Survey templates
Likert scale examples: 5-point and 7-point templates
Copy balanced Likert scale examples, choose 5 or 7 points, separate neutral from unsure, and avoid common survey-analysis mistakes.
A five-point agreement item commonly uses strongly disagree, disagree, neither agree nor disagree, agree, and strongly agree. A seven-point version adds more gradations. Choose labels that match the judgment you need, and keep “cannot judge” separate from a neutral opinion. The examples below show both agreement items and other useful rating formats.
Which scale should you use?
Start with the decision, not the number of boxes. If you need to know whether a price explanation is understandable, ask about clarity. If you need to measure agreement with a particular proposition, an agreement item may fit. Asking respondents to agree that an explanation is clear adds an extra layer between the question and the judgment.
The examples below distinguish agreement items from direct ratings of clarity, ease and relevance. Decide what each question measures before choosing its labels. Sharing a five-point format does not make two questions interchangeable or justify combining their answers into one score.
| Your decision | Useful format | What it tells you |
|---|---|---|
| Is the explanation clear? | Direct clarity rating | Perceived clarity, which still needs a comprehension check. |
| Do people agree with this statement? | Balanced agreement item | Agreement with the particular statement shown. |
| How often did people use a feature? | Defined frequency categories or a count | Reported behavior in a specified period. |
| Which version is preferred? | Explicit preference question | Relative preference in the comparison shown. |
Sources: Pew Research Center on questions and response options
Copy these 5-point and 7-point agreement labels
Use one clear statement at a time. For a fictional account page, an example is: “I can tell when the subscription will renew.” A person should not have to combine their views about price, cancellation and renewal to answer one item.
Keep the labels in their logical order. Use the same direction across a short questionnaire unless a deliberate research design calls for something else. An unexplained switch from positive-first to negative-first can create avoidable mistakes. These are starting templates to pretest, not a universal validated scale.
| Position | 5-point agreement | 7-point agreement |
|---|---|---|
| 1 | Strongly disagree | Strongly disagree |
| 2 | Disagree | Disagree |
| 3 | Neither agree nor disagree | Somewhat disagree |
| 4 | Agree | Neither agree nor disagree |
| 5 | Strongly agree | Somewhat agree |
| 6 | — | Agree |
| 7 | — | Strongly agree |
| Separate option | Cannot judge from the information shown | Cannot judge from the information shown |
Five direct-rating examples for marketing research
Sometimes the question is easier to answer when it names the judgment directly. Keep the same object and time period in the question and answer options. Include an experience or uncertainty option when respondents may lack a basis for judging.
Do not use a seven-point format just because it looks more precise. More response categories do not automatically produce a more accurate decision. Pretest whether the audience understands the distinctions, particularly on mobile.
| Question | Five ordered answers | Separate answer if relevant |
|---|---|---|
| How clear or unclear is the renewal explanation? | Very unclear; somewhat unclear; neither clear nor unclear; somewhat clear; very clear | Cannot judge |
| How easy or difficult was it to find the total price? | Very difficult; somewhat difficult; neither easy nor difficult; somewhat easy; very easy | Did not try |
| How believable or unbelievable is this claim? | Very unbelievable; somewhat unbelievable; neither believable nor unbelievable; somewhat believable; very believable | Not enough information |
| How satisfied or dissatisfied were you with the last support interaction? | Very dissatisfied; somewhat dissatisfied; neither satisfied nor dissatisfied; somewhat satisfied; very satisfied | No support interaction |
| How relevant is this offer to your current needs? | Not at all relevant; slightly relevant; moderately relevant; very relevant; extremely relevant | Cannot judge |
How to handle neutral, unsure and missing answers
Consider two people looking at a fictional cancellation policy. One understands the policy and has no positive or negative opinion. The other cannot find the cancellation terms. Coding both as the midpoint hides a useful product problem.
Keep a distinct code for cannot judge, not applicable, and skipped where those states matter. Decide which cases belong in each denominator before reporting. If you show a favorable share among people who expressed an opinion, also show how many people could not judge. Otherwise, a clearer-looking number may be created by removing the very readers who need help.
For example, in an invented set of 20 responses, eight favorable answers are 40% of all 20. If four people could not judge, the same eight answers are 50% of the remaining 16. Both calculations can be described accurately, but they answer different questions. These are teaching numbers, not collected responses.
What to check when AI writes the scale
Ask the assistant to state the exact construct, identify the opposite endpoints, and explain why a respondent might need an uncertainty option. Review each label yourself. A frequent draft problem is an enthusiastic set of positives paired with only one mild negative.
Use this review instruction with the draft: “Check whether every answer option measures the same thing. Flag overlapping labels, unequal positive and negative intensity, missing uncertainty options, and questions that ask about two separate judgments. Suggest changes without claiming the instrument is validated.”
Our survey question checker can flag some wording patterns and prepare a tailored review prompt. Its local checks do not validate a scale or prove that a question is unbiased. Testing the questionnaire with relevant people remains a separate step.
How should you report the results?
Start with counts and percentages for each response category, together with the wording, order and analysis base. If you report a mean, state the numeric coding and why that summary is useful for this measure. If you combine items, explain the construct and provide a justification for the scoring rule.
A mean should not replace the distribution. Two messages can receive the same average while one produces mostly middle ratings and the other sharply divided views. That difference may change the next research question.
Can you compare five-point and seven-point results directly? Usually you should not treat them as the same measure without a justified conversion and comparability assessment. Can a clarity rating replace a comprehension question? No: perceived clarity and accurate understanding are different outcomes.
Sources and scope
- Writing Survey QuestionsPew Research Center
Guidance on matching question wording and response options, preserving meaningful order and pretesting questions. The examples and reporting workflow here are MTB editorial guidance.
- Statistical Quality Standard A2U.S. Census Bureau
Primary standards for developing and pretesting data-collection instruments; not certification of these templates.
Sources support the specific statements described above; they do not validate this publication’s rubric, guarantee a compliant execution, or replace context-specific professional advice.
What this page is: a research-methods guide, not a report of completed consumer fieldwork. Corrections and material revisions are recorded under the publication’s editorial standards.