5 point Likert scale examples, ready to paste
The five labels in order, the eight answer rows that cover almost every rating question we publish, and 35 statements lifted from a library of 1,422 questions. Plus the two decisions, how many points and which end labels, that move a result further than the wording of the statement does.
A five-point Likert scale puts one statement in front of a respondent and offers five ordered answers. The standard row reads Strongly disagree, Disagree, Neither agree nor disagree, Agree, Strongly agree, coded 1 to 5. It is the most common answer row in survey work by a distance: 443 of the 644 rating items in our own published template library carry exactly those two end labels, 68.8 per cent of them.
One statement with that row underneath it is a Likert item. The scale is the set of statements you score together. Almost everything that goes wrong with this format comes from confusing the two.
SuperSurvey Question Corpus, 109 live templates and 1,422 questions, read 11 September 2026.
Copy and paste a 5 point Likert scale
Eight answer rows, each with the number of live rating items behind it.
Most of the work sits in the answer row rather than the statement. Choose one row, keep it identical down the whole block, and every later decision gets easier: the coding, the chart, the comparison against last quarter. Change it halfway and none of those work.
Eight 5 point Likert scale answer rows
- Strongly disagreeDisagreeNeither agree nor disagreeAgreeStrongly agree
- Very dissatisfiedDissatisfiedNeither satisfied nor dissatisfiedSatisfiedVery satisfied
- Almost neverRarelySometimesOftenAlmost always
- Very unlikelyUnlikelyNeither likely nor unlikelyLikelyVery likely
- Not at all confidentSlightly confidentModerately confidentVery confidentCompletely confident
- Very difficultDifficultNeither easy nor difficultEasyVery easy
- Not at all trueSlightly trueModerately trueMostly trueCompletely true
- Not at all importantSlightly importantModerately importantVery importantExtremely important
| Construct | End labels | Items | Use it for |
|---|---|---|---|
| Agreement | Strongly disagree to Strongly agree | 443 | Attitude statements. The default, and the one to reach for unless the construct argues otherwise |
| Satisfaction | Very dissatisfied to Very satisfied | 64 | Service and product experience, where the judgement is already an evaluation |
| Frequency | Almost never to Almost always | 10 | Behaviour, where asking someone to agree with a behaviour is a wasted step |
| Likelihood | Very unlikely to Very likely | 6 | Stated intent. Intent is not action, so treat the number as a direction |
| Confidence | Not at all confident to Completely confident | 7 | Capability after training, and forecasts |
| Ease | Very difficult to Very easy | 9 | Effort in a process or an interface |
| Truth of a description | Not at all true to Completely true | 12 | Self-description, where agree and disagree read oddly |
| Importance | Not at all important to Extremely important | 0 | Ranking priorities, although the alternative below is usually better |
Every row in the copy block above, with what our library uses it for. Source: SuperSurvey Question Corpus, 109 live templates and 1,422 questions, read 11 September 2026. Base: 644 rating items. The library records the two end labels for every rating item, so the middle labels in the copy block are the conventional fill and are not part of the count. Importance shows zero because no live template uses it.
Share of rating items using each end-anchor pair
Two things follow from that shape. The agreement row is so familiar that readers stop reading it, which is an advantage on a long block and a liability on a single question. And the moment your construct is genuinely not agreement (how often something happens, how hard something was, how sure someone is), swapping the row costs nothing and removes a translation step for the respondent. Numeric and star formats are a separate family with their own wording rules, so compare rating-scale wording against the worded rows before you settle on either.
If none of the eight fits your construct, Marquette University publishes a longer library of sample label sets covering frequency, quality and importance wordings.6
"Not applicable" and "Do not know" are not points on the scale. Put them after the five, mark them visually apart, and keep them out of the scored denominator: the average and the top-two box are computed over the people who answered on the row, and the Not applicable count is reported beside them as a figure of its own. Fold them in and a question that half your audience cannot answer will quietly read as lukewarm agreement.
3, 4, 5 and 7 point Likert scales compared
What our own library uses, and why that is not the same as what you should use.
Here is an awkward number. Of the 644 rating items in our template library, 637 offer five labelled points and seven offer eleven numbered options, 0 to 10. Not one uses three, four, six or seven. The seven 0-10 items are all the same question, how likely are you to recommend, which is a single-item metric borrowed from customer research rather than an attitude battery. Its numerical midpoint is 5, and that is only the middle number: nothing on a 0-10 row tells the respondent that 5 means a neutral view.
That is a house convention, not a finding about respondents. We standardised on five because a five-point row fits a phone screen, because top-two box reporting is immediately legible to someone who has never seen a survey report, and because a consistent row lets us compare items across templates. None of those reasons is about measurement quality. So the honest position is that our library cannot tell you whether seven points would measure your construct better, and this page will not pretend otherwise.
Response options per item across 644 live rating items
| Options | Labelled midpoint | What you gain | What it costs | In our library |
|---|---|---|---|---|
| 2 | None | The fastest possible answer | Strength of feeling disappears. This is a yes/no wearing a scale costume | 0 |
| 3 | Yes | Fits a narrow phone column; usable for children and low-literacy audiences | Most of the discrimination. Everyone who mildly agrees and everyone who strongly agrees land in the same box | 0 |
| 4 | None | Forces a direction, which suits audit and compliance items | People who genuinely have no view have nowhere honest to go, so item non-response rises | 0 |
| 5 | Yes | The default. Legible top-two and bottom-two reporting, one screen line on a phone | Fine gradations. "Slightly agree" has nowhere to sit | 637 |
| 6 | None | A forced direction with more grades than four | A longer row to read for a distinction most respondents will not make | 0 |
| 7 | Yes | More spread, if your audience can tell "slightly" from "somewhat" | Speed, and horizontal room. Seven labelled points do not fit a phone in one line | 0 |
| 0-10, eleven options | None. The numerical midpoint is 5, with no label attached to it | A familiar recommend metric with an established benchmark | It is one question, not a battery, and it cannot be averaged with agreement items | 7 |
What each length buys and what it costs. Source: SuperSurvey Question Corpus, 109 live templates and 1,422 questions, read 11 September 2026. Base: 644 rating items across 109 templates. The 0-10 row has eleven response options, 0 through 10, as the live recommend widget renders it. Gains and costs are editorial judgements, not counts.
2, 3, 4 and 7 point Likert scale rows
2 points, agreement:
DisagreeAgree3 points, agreement:
DisagreeNeutralAgree3 points, frequency:
NeverSometimesOften3 points, satisfaction:
DissatisfiedNeutralSatisfied4 points, agreement:
Strongly disagreeDisagreeAgreeStrongly agree4 points, frequency:
NeverRarelyOftenAlways7 points, agreement:
Strongly disagreeDisagreeSlightly disagreeNeither agree nor disagreeSlightly agreeAgreeStrongly agree
None of these seven rows appears anywhere in our template library. They are here because people ask for them, and because a row you have to invent under time pressure is a row that ends up inconsistent.
When a 3 point Likert scale is the right call
Three points earn their place when reading is the bottleneck rather than measurement. Surveys for children, for audiences answering in a second language, and grids that have to survive a 360px screen all work better at three. The price is real: a three-point row cannot separate a mild opinion from a strong one, so a move from a lukewarm Agree to an emphatic one is invisible, and you lose the top-two box as a summary because the top box is the whole positive side.
Use three points for the whole block or none of it. A survey that runs three points on question four and five points on question five is unanalysable across the two, and respondents slow down at the switch.
4 point Likert scale examples: dropping the middle
A four-point row is the five-point agreement row with the middle removed: Strongly disagree, Disagree, Agree, Strongly agree. It is common in compliance and audit work, where the organisation has decided that "no view" is not an acceptable answer about a rule that applies to everyone.
Removing the midpoint does not remove ambivalence, it relocates it. Some of the people who would have chosen the middle pick a side at random; some abandon the question. If you take the middle away, add an explicit escape route beside the row, "Not applicable" or "Do not know", so the people who really cannot answer are counted as such rather than guessing.
When 7 point Likert scales are worth it
Seven points are defensible when three conditions hold together: your respondents are practised at self-report, the construct genuinely varies finely, and you are modelling the result rather than reporting a percentage to a stakeholder. Academic instruments live at seven for exactly this reason. The review of Likert scale development between 1995 and 2019 by Jebb and colleagues treats the number of points as one design decision among several, and puts far more weight on how well the items cover the construct than on the length of the row.2
On a phone, seven labelled points wrap. If you use seven, label the ends and the middle only, and check the render at 360px before you send it.
Likert scale examples by survey type
Thirty-five statements, all of them live in a published template today.
Every statement below is running in a real template, which means each one has already survived an editing pass and a preview on a phone. Copy a block whole, then cut it back to the items that feed a decision you are actually going to make. A block of ten that changes something beats a block of thirty that gets read once.
All four blocks use the same answer row: Strongly disagree, Disagree, Neither agree nor disagree, Agree, Strongly agree. Each copy button pastes the heading, the numbered statements, that row and a short analysis note, so the block stands on its own in a document. These are editable examples, not a validated scale: no norms, no published reliability, and nothing stopping you from rewording an item to fit your case.
Likert scale examples for employee engagement
Ten items, from three engagement templates
I know what my team needs from me before the week is out.
My workload is manageable.
My manager gives me feedback that helps me improve.
When I do good work, it gets noticed.
I can raise a concern or a risk without it counting against me.
I can see a path to develop my skills or career here.
Decisions that change my work are explained to me.
I get the training I need to do my job well.
Most days, I am glad to be doing this job.
I see myself still working here in two years.
Answer row, the same under every statement
Analysis note. Editable examples, not a validated scale. Code Strongly disagree as 1 through Strongly agree as 5, the same direction on every item. No item is reverse-worded, so no recoding is needed. Average the answered items per respondent only when at least 8 of the 10 are answered; otherwise leave that person's score blank rather than averaging fewer. Not applicable counts as unanswered for that rule and is reported as its own share, outside the scored denominator.
Taken from the employee engagement survey template, Temperature Check and Voice of Employee. Item ten is an intent item: read it as a direction, not as a forecast.
Likert scale examples for customer experience
Eight items along the customer journey
When I was looking, it was clear what you actually offered.
Buying from you was straightforward once I had decided.
The price I paid was the price I had been shown.
Getting started took about as long as I thought it would.
Since then, it has done the job I got it for.
When something went wrong, I could reach a person who could help.
The problem I had was put right without me having to chase it.
If I were making this decision again today, I would end up in the same place.
Answer row, the same under every statement
Analysis note. Editable examples, not a validated scale. Code Strongly disagree as 1 through Strongly agree as 5, the same direction on every item. No item is reverse-worded, so no recoding is needed. Average the answered items per respondent only when at least 7 of the 8 are answered; otherwise leave that person's score blank rather than averaging fewer. Not applicable counts as unanswered for that rule and is reported as its own share, outside the scored denominator.
From the customer experience survey template, in order from the first look to the decision to stay, so you can cut from either end.
Likert scale examples for training feedback
Nine items from the post-training template
What we covered applies directly to the work I do.
This session was worth the time it took.
I can name at least one thing I will do differently because of it.
I am confident I can do what this session taught, without help.
The material was explained clearly enough to follow.
The session moved at a pace I could keep up with.
There was enough time to practise, not just watch.
I know what to do first with this when I am back at work.
Overall, I would recommend this session to a colleague doing my job.
Answer row, the same under every statement
Analysis note. Editable examples, not a validated scale. Code Strongly disagree as 1 through Strongly agree as 5, the same direction on every item. No item is reverse-worded, so no recoding is needed. Average the answered items per respondent only when at least 8 of the 9 are answered; otherwise leave that person's score blank rather than averaging fewer. Not applicable counts as unanswered for that rule and is reported as its own share, outside the scored denominator.
From the post-training survey template. Items three, four and nine are about what happens after the room empties, which is the part a happy-sheet usually misses.
Likert scale examples for a product or app
Eight items from the product and app templates
It does the job I got it for.
I can rely on it to do that job every time.
This app works the way I expect it to.
It fits the way I already work.
I have had to work around something it does not do.
It has changed for the better since I started using it.
The information I needed was easy to find.
I expect to still be using it a year from now.
Answer row, the same under every statement
Analysis note. Editable examples, not a validated scale. Code Strongly disagree as 1 through Strongly agree as 5, the same direction on every item. Item 5 is reverse-worded: recode it so 5 becomes 1, 4 becomes 2 and 3 stays 3 before combining anything. Average the answered items per respondent only when at least 7 of the 8 are answered; otherwise leave that person's score blank rather than averaging fewer. Not applicable counts as unanswered for that rule and is reported as its own share, outside the scored denominator.
From the app feedback survey template and its product sibling. Item five runs in the negative direction on purpose, and the analysis note says what to do about it.
Writing the 5 point Likert scale statement
What 443 published agreement items have in common.
We read every one of the 443 agreement items in the library and measured them. The median is 11 words. The shortest is four words, the longest 19, and eight in ten sit between seven and fifteen. Not one of the 443 ends in a question mark: every single one is a statement, which is the part people get wrong most often. "Was the training useful?" cannot be answered with Strongly agree, and a respondent who meets that mismatch has to work out what you meant before they can answer at all.
149 of the 443 open with "I", 83 with "The", 26 with "When" and 19 with "My". That is the whole grammar of the format: a first-person claim about the respondent, or a flat claim about the thing being rated.
Five rules for a 5 point Likert scale statement
-
Write a statement, not a question
If the sentence would take a question mark, it is the wrong shape for this row. Turn it round: "How clear were the instructions?" becomes "The instructions were clear."
-
One idea per statement
Twenty of our 443 items contain the word "and". We read all twenty. Most join halves of a single idea: "I could hear and see clearly from where I was sitting" is one experience. A handful are genuinely two questions wearing one coat. "The communication from the recruiting team was clear and timely" is one of them: a candidate whose recruiter was clear but slow has no honest answer available, and we are fixing it.
-
Keep the subject constant down the block
Every item in a block should be about the same thing: your manager, this session, the product. Switching subject halfway is the most common reason a set of items refuses to hold together as one score.
-
Put a timeframe in when memory is doing the work
"My workload is manageable" is answerable today. "My workload in the last six months was manageable" is answerable about a period, and the two produce different numbers from the same person. Pick deliberately.
-
Name the decision the item feeds, before you keep it
If you cannot say what you would do differently at a score of 2.8 rather than 4.1, the item is costing your respondents time and buying you nothing. Cut it. The median template in our library asks 13 questions in total, and that ceiling is doing useful work.
One more thing the agreement row does to you: it invites agreement. Acquiescence is the tendency to say yes to a statement regardless of its content, and it is strongest on vague items and on long blocks late in a survey. Mixing constructs, a frequency row here and an ease row there, costs a respondent less than you would think and breaks the rhythm that produces straightlining. The rest of that family, and how to reduce agreement bias without reverse-wording half the block, is covered on its own page.
Do not reverse-word items to catch inattentive respondents unless you are prepared to handle what it does to your data. A negatively worded item inside an otherwise positive block reliably confuses a slice of respondents, and the confusion shows up as a weaker correlation with the rest of the block. If you include one, keep it simple, and reverse-code it before you combine anything.
Likert scale batteries and reliability
How many items make a score, and how to tell whether the score means anything.
One statement measures one statement. If the thing you care about is bigger than a single sentence (engagement, trust, how easy a product is to live with), you need several statements and one combined score. That is what Rensis Likert proposed in 1932: not a five-point row, which existed before him, but the practice of summing several ordered items into a single measure of an attitude.1
Our library shows how that plays out in production. Sixty-seven of the 109 templates use the agreement row at all, and those that do carry a median of seven agreement items. Reading each template in order, we counted 112 unbroken runs of items sharing that row. The median run is three items long. Thirty-five of the runs are a single item, which is a rating question rather than a scale, and the longest unbroken run is 17.
Length of 112 uninterrupted runs of agreement items
Four to eight items per construct is the working range. Below four, one badly worded item moves the whole score. Above eight, you are spending respondent attention on precision you will not use, and drop-off starts to eat the tail of the survey.
Scoring a 5 point Likert scale battery
Code each item 1 to 5 with the negative pole as 1. Reverse any item you worded the other way round, so 1 becomes 5 and 2 becomes 4, before you do anything else. Then average the items per respondent rather than summing them, so the score stays on the original 1 to 5 range.
Decide the minimum number of answered items before you field the survey, and apply it without exception: a respondent gets a score only if they answered at least four of five, or at least 80 per cent of a longer block rounded up, and otherwise the score is left blank rather than averaged from whatever they happened to answer. Not applicable counts as unanswered for that rule.
If your block comes from a published instrument, its scoring manual overrides this paragraph: use its item weights, its list of reverse-keyed items and its own missing-data rule. The example blocks on this page are editable starting points with no manual behind them, which is why each one carries the rule in its analysis note. Once the scores exist, analyse your scale responses as ordered categories first and as numbers second.
Check internal consistency before you publish the score. Cronbach's alpha is the usual check and it is a diagnostic, not a trophy: a value around 0.7 to 0.9 says the items are measuring something in common, and a value above 0.95 usually means you asked the same question several ways. If one item pulls the alpha down, read it again: it is normally measuring a different thing, not measuring badly. The 25-year review of Likert scale development by Jebb, Ng and Tay is the place to go for how much of this matters when the score has consequences.2
Report the index and the item-level breakdown together, always. The index tells a leadership team whether the number moved. The items tell them which sentence to do something about.
How to analyse 5 point Likert scale results
Distribution first, average second, and never the average alone.
Before anything clever, produce the distribution. It is the fastest way to see a polarised result, a pile at each end and a hollow middle, and it is the only summary a stakeholder can argue with usefully. Here is one item from a customer support survey, 200 replies, on the ease row.
| Answer | Count | Share | Box |
|---|---|---|---|
| Very difficult | 18 | 9% | Bottom two |
| Difficult | 32 | 16% | Bottom two |
| Neither easy nor difficult | 54 | 27% | Middle |
| Easy | 66 | 33% | Top two |
| Very easy | 30 | 15% | Top two |
| Total | 200 | 100% |
"It was easy to get the help I needed", answered on the Very difficult to Very easy row. Worked example: the five counts are illustrative rather than observed, and they sum to 200.
One item, 200 replies, drawn as a distribution
From that one table you can report everything a stakeholder needs. Top-two box is 48 per cent, bottom-two box is 25 per cent, and the net, top two minus bottom two, is plus 23 points. The mode is Easy. Twenty-seven per cent chose the middle, which on an ease question is high enough to be worth a second look at the wording. The average is 3.29, and on its own it would have told you almost none of that.
Should you average a 5 point Likert scale?
The five answers are ordered but not evenly spaced: nothing guarantees that the step from Agree to Strongly agree is the same size as the step from Neither to Agree. Guidance written for clinical trials, where the answer has consequences, recommends methods that respect that ordering (distributions, medians, ordinal models) rather than treating the codes as measurements.4 A 2025 exploration of the same question sets out the conditions under which parametric methods behave acceptably on Likert data, which is the honest counterweight: the practice is defensible in some settings and sloppy in others.5
The practical rule that survives both arguments: publish the distribution next to any average you quote, and never let an average travel on its own into a slide. If you are comparing two groups or two quarters, whether the gap is real is a separate question, and the average alone cannot settle it.
If every item comes back rated important, you have learned nothing about priority. Forced-choice methods make respondents trade one thing off against another, and a comparison of best-worst scaling against Likert rating on accommodation attributes found the two produce genuinely different orderings.3 Ask people to choose rather than to rate: multiple choice questions covers how many options to offer and how to order them.
A scale also tells you how much and never why. One good free-text follow-up after the two items you care about is worth more than four more agreement statements, and open-ended vs closed questions covers when to spend the effort.
Frequently asked questions
Who invented the Likert scale?
Rensis Likert, an American psychologist, in a 1932 monograph published in Archives of Psychology while he was at Columbia.1 He later founded the Institute for Social Research at the University of Michigan. His surname is pronounced LICK-ert, not LIKE-ert, a detail almost every survey team gets wrong, including ours for years.
Is a 1 to 5 rating a Likert scale?
Only if the points are labelled and the item is measuring an attitude. Five bare numbers with nothing written under them is a numeric rating: two respondents can read the same 4 completely differently. And the label set does not have to be agreement: satisfaction, frequency, likelihood and confidence rows are all Likert-type items, which is the broader reading and the one we use across this site. The comparison linked from the first section sets the two families side by side.
Can I use a 5 point and a 7 point scale in the same survey?
Across different sections, yes. Inside one block, never. Two consequences follow. You cannot average or combine items measured on rows of different lengths, so each block gets its own score and its own chart. And respondents visibly slow down at the switch, because the row they had learned to read has changed shape, so put a section heading between them.
How do you code a 5 point Likert scale for analysis?
1 for the negative pole through 5 for the positive one, and keep that direction identical for every item in the block. Export the label text alongside the number if your tool allows it, because a column of bare 1-to-5 codes six months later is ambiguous: somebody will eventually have to work out whether 5 meant Strongly agree or Strongly disagree, and a spreadsheet does not remember.
What is the middle of a 5 point Likert scale called?
"Neither agree nor disagree" is the standard, and it is the one to use if you want the middle to mean a balanced view. "Neutral" and "Undecided" are common substitutes and they are not synonyms: neutral suggests a settled position in the centre, undecided suggests no position yet. Pick the one that matches what you would do about it, and use the same word on every item in the block.
Can a 5 point Likert scale include Not applicable?
Add one whenever a respondent could reasonably have no basis for answering, such as an item about a manager, sent to someone who started last week. Keep it visually separate from the five, leave it out of the scored denominator, and publish its share as a separate line next to the box percentages, so a reader can see both the result among people who could answer and how many could not. If more than roughly one in ten people select it, treat that as evidence the item is wrong for that audience rather than as a data-cleaning problem.
How we counted
Every corpus figure on this page comes from one pass over the live SuperSurvey template library on 10 September 2026: 109 published templates, 1,422 questions, of which 644 are rating items carrying two labelled end anchors. Question type, point count, end anchors and position within the template were read directly from the published markup rather than typed in by hand, so the counts are reproducible.
What that base excludes: drafts, redirect rows and unpublished templates. What it cannot tell you: anything about surveys other people run. It shows what one survey vendor publishes as recommended practice, which is why every figure here is written as a share of our own library and never as a fact about surveys in general. The zeros in the point-count table are the clearest example: they say we do not ship three-point rows, and nothing at all about whether you should.
Two figures are not from the corpus and are marked where they appear. The n = 200 distribution is a worked example with illustrative counts. The academic findings are attributed to their papers in the list below.
References
- Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 22(140), 5-55.
- Jebb, A. T., Ng, V., and Tay, L. (2021). A review of key Likert scale development advances: 1995-2019. Frontiers in Psychology, 12, 637547. doi.org/10.3389/fpsyg.2021.637547
- Heo, C. Y., Kim, B., Park, K., and Back, R. M. (2022). A comparison of best-worst scaling and Likert scale methods on peer-to-peer accommodation attributes. Journal of Business Research, 148, 368-377. doi.org/10.1016/j.jbusres.2022.04.064
- Al-Jaishi, A. A., Cuerden, M. S., Luo, B., et al. (2026). Statistical analysis of Likert-based ordinal scales: a guide for clinical trialists. BMC Medical Research Methodology. doi.org/10.1186/s12874-026-02793-5
- Huh, I., and Gim, J. (2025). Exploration of Likert scale in terms of continuous variable with parametric statistical methods. BMC Medical Research Methodology, 25, 218. doi.org/10.1186/s12874-025-02668-1
- Marquette University, Division of Student Affairs. Sample Likert Scales. marquette.edu/student-affairs/assessment-likert-scales.php
- SuperSurvey Question Corpus, 109 live templates and 1,422 questions, read 11 September 2026. Question type, point count and end anchors parsed from the published template markup.
Michael Hodge: Survey methodologist and editor, SuperSurvey. Bachelor of Science (Psychology), University of Wollongong, with coursework in psychometrics and research methods. Designing surveys since 2003. About the author and how these guides are reviewed
What to read next
Put a 5 point Likert scale in front of people
The 35 statements above are already live in templates, with the answer row attached and the reporting built.
Create a survey with a Likert scaleFree to send, no card needed. Each example block above links to the template it came from, or work through the rest of the survey learning centre first.