Likert scale examples: 5-point and 7-point templates

Free 1–5 scale survey template

Download the CSV to open in Excel or Google Sheets. It contains three separate examples: agreement with a pricing statement, perceived clarity, and ease of finding a price. Each row includes the wording, five answer labels, a separate cannot-judge code, and a note on what the answer can tell you. These are templates to adapt, not responses collected from people.

Replace the example subject with the exact message or task you will show. Keep the answer labels in order. Use a separate code such as 99 for cannot judge, and leave skipped responses blank; neither belongs in a 1–5 average. Test the wording with people from your intended audience before collecting results.

A single agreement question is a Likert-type item. A multi-item Likert scale combines related items under a justified scoring rule. The three examples in this file measure different things: do not add them together to produce a single score.

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.

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.

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.

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.

Worked scoring example: one item, two honest denominators

Take the single agreement item “The price information gives me enough detail to decide whether this offer fits my budget.” The table below uses invented counts to show the arithmetic. It is not a study result, a norm, or a recommended pass threshold.

There are 20 constructed responses: 16 on the 1–5 scale and four marked cannot judge. The favorable count is 6 + 2 = 8. Among all 20, the favorable share is 8 / 20 = 40%. Among the 16 valid scale answers, it is 8 / 16 = 50%. Report which denominator you use and keep the 20% cannot-judge share visible.

If a mean is justified for your analysis, exclude code 99 and skipped answers. The weighted total is (1 × 1) + (2 × 2) + (3 × 5) + (4 × 6) + (5 × 2) = 54; 54 / 16 = 3.375, or 3.38. Treating code 99 as a rating would instead produce (54 + 396) / 20 = 22.5, outside the scale. That is a coding error, not an unusually positive result.

A mean assumes the numeric spacing is useful for your purpose. Keep the category distribution alongside it, and do not infer comprehension from this opinion item. In a spreadsheet, use COUNTIF for individual codes and make the valid-code filter explicit; a plain AVERAGE over a column containing 99 will produce the wrong result.

Check the complete answer list

Read the question and every answer option together. Do the options measure the same thing? Are positive and negative labels equally strong? Could someone understand the question but lack the experience needed to answer?

Try a made-up response at each end of the scale, at the midpoint and outside the scale. For example, a person who has never used checkout needs a different answer from a person who feels neither satisfied nor dissatisfied.

The wording checker can flag some patterns, but it cannot settle these choices. Ask relevant people to explain how they interpret the question and choose an answer before using it in a study.

Sources: Census instrument-development standard

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.

Try one complete item before choosing a scale

For a fictional subscription page, use this item after showing the renewal information: “Please indicate how much you agree or disagree with this statement: I can tell when this subscription will renew.” Provide all five ordered choices: Strongly disagree; Disagree; Neither agree nor disagree; Agree; Strongly agree. Add “Cannot judge from the information shown” separately when that answer is possible.

Follow with “What renewal date did you find?” as an open answer. Someone may agree that the date is clear yet report the wrong date. Keep the agreement response and the date check separate in your report. Store cannot judge outside the 1–5 scale and leave skipped answers missing.

The editable template includes the question, all answer choices, the open follow-up and coding notes. Its two question blocks produced no signals in checker rules version 1.0 on September 23, 2026. That is a reproducible software observation, not validation of the item.

Sources: Edit the complete 5-point item; Load this example in the wording checker

Sources and scope

1. 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.

2. Statistical Quality Standard A2U.S. Census Bureau

   Primary standards for developing and pretesting data-collection instruments; not certification of these templates.