A list of common Likert scales

A collection of the scales we've kept running into since 2002, with notes on when they fit and when they're better avoided.

Likert scales are among the most popular response formats in survey research. Originally developed to measure attitudes, they've long since become established for capturing many other concepts in a structured and comprehensible way, too. No wonder: they're simple and versatile, and they still deliver meaningful, easily interpretable results.

We keep getting questions like:

• "Which wordings are suitable for my scale?"
• "Are there standard wordings for agreement, frequency, or importance?"

That's why we've put together a practical overview for you: in this article you'll find 62 common Likert scales, clearly organized into 14 categories. Ideal for looking things up, getting inspired, or dropping straight into your survey.

In short: a Likert scale is a response format with fixed, graded answers. Respondents use it to say how much they agree with a statement or how they rate something, for example how satisfied they are, how often they do something or how important it is to them. The usual choice is 5 or 7 points with a neutral midpoint. If you don't want a midpoint, use 4 or 6 points. In everyday use, these formats are also called rating scales, and a question that uses one is often called a Likert scale question.

Let's look at an example. For the statement "The staff at the counter were friendly." you offer these five answers:

Strictly speaking, that's a Likert item: a single statement with graded answers. When Likert spoke of a scale, he meant several such statements on the same topic whose answers are added up. In everyday usage, though, the single response format is called a Likert scale too, and that's how we use the word on this page. We explain where the name comes from and what to watch for when you build one in Likert scales: everything you ever wanted to know, but never dared to ask.

👉 For anyone who wants to dig deeper into the topic: in a separate blog article we explain what Likert scales actually are and what to keep in mind when choosing one (e.g. an even or odd number of response options, how to word the poles, and so on), and how to avoid common mistakes.

But now, without further ado: here is the overview of the most common Likert scales:

The scales are grouped by what they measure. Under each heading you'll find a short note on what the category is good for. New on this list: 4-point and 6-point variants without a midpoint for agreement, importance and satisfaction. The wordings are proven examples from practice, not validated psychometric scales. Test in a pretest whether your target group understands them the way you mean them.

Quick navigation:

• Agreement
• Importance
• Probability
• Satisfaction
• Attitude
• Preference
• Frequency
• Quantity
• Intentions
• Usefulness
• Credibility
• Difficulty
• Effectiveness, competence, speed
• Self-Comparison

Agreement Likert scales

The classic Likert scale. You present a statement, such as "Onboarding was well organized.", and ask how much respondents agree with it.

Importance Likert scales

Use these to find out what really matters to respondents, for example which service actually counts when they choose a provider.

Probability and likelihood Likert scales

For judging whether something will happen. For example: "How likely is it that you'll change employers in the next year?"

Satisfaction Likert scales

Satisfaction asks about something respondents have experienced themselves, such as a purchase or their own workplace.

Attitude Likert scales

These are about how good someone thinks something is, a brand for example, even if they've never used it themselves.

Preference Likert scales

With these, respondents say how much they like something or prefer it to an alternative, for example one working-hours model over another.

Frequency Likert scales

These fit questions about behavior, such as "How often do you use the intranet?". If you need precise figures, use the 7-point variant with concrete time periods instead of words like "often" or "rarely".

Quantity Likert scales

For anything that can be too much or too little, and for comparisons, for example with the past or with expectations.

Intention Likert scales

For planned behavior, for example whether someone will buy a product or recommend you.

Usefulness Likert scales

Use this to ask whether a training session or a feature actually helps respondents.

Credibility Likert scales

For questions about how much respondents believe a statement or an ad.

Difficulty Likert scales

For tasks and processes, such as "How easy was it to sign up?"

Effectiveness, competence and speed Likert scales

For judging how effective or how fast a process or a tool is, and for the development level of skills, for example in a performance review.

Self-comparison Likert scales

Respondents say how well a statement describes them. This is typical for questions about personality or motivation.

5 or 7 points? A quick decision

There's no number of points that's right for every survey. What matters is who answers, how you ask and what you'll do with the data later. The table sums it up.

With a midpoint (5 or 7 points) is the default, because a genuine neutral position is a valid answer. Without a midpoint (4 or 6 points) can make sense when neutrality hardly occurs for the topic. But don't remove the midpoint just to force a decision. "Don't know" and "not applicable" are separate answers, not a midpoint: offer them separately when they can occur.

Research doesn't offer a universal rule either. Simulation studies tend to favor at least seven categories, while more recent empirical studies point to four or five (Schnell 2019, p. 80). For scales where every point is labeled, the sweet spot is four to seven points. Beyond that it's hard to find words that are still clearly distinct (Thielsch, Lenzner & Melles 2012, p. 233). My advice: when in doubt, use the 5-point scale. Malhotra & Birks (2007, p. 351) give five to nine categories as the traditional guideline, but stress that there's no single optimal number.

By the way, if you repeat a survey regularly, stick with the number of points you chose the first time. Answers to a 5-point and a 7-point version of the same question can only be compared under strong assumptions (Schnell 2019, p. 80).

You'll find the reasoning behind each row of the table in Likert scales: everything you ever wanted to know, but never dared to ask, in the sections on the number of response categories and on an even or odd number of response options.

Analyzing Likert scale data: the first steps

The answers are in. Now what? For a single Likert question, three steps are usually enough: look at the frequencies, find the median, and report the shares at the very top and the very bottom of the scale. You can add the mean, but never on its own.

Let's look at an example. A bakery surveys 20 customers in each of its two stores: "How satisfied are you with the service at the counter?" The answers use the 5-point satisfaction scale from this list and are coded from 1 (Extremely dissatisfied) to 5 (Extremely satisfied).

Step 1: Frequencies. Count how often each point was chosen and convert the counts to percentages. For store A it looks like this:

With only 20 answers, each person stands for 5 percentage points. If a single customer answers differently, a share shifts by 5 percentage points.

Step 2: Median. The median is the answer in the middle when you line up all answers in order. With 20 answers, the middle lies between the 10th and the 11th answer. The "Cumulative" column shows that up to and including "Neutral" you have 45%, which is 9 answers. So the 10th and the 11th answers are both "Somewhat satisfied", and the median is 4. It's a real point on the scale, and you can report it in words: at least half of the customers are somewhat or extremely satisfied.

Step 3: Top-2-box and bottom-2-box. The top-2-box is the share of the two highest points combined (Linke 2018, p. 105). In store A that's 7 + 4 = 11 of 20 answers, or 55%. Any store manager understands this number without knowing statistics. The bottom-2-box is the mirror image: the share of the two lowest points, here 5 of 20, or 25%.

And the mean? Let's calculate it: (2 × 1 + 3 × 2 + 4 × 3 + 7 × 4 + 4 × 5) / 20 = 68 / 20 = 3.4. Now add store B, also with 20 answers:

Both stores have the same mean. Yet they're two completely different situations. In store A, most customers are reasonably satisfied. In store B there are two camps (Linke 2018, p. 105): more than half are extremely satisfied, and just over a third are extremely dissatisfied. Maybe it comes down to a particular shift or a particular person at the counter. In store B, the mean of 3.4 describes an answer that almost nobody there gave.

Now look at the median and the top-2-box again. Both make store B look even better than store A: a median of 5 instead of 4, a top-2-box of 60% instead of 55%. You only see the 40% who are dissatisfied in the bottom-2-box and in the frequency table. That's why the distribution always belongs in your report.

With 20 responses per store, 5 percentage points equal a single person. The difference between 55 and 60 percent proves nothing on its own.

There's also a more fundamental point. The numbers 1 to 5 are just codes for ordered answers. Whether the distance between "Neutral" and "Somewhat satisfied" is as large as the distance between "Somewhat satisfied" and "Extremely satisfied" can't be read from the answers alone. Respondents even draw the boundaries between the points differently (Schnell 2019, pp. 77-78). In practice, people report Likert means all the time anyway, and for comparing groups they're useful. My advice: never report the mean on its own, always together with the distribution or with the top-2-box and bottom-2-box.

Several questions on one topic? If you measure job satisfaction with several statements, for example, you can analyze each statement separately or calculate an overall score per person. Before you do, reverse-worded statements such as "The service is poor" have to be recoded so that a high value means the same thing everywhere. How to build the overall score and when more effort pays off is shown in Aggregating multi-item (Likert) scales. For the next steps of an analysis, from charts to statistical tests, see Analyzing online surveys.

By the way, in QUESTIONSTAR the analysis shows you, for every question, the count, percentage and cumulative percentage per answer option, plus the median, minimum, maximum, variance and standard deviation. The top-2-box is then simply the sum of the two highest percentages.

Standard wordings used in market research

Many questionnaires in market research use a small set of standard wordings. The scales on this list are close to them, but not always identical. If you want to compare your results with a published study or a textbook example, that difference matters. The middle column shows the standard wordings. You'll find the agreement scale in Malhotra & Birks (2007, pp. 348-349), for example, and all six rows in my slide deck "Principles of Survey Research" (Marx 2026, slide 51). The right column shows the closest scale from this list.

Similar is not the same. Look at the frequency row. The standard scale ends with "very often", the scale on this list ends with "always". "Always" is an extreme pole, "very often" isn't. Anyone who treats the two versions as equal is comparing two different scales. The satisfaction row shows a similar step: "extremely satisfied" is a stronger pole than "very satisfied", and extreme poles usually lead to answers that cluster more toward the middle (more on this in peaked vs. flat response distribution). The midpoints differ too: "neutral" on one side, "neither agree nor disagree" on the other. None of these versions is wrong. But if you compare your results with a study, use exactly the wording that study used. For a stand-alone survey, choose wording that fits the construct and check it with your target respondents.

Using a scale in another language? Then these differences grow. Respondents understand vague words like "often" or "rarely" differently, and the boundaries between points can differ between groups, including between nationalities (Schnell 2019, p. 77). For ratings and attitudes, cultural differences play a particularly large role (Möhring & Schlütz 2019, p. 194). A small example: translated word for word into German, "strongly agree" becomes "stimme stark zu". German questionnaires usually say "stimme voll und ganz zu" (agree completely) instead, and that tends to sound stronger than the English original. So if you want to compare results across languages, have the translated scales checked by native speakers. A back-translation also helps: someone translates the translated version back into the original language, and you compare the two versions for ambiguities (Möhring & Schlütz 2019, p. 195). Even that doesn't guarantee that both versions measure the same thing, though.

Frequently asked questions about Likert scales

What does a Likert scale question look like?

A Likert scale question combines a statement or a question with one of the scales from this list. A few examples: "Onboarding was well organized." with the Agreement 5 Point Scale. "How satisfied are you with the service at the counter?" with the Satisfaction 5 Point Scale. "How often do you use the intranet?" with one of the frequency scales. "How likely is it that you'll change employers in the next year?" with the Probability 5 Point Scale. The important thing is that the scale fits the question: don't answer a frequency question with agreement labels.

How many points should a Likert scale have?

Usually 5 or 7. Five points are the safer choice when respondents know the topic only superficially, when the questions are read aloud or when many people answer on a smartphone. Seven points pay off with knowledgeable respondents who can reliably tell fine gradations apart. You'll find a decision aid above in the section "5 or 7 points?".

With or without a midpoint?

With a midpoint (5 or 7 points) is the default, because a genuine neutral position is a valid answer. Without a midpoint (4 or 6 points) can make sense when neutrality hardly occurs for the topic. But don't remove the midpoint just to force a decision. "Don't know" and "not applicable" are separate answers, not a midpoint: offer them separately when they can occur. Scales with an even number of points have no midpoint. The new variants for agreement, importance and satisfaction are marked "No Midpoint". Also without a midpoint on this list are the 4-point scales for probability and for quantity (Comparison to Ideal). The 4-point scales for frequency and quantity start at "Never" or "None", so they have no neutral middle anyway. More detail: Even or odd number of response options.

Can you calculate the mean of Likert scale data?

You can, and in practice it's common. Strictly speaking, though, the answers are only ordered, and equal distances between the points aren't guaranteed. So always report the mean together with the distribution or with the top-2-box and bottom-2-box. The example with the two stores in the section "Analyzing Likert scale data" shows why.

How do you label a 7-point Likert scale?

The simplest way is symmetrical: two clearly named poles, a clearly neutral midpoint and steps in between that are clearly distinct from each other. The 7-point agreement scale on this list follows exactly this pattern: Strongly disagree, Disagree, Somewhat disagree, Neutral, Somewhat agree, Agree, Strongly agree. Labeling every point often feels clearer, but with self-explanatory scales it can also add visual clutter. Research shows no consistent, clear difference in answers between labeling all points, only the poles and the midpoint, or only the poles. What matters is that the poles are clearly labeled and the steps look evenly spaced. With seven points, the midpoint should be clearly named as well as the two poles. And if you label every point, make sure no two labels mean almost the same thing. When space is tight, for example in a matrix or with a slider, the poles and the midpoint are enough. More on this in Labeling the scale points.

Are these scales available as templates?

Yes. In QUESTIONSTAR, the scales on this list are stored in the scale library. With the Magic wand, you add a scale to your question in one click. For whole questionnaires, there are also templates by use case. With any template, check whether the questions and scales fit your target group.

Where the wordings come from

The scales this list started with are ones we've kept running into in practice since 2002. In this revision we fixed a few of these rows: two scale names gave the wrong number of points, and we renamed the category Efficiency to Effectiveness, because its scales measure how well and how fast something works, not how economically.

Of the six new variants without a midpoint, three are based on the collection "Antwortskalen im ALLBUS und ISSP" (Response scales in ALLBUS and ISSP) published by ZUMA in Mannheim. It gathers the response scales used up to 2003 in two large social-science surveys. The originals are German, and the English labels on this list are our renderings. From there come the 4-point agreement scale (p. 8), the 4-point importance scale (p. 11) and the 6-point satisfaction scale (p. 38). We built the other three ourselves. The 6-point agreement and the 4-point satisfaction scales are the 7-point and 5-point scales from this list without their midpoint. The 6-point importance scale follows the pattern of the 6-point satisfaction scale: very, fairly, somewhat.

By the way, there's a classic empirical study behind many German response scales. In 1978, Rohrmann tested which labels different respondents understand as uniformly as possible. Standard sets such as "nie, selten, gelegentlich, oft, immer" (never, rarely, occasionally, often, always) for frequencies and "keinesfalls, wahrscheinlich nicht, vielleicht, ziemlich wahrscheinlich, ganz sicher" (definitely not, probably not, maybe, quite likely, definitely) for likelihoods come from this work (cited in Schnell 2019, p. 79). You'll find similar patterns in several rows of this list, for example in the frequency and intention scales.

Sources:

• Linke, R. (2018): Mitarbeiterbefragungen optimieren. Von der Befragung zum wirksamen Management-Instrument [Optimizing employee surveys]. p. 105.
• Malhotra, N. K. & Birks, D. F. (2007): Marketing Research: An Applied Approach, 3rd European edition. pp. 348-349 and 351.
• Marx, P.: Principles of Survey Research (online course).
• Möhring, W. & Schlütz, D. (2019): Die Befragung in der Medien- und Kommunikationswissenschaft [Surveys in media and communication research], 3rd edition. pp. 194-195.
• Prüfer, P., Vazansky, L. & Wystup, D. (2003): Antwortskalen im ALLBUS und ISSP. Eine Sammlung [Response scales in ALLBUS and ISSP: a collection]. ZUMA-Methodenbericht 2003/11. pp. 8, 11 and 38.
• Rohrmann, B. (1978): Empirische Studien zur Entwicklung von Antwortskalen für die sozialwissenschaftliche Forschung [Empirical studies on developing response scales for social-science research]. Zeitschrift für Sozialpsychologie, 9, pp. 222-245. Cited in Schnell (2019), p. 79, and in Thielsch, Lenzner & Melles (2012), p. 233.
• Schnell, R. (2019): Survey-Interviews. Methoden standardisierter Befragungen [Survey interviews: methods of standardized surveys], 2nd edition. pp. 76-80.
• Thielsch, M. T., Lenzner, T. & Melles, T. (2012): Wie gestalte ich gute Items und Interviewfragen? [How do I design good items and interview questions?] In: Thielsch, M. T. & Brandenburg, T. (eds.): Praxis der Wirtschaftspsychologie II. p. 233.

Date: April 9, 2025
Author: Dr. Paul Marx
Updated: November 3, 2026

You might also be interested in:

• Likert scales: everything you ever wanted to know, but never dared to ask
• Aggregating multi-item (Likert) scales