Likert scale generator

How this tool works

Each scale uses a complete set of labels chosen for the concept being measured. Read every option in order and check that the distinctions make sense for your question. Balanced labels do not establish that the distances between responses are numerically equal.

Bipolar scales run from one pole through a middle to its opposite: agree to disagree, satisfied to dissatisfied. Unipolar scales climb from none of something to a lot of it: never to always, unimportant to extremely important. The distinction decides the midpoint. On a bipolar scale the middle is a real position somebody can hold. On a unipolar one it is a quantity, so labelling it “Neutral” invites people with no view to pick a specific amount — there is no neutral amount of “often”.

An even number of points removes the midpoint. Choose this only when it fits the construct and audience; some respondents may genuinely hold a middle position. Offer “Don’t know” or “Not applicable” where appropriate and exclude those responses from numerical scale summaries.

Choose enough points to capture useful distinctions without asking respondents to make distinctions they cannot reliably judge. Check long labels on a phone. Keep the scale and direction consistent across items you plan to compare or combine, unless your validated instrument specifies otherwise.

Answers on a scale like this are ordinal: you know the order of the rungs, not that the step from “Agree” to “Strongly agree” is the same size as the one below it. Averaging them is common and often defensible, but report the distribution as well, and prefer the median or the share choosing each option when the shape is lopsided. The technique is Rensis Likert’s, from 1932; the format has outlived most of the ways people score it. More on the trade-offs in the Likert scale guide.