion example
Suppose the protocol specifies at least 75% of valid ratings in 7–9 and fewer than 15% in 1–3. Of 20 valid ratings, 16 are high, 2 are middle, and 2 are low. High is 80% and low is 10%, so the stated inclusion rule is satisfied.
If 16 are high, 1 middle, and 3 low, the low share is 15%, which fails a less than 15% condition. Test boundary cases in Surveylet because the required comparison signs are part of the rule, and a rounded display is not enough to resolve a boundary.
Keep the denominator explicit
Exclude the virtual “I don’t know” non-rating from scored calculations and distinguish valid ratings from total questionnaire participation. If the protocol uses another denominator or stakeholder-specific rules, verify that the selected built-in method matches it. Export the counts used for reporting.