Likert scale

Likert scale meaning and pronunciation

The scale is named after Rensis Likert, an American social psychologist who described the method in 1932. His name is pronounced LICK-ert, not LIKE-ert, although you will hear both in offices and lecture halls.

The idea was simple and still works: instead of asking «do you like your manager?» and getting a yes or a no, you give people a statement and a graded range of agreement. The grading is what turns a vague opinion into something you can count.

Likert item vs Likert scale

Strictly speaking, one statement with five answer options is a Likert item. A Likert scale is the set of several items that measure the same thing, whose answers are added up or averaged into one index. «My workload is manageable», «I rarely work late» and «I have time to rest at weekends» are three items of one workload scale.

In everyday survey work people call a single item a Likert scale too, and nobody gets hurt. The distinction matters when you report results: a single item gives you a distribution, a proper scale gives you a score per respondent.

Likert scale examples for a questionnaire

Statements that work well are short, about one idea, and written in the first person:

• I understand what is expected of me at work.
• The checkout process was easy to complete.
• I would trust this company with my personal data.
• The course material was relevant to my job.
• Support resolved my issue without me having to repeat myself.

The standard five-point labels are: strongly disagree, disagree, neither agree nor disagree, agree, strongly agree. Frequency versions («never» to «always») and importance versions («not important» to «very important») follow the same logic.

5-point or 7-point Likert scale?

Five points are the default for a reason: they are quick to read on a phone and people use the whole range. Seven points add resolution for audiences who think in nuances, such as researchers or employees answering a long engagement study. Three points suit children, very short pulse checks and moments when you only need direction.

Keep the same number of points for every item in one scale. Mixing five and seven makes the summed score meaningless.

How to analyse and interpret Likert scale data

Likert answers are ordinal: «agree» is more than «neutral», but nobody promised the gap between them equals the gap between «neutral» and «disagree». That is why the safest summaries are the median, the mode and the distribution. The mean is widely reported and fine for tracking change over time, but it is debated as a precise measure, so do not read decimals into it.

The most readable number for a stakeholder is top-2 box: the share of people who chose the two most positive points.

Top-2 box = (answers «agree» + «strongly agree») / all answers × 100%
Example: 120 answers, 54 agree, 30 strongly agree
Top-2 box = (54 + 30) / 120 × 100% = 70%

For a multi-item scale, first reverse any negatively worded item (on five points, 1 becomes 5 and 2 becomes 4), then sum or average the items per respondent. Only after that compare groups.

When a Likert scale is the wrong tool

• You need a priority. Everyone agrees that price, speed and quality matter; use ranking or a constant sum instead.
• You are measuring perception with two opposites, like «dull» and «exciting». A semantic differential fits that better.
• You want a loyalty benchmark comparable with other companies. That is what NPS exists for.
• The statement is a fact, not an attitude. «I use the app daily» belongs in a frequency or yes/no question.

Common Likert question mistakes

• Double-barrelled items: «The app is fast and easy to use» cannot be answered honestly if it is fast but confusing.
• Only positive wording. People slide into agreeing with everything; mix in one or two reversed items.
• Unbalanced labels, such as three shades of agreement against one of disagreement.
• Twenty items in a row. Fatigue sets in and answers drift to the middle; split long batteries into sections.

Building a Likert question in Askyo

The Likert block starts with five points and the end labels «Strongly disagree» and «Strongly agree». A three-way switch in the editor changes it to 3, 5 or 7 points, and both end labels can be rewritten, so the same block covers frequency or importance scales.

Respondents see one horizontal row of numbered buttons from 1 to N, including on a phone. On seven points that row gets tight on small screens, which is one more argument for five. For a battery of statements, add one block per statement.

In results each item shows its average out of N, the top-2 and bottom-2 shares and a bar per point. The responses CSV gives every block its own column, so reverse coding and summed scores take a couple of spreadsheet formulas.