# Top-two-box scores: count the eligible ratings

Calculate a top-two-box percentage with explicit rating direction, valid-answer count and not-applicable exclusions.

If nothing was unclear or difficult, say so. Skip questions about steps you did not experience; use not applicable where appropriate.

## 1. What does the lowest rating label mean to you?

Answer: ____________________

## 2. What does the highest rating label mean to you?

Answer: ____________________

## 3. Which two labels would describe a favourable experience, and why?

Answer: ____________________

## 4. What would you choose if you had not experienced the service?

Answer: ____________________

## Author notes / illustrative keys where applicable

1. Check that the low endpoint refers to the intended negative experience.
2. Verify the high endpoint’s direction rather than relying on numeric code alone.
3. Test how people interpret the selected favourable categories.
4. Keep non-exposure separate from neutral experience.

Interpretation caution: A top-two-box share depends on the chosen labels and denominator. It is not NPS, a validated scale or evidence that numeric codes have equal distances.


## Illustrative finding and follow-up

Observation: Synthetic counts on a low-to-high five-point satisfaction scale are 1,1,2,3,5; three separate cases are not applicable.

Action to test: Use the top labels 4 and 5 as eight favourable ratings among 12 valid ratings; report the three not-applicable cases separately.

Follow-up: After revising ambiguous labels, pilot both endpoints again; do not directly compare scores collected under different labels.


## Collection protocol

When and whom to ask: Pilot scale wording before fielding the main survey with people who have and have not experienced the service.

Decision owner: The questionnaire owner approves endpoint labels; the analyst retains the valid-answer rule.

Follow-up: Review changed label interpretation before deciding whether later numeric summaries are comparable.

Response handling: Pilot questions are free text. This worksheet’s scale counts are manually supplied examples, not a claim about native rating-field or scoring availability.


## Worked measurement study

All example records are synthetic; manual worksheet, no automatic import.

### Inspect the synthetic example

Scale label or disposition | Synthetic count
--- | ---
1: very dissatisfied | 1
2: dissatisfied | 1
3: neither | 2
4: satisfied | 3
5: very satisfied | 5
Not applicable | 3

### Derivation

Valid ratings = 1+1+2+3+5=12. Top two = 3+5=8. Top-two-box = 8÷12×100≈66.7%. Three not-applicable cases are disclosed outside the rating denominator.

Score note: eight of 12 valid ratings select the two stated favourable labels, approximately 66.7%; three not-applicable cases remain separate.

Calculator inputs and convention: {"mode": "ratio", "labels": ["Favourable valid ratings", "All valid ratings"], "values": [8, 12], "unit": "% top-two-box", "integer": true, "subset": true}

### Evidence boundary

Observed / Category labels and counts / Supports a labelled eight-of-12 share.
Unknown / Comparability with another scale / Do not equate top-two-box with NPS or an unlabeled average.
Next check / Pilot endpoint meaning and exclusion policy / Preserve the rule when comparing later waves.

### Method references

Pew Research Center question design: https://www.pewresearch.org/writing-survey-questions/ — Wording and response-option consistency.