Likert Scale Examples for 5, 7 and 10-Point Questions

A row of “strongly agree to strongly disagree” is the easy part of a Likert scale. How many points to use, how to word them so the answers mean anything, and what to do with the numbers afterwards are the parts that decide whether your survey tells you the truth.

So here are ready-to-use Likert scale examples at 5, 7 and 10 points, grouped so you can copy the format that fits, plus the research on how many points is actually best, the design rules that keep the answers honest, and how to read the results without kidding yourself.

What a Likert scale actually is

A Likert scale asks a respondent to rate how strongly they agree with a statement, on a fixed set of ordered points. The classic version has five: strongly disagree, disagree, neither agree nor disagree, agree, strongly agree. The psychologist Rensis Likert introduced it in his 1932 paper “A Technique for the Measurement of Attitudes” as a faster, simpler alternative to the fiddly attitude-scaling methods of the day. The five-point agreement format is the one that stuck.

One point of language, because it trips people up. A single question is a Likert item. A Likert scale, strictly, is several items combined into one score. In everyday use most people call a single agreement question a Likert scale, and I will too, but the distinction matters when you get to analysis.

It matters for a practical reason, not a pedantic one. A single question is noisy: one odd interpretation, one distracted respondent, one badly chosen word, and the answer wobbles. Likert’s original method summed several items precisely so those wobbles average out. So every single-question example below is a Likert item, and if you are measuring something that genuinely matters, ask three or four related items and combine them rather than resting a decision on one line.

It is also worth separating a Likert scale from rating scale questions in general. Rating scale questions are any question answered on an ordered set of points: star ratings, zero-to-ten satisfaction, frequency scales from never to always. A Likert scale is the specific kind that measures agreement or a similar attitude, with balanced points either side of a neutral middle and a word on every point. All Likert scales are rating scales. Not all rating scales are Likert scales.

5-point Likert scale examples

Five points is the natural starting point and the format Likert built. Two negative options, one neutral, two positive. It is quick to answer, reads cleanly on a phone, and is almost always the right choice for a checkout survey or a post-purchase email.

The standard agreement anchors:

Strongly disagree · Disagree · Neither agree nor disagree · Agree · Strongly agree

Statements to attach to them, written for an ecommerce store:

• The product page gave me the information I needed to decide.
• This product is worth what I paid for it.
• Checkout was quick and easy.
• The delivery arrived when I expected it to.
• I would buy from this store again.

Agreement is not the only 5-point scale you can run. Match the anchor to what you are actually measuring, because a mismatched anchor is where most bad rating scale questions go wrong.

Satisfaction: Very dissatisfied · Dissatisfied · Neither satisfied nor dissatisfied · Satisfied · Very satisfied

Likelihood: Very unlikely · Unlikely · Neither likely nor unlikely · Likely · Very likely

Frequency: Never · Rarely · Sometimes · Often · Always

Importance: Not at all important · Slightly important · Moderately important · Very important · Extremely important

There are more formats to borrow from in our survey questions examples. Keep one scale per survey where you can. If you ask ten agreement questions and then switch to a frequency scale halfway down, people answer the new one on autopilot with the old scale still in their head.

7-point Likert scale examples

Seven points give respondents more room to place themselves, which matters when you are comparing two groups and the difference between them is small. The extra options sit between the middle and the extremes.

Strongly disagree · Disagree · Somewhat disagree · Neither agree nor disagree · Somewhat agree · Agree · Strongly agree

The same statements work; the scale just resolves finer. A 7-point scale suits attitude and brand questions, where you genuinely want to tell “somewhat agree” apart from “agree”:

• This brand feels like it was made for someone like me.
• The price feels fair for what you get.
• I trust this brand to deliver what it promises.
• This is a brand I would recommend to a friend.

The trade-off is length and effort. Seven fully labelled options take longer to read than five, so on a long survey or a small phone screen, five usually wins. Reserve seven for the handful of questions where the extra precision earns its place.

The practical version of that argument is worth stating plainly, because the psychometrics get all the attention and the layout decides the outcome. Seven fully labelled options do not fit across a phone screen. They either wrap onto two lines, shrink to unreadable type, or stack vertically and push the next question below the fold. A scale a respondent cannot read at a glance is a scale they satisfice their way through, tapping whatever is nearest. If most of your traffic is mobile, that consideration outranks the reliability gain.

10-point and 0-to-10 scale examples

Ten-point and zero-to-ten scales are common for overall satisfaction and recommendation, and they are the format respondents tend to like best. They give a wide spread, which is useful when you want to track a number over time and see it move.

A 0-to-10 recommendation question is the Net Promoter format:

“How likely are you to recommend us to a friend or colleague, from 0 (not at all likely) to 10 (extremely likely)?”

A 0-to-10 satisfaction question works the same way, anchored at both ends:

“Overall, how satisfied are you with your order? 0 means not at all satisfied, 10 means completely satisfied.”

The catch with wide scales is that only the end points carry words. Everyone agrees what 0 and 10 mean; one person’s 7 is another person’s 8. That is fine for a single tracked metric where you care about the trend more than the absolute number, and it is why zero-to-ten is the home of NPS and satisfaction tracking rather than detailed attitude work. For anything you need to interpret point by point, a fully labelled 5 or 7-point scale gives you cleaner data.

How many points should you actually use

This is the question everyone asks, and the research answers it more clearly than the usual “it depends”.

Preston and Colman (2000), in Acta Psychologica, had respondents rate the same service on scales ranging from 2 to 11 points. Scales with two, three and four points performed poorly on reliability, validity and the ability to discriminate between respondents. The scores improved as points were added, up to about seven, and then flattened. Test-retest reliability actually started to fall once scales went beyond ten points. When they asked people which scale they preferred to use, the 10-point came top, closely followed by the 7 and 9-point.

So there is a genuine tension: seven points is around where the psychometric quality peaks, but people find ten the most comfortable to fill in.

The other reassuring finding is that the choice matters less than you fear for the headline number. Dawes (2008), in the International Journal of Market Research, ran the same questions on 5, 7 and 10-point scales and found that once the answers were rescaled to a common range, the mean scores were very similar across all three. The wider scales spread the responses out more, which captures a little more nuance, but they did not systematically push the average up or down. You are not going to get a wildly different answer because you picked five instead of seven.

Here is how I decide:

The one rule underneath all of this: pick a length, then use it consistently across the survey and across time. Half your value from a tracked metric comes from measuring it the same way every quarter.

The rules that keep the answers honest

The number of points is the easy part. Wording is where surveys quietly break.

Label every point, not just the ends. When only the extremes carry words, people interpret the unlabelled middle points differently, which adds noise you cannot see. Fully labelled scales produce more reliable data, which is exactly why the 5 and 7-point agreement scales above put a word on all of them. On a 0-to-10 scale you cannot label every point, and that is the accepted cost of using one.

Keep the scale balanced. Equal numbers of positive and negative options, with the neutral point genuinely in the middle. A scale that runs “satisfied, very satisfied, extremely satisfied” with one negative option will flatter you and tell you nothing.

Decide the neutral midpoint on purpose. An odd number of points gives you a middle option, and there is a long argument about whether to keep it. Remove it and you force an opinion out of people who genuinely do not have one, which adds error. Keep it and a few respondents will hide there to avoid thinking. Chyung and colleagues (2017), reviewing the evidence, advise keeping the midpoint when a neutral answer is real and meaningful, and only considering its removal when you have a specific reason to think people are using it as an escape hatch. For most commercial surveys, keep it.

Mind acquiescence. People lean towards agreeing, regardless of the statement, a bias that inflates every agree-disagree question. The classic fix is to reverse-word some items, so agreement points the other way. Except it often does not work. A PLOS One study (2013) found reverse-worded items suffered the same acquiescence and mostly confused respondents, correlating with the items they were meant to counterbalance rather than against them. The better defences are plain statements, a neutral tone, and not stacking a survey entirely out of agree-disagree items when a direct question would do.

One idea per statement. “Checkout was fast and the delivery was on time” cannot be answered by someone whose checkout was fast but whose parcel was late. Split it.

How to read Likert results without fooling yourself

You have the answers back. Resist the urge to average them and move on.

The honest starting point: Likert data is ordinal. The points are in order, but the gaps between them are not guaranteed to be equal. The distance from “strongly disagree” to “disagree” is not provably the same as the distance from “agree” to “strongly agree”, so the moment you average them you are assuming something the scale never promised. Jamieson (2004), in Medical Education, made the point sharply: the mean of a single Likert item is not strictly legitimate, because the average of “fair” and “good” is not “fair and a half”.

In practice, three readings do the job and none of them require you to defend a contested average:

Report the distribution. What share landed on each point. A question with everyone clustered on “agree” and one with a fifty-fifty split at the extremes can share a mean and mean opposite things.

Use top-two-box. The percentage who chose the two most positive points, so agree plus strongly agree. It is easy to explain to anyone in the business, hard to misread, and it is the standard way operators track these questions. Watch bottom-two-box, the two most negative points, with equal attention.

Report the median if you want a single midpoint figure, because the median is legitimate for ordinal data where the mean is shaky.

A worked example. Two hundred customers rate “This product is worth what I paid for it” on a 5-point scale. Forty say strongly agree, ninety say agree, forty neutral, twenty disagree, ten strongly disagree. Top-two-box is (40 + 90) / 200, which is 65 per cent. Bottom-two-box is 15 per cent. The median lands on “agree”. That is a clear, defensible read of a broadly positive but not universal result, and you never had to pretend the intervals were equal.

When a rating scale is the wrong tool

A Likert scale measures what people say about something in front of them. That is exactly right for satisfaction, for brand attitude, for feedback on a product they already own. It is far weaker for a buying decision that has not happened yet. “I would buy this” on a five-point scale is a stated intention, and the gap between stated intention and actual purchase is wide and well documented, which is why plenty of products that surveyed beautifully still launched flat.

So if you are about to spend real money on a new product, a pack, an ad, a claim or a price, an agreement scale will tell you how people feel about the idea, not whether they will part with cash for it.

There is a better instrument for that specific job, and almost nobody writing about Likert scales mentions it. The Juster purchase probability scale asks for a probability rather than an agreement: an 11-point scale running from 0, “no chance, almost no chance”, to 10, “certain, practically certain”, with a stated likelihood attached to every point. Because respondents answer in probabilities rather than sentiment, it predicts actual purchase considerably better than an agree-disagree item does. If the question you are really asking is “will they buy this”, reach for Juster rather than bending a Likert scale to a job it was not built for. Combine the rating with an open-ended “why”, and if you can, put the decision in a context closer to a real purchase. If you want a habit of listening to customers at scale, it is worth comparing the voice-of-customer tools built for reading the reasons behind the ratings, because the verbatims are usually where the value sits.

For the pre-launch case specifically, the honest gap is that there is nobody to survey yet. That is what TestFeed is built for. You put a product, a pack, an in-context price, an ad or a concept in front of shoppers modelled on your target customer and get back a purchase-intent read, their reasons in their own words, and a clear next move, in days rather than weeks. Treat it as directional pre-spend signal to decide what deserves a real launch, not a sales forecast or a guaranteed number, and it does not judge taste, texture or smell. If you want to see how that differs from a survey panel, the comparison with a purpose-built test audience is the thing to read next. Then point a Likert survey at the customers a launch brings you, and track the score over time.

Frequently asked questions

What is a Likert scale?

A Likert scale is a survey question that asks people to rate how strongly they agree or disagree with a statement, on a fixed set of ordered points. The classic version has five: strongly disagree, disagree, neither agree nor disagree, agree, strongly agree. Rensis Likert introduced it in 1932 as a simpler way to measure attitudes. A single question is technically a Likert item; a set of items combined into one score is the Likert scale.

How many points should a Likert scale have?

Five or seven for most surveys. Preston and Colman (2000) found that scales with two, three or four points performed poorly on reliability and validity, that scores improved up to about seven points, and that gains beyond seven were marginal. Use five points for short or mobile surveys where speed matters, and seven when you need to detect smaller differences between groups. Ten-point and zero-to-ten scales are common for satisfaction and recommendation and are the format respondents tend to prefer.

Should a Likert scale have a neutral middle option?

Usually yes. Removing the midpoint forces an opinion from people who genuinely do not have one, which adds error rather than removing it. Chyung and colleagues (2017) advise including a midpoint when a neutral position is a real, meaningful answer, and considering its removal only when you have specific reason to believe respondents are hiding behind it. Keeping the midpoint and reading the results carefully is the safer default.

Can you take an average of Likert scale answers?

Cautiously, and it is contested. Likert data is ordinal: the points are ranked, but the distance between strongly disagree and disagree is not guaranteed to equal the distance between agree and strongly agree, so an average assumes something the scale does not give you. Jamieson (2004) argued the mean is not strictly legitimate for a single Likert item. In practice, report the median and the distribution, and use top-two-box, the share choosing the two most positive points, which is easy to explain and hard to misread.

What is the difference between a Likert scale and a rating scale?

A Likert scale is one type of rating scale. Rating scale questions are any question that asks for a rating on an ordered set of points, including star ratings, zero-to-ten scales and frequency scales. A Likert scale specifically measures agreement or a similar attitude, with balanced points either side of a neutral middle and a verbal label on each point. All Likert scales are rating scales; not all rating scales are Likert scales.

The next step is not to add more questions. Pick your length, five for speed or seven for precision, write a word on every point, keep one idea per statement, and decide up front that you will read the results as top-two-box and a distribution rather than a single average. Do that and the scale will tell you something you can act on.