Forms · Glossary

What is a Likert scale?

A Likert scale measures agreement with a statement across a set of ordered options, most often five points from strongly disagree to strongly agree. It was introduced by Rensis Likert in 1932. Each point is labelled rather than numbered, and the responses are ordinal, which is what decides how you are allowed to analyse them.

Almost every satisfaction survey uses one, and most of them get two decisions wrong: how many points, and whether the middle one exists.

· Co-founder

3 min read · Published

Five points or seven

Five is the workhorse: strongly disagree, disagree, neither, agree, strongly agree. It is quick to read and fits on a phone. Seven gives more room to discriminate and is worth it when you expect responses to bunch at one end, which is common in customer satisfaction. Beyond seven the extra points stop meaning anything distinct to a respondent, and the data does not get better.

Whether to include a midpoint

A midpoint gives people who genuinely have no view somewhere honest to go. Removing it, a forced choice scale, pushes those people to pick a side and adds noise you cannot distinguish from real opinion. The case for removing it is that some respondents use the middle to avoid thinking. On balance, keep it, and add a separate not applicable option for people the statement does not apply to, which is a different thing from being neutral.

Label every point, not just the ends

A scale labelled only at the ends leaves the middle points to interpretation, and different respondents interpret them differently. Labelling all five costs a little space and makes the answers comparable. It also helps translation, since a numbered scale with two anchors travels much worse between languages than five named points.

Questions people ask

Can I average Likert responses?

Strictly the data is ordinal, so the distance between agree and strongly agree is not known to be the same as between neutral and agree, and a mean is a compromise. In practice means are widely reported and useful for tracking movement over time. Report the distribution alongside, because a mean of three can be everybody neutral or half at each extreme.

Should I use a neutral middle option?

Usually yes. Forcing a choice does not create an opinion, it just makes the people without one pick at random, which is worse than a clear count of people who do not care. Keep the midpoint and add a separate not applicable option for anyone the statement genuinely does not reach.

Is a Likert scale the same as a rating scale?

Not quite. A Likert scale measures agreement with a statement across labelled, ordered points. A rating scale asks for a score, often out of five or ten, without agreement wording. They are analysed similarly but written differently: a Likert item is a sentence somebody agrees or disagrees with.

How many statements can I put on one scale?

On a form, five to eight before people start straight lining down a column. That is a limit of attention rather than of the scale. If your instrument genuinely has twenty items, break them into groups with a heading between, which resets attention at each break.

Should the positive end be on the left or the right?

Pick one and keep it consistent across the whole survey, and across surveys you compare over time. Flipping the direction between questions to catch inattentive respondents mostly catches attentive ones having a bad day, and it makes your own analysis error prone.

Make one with forms

The button opens the generator with this use case already described. Change the wording to match your own.

Create a form with OneCraft

Related questions

Sources

Written and checked by the OneCraft team. Last checked .