Likert Scale in Research: Types, Examples & Analysis
Aug 24, 2026
What Is a Likert Scale in Research?
A Likert scale is a research instrument made up of several ordered response items — usually ranging from "strongly disagree" to "strongly agree" — combined into a single score to measure an attitude, opinion, or perception that can't be observed directly. Respondents pick the option that best reflects how strongly they agree, and their choices are converted into numbers for analysis. The instrument has the name of psychologist Rensis Likert, who developed it in 1932 as a more methodical technique to gage attitudes than a straightforward yes/no question could.
Likert Item vs. Likert Scale
This distinction gets flattened almost everywhere, but it matters for your methods section. A Likert item is one statement with its ordered response options — a single question rated from 1 to 5. A Likert scale, strictly speaking, is a set of related items whose values are added together or averaged to produce a single composite measure of an underlying construct, such as "attitude toward online learning."
In practice, many researchers use "Likert scale" loosely to describe even a single rated item — common usage, not an error to panic about. But in a thesis or paper, be precise about whether you're analyzing one Likert-type item or a multi-item scale, since that choice affects which statistics are appropriate later.
How Is a Likert Scale Used in Research?
Likert scales let researchers turn subjective, hard-to-observe constructs into structured, comparable data. In academic work, they typically measure attitudes and beliefs (e.g., toward a teaching method), perceptions (e.g., perceived workload or usability), agreement with a statement, satisfaction with a course or intervention, frequency or likelihood of a behavior, and the perceived importance of a factor.
Because responses are structured and ordered, they can be summarized statistically and compared across groups, time points, or conditions — which is why Likert-based instruments are common across education, psychology, health sciences, and the social sciences.
Types of Likert Scales
The majority of Likert scales vary in two ways: the number of response points and whether the scale is bipolar (two opposing poles around a neutral center, such as "strongly disagree" to "strongly agree") or unipolar (intensity of a single attribute, such as "not at all important" to "extremely important"). Agreement, satisfaction, frequency, and likelihood scales are variations on this same structure, applied to different wording.
4-Point vs. 5-Point vs. 7-Point Likert Scale
4-Point Likert Scale
Typical points:
4 (even, no midpoint)
Example response options:
Strongly disagree, Disagree, Agree, Strongly agree
Best use:
When you want to avoid a "fence-sitting" neutral
response and force a directional answer.
5-Point Likert Scale
Typical points:
5 (odd, with midpoint)
Example response options:
Strongly disagree → Neutral → Strongly agree
Best use:
The most common default; balances simplicity
with a genuine neutral option.
7-Point Likert Scale
Typical points:
7 (odd, with midpoint)
Example response options:
Strongly disagree → Neutral → Strongly agree,
with finer gradations
Best use:
When you need more sensitivity to detect small differences between respondents or groups.
5-Point Likert Scale: Example
Research statement: "This course's teaching strategies help me grasp the material better."
Response options:
Strongly disagree
Disagree
Neither agree nor disagree
Agree
Strongly agree
Each response is typically coded numerically (1 through 5) in the order shown, so higher numbers represent stronger agreement. This coding only works this way if the item is worded in a consistent, positive direction — reverse-worded items need reverse coding before they're combined with the rest.
7-Point Likert Scale: Example
The same statement on a 7-point scale might run from "Strongly disagree" through "Neither agree nor disagree" to "Strongly agree," with additional intermediate labels such as "Disagree," "Somewhat disagree," "Somewhat agree," and "Agree."
The practical difference isn't just more options — it's resolution. A 7-point scale can capture subtler differences among respondents who are all generally "agreeing" but at different strengths, which can improve statistical sensitivity in some analyses. The trade-off: respondents may find it harder to distinguish between adjacent points, and the extra categories add little value if your construct or sample doesn't need that precision.
How to Create a Likert Scale Questionnaire for Research
Define the construct. Be specific — "engagement" means different things in different studies.
Write focused items. Every item should only measure one concept.
Use clear, direct language. Avoid jargon and ambiguous qualifiers like "sometimes."
Avoid double-barrelled questions. It simultaneously poses two queries: "The course was well-organized and useful."
Keep response options balanced. Equal positive and negative options prevent the scale from skewing responses.
Decide on a neutral midpoint deliberately. Include one if genuine neutrality is plausible; omit it to force a directional answer.
Keep item direction consistent, or plan reverse coding. Mixing positive and negative wording can reduce response-set bias, but only if you reverse-code before analysis.
Pilot-test the questionnaire. A small trial run catches confusing wording before it costs you real data.
How to Score a Likert Scale
Scoring starts with assigning a number to each response category — for example:
Strongly disagree = 1, Disagree = 2, Neutral = 3, Agree = 4, Strongly agree = 5
From there:
Individual item scores are the number assigned to the respondent's chosen category.
Summed scores add the item scores together across a multi-item scale.
Average (composite) scores divide that sum by the number of items, keeping the score on the original range for easier comparison.
Reverse scoring flips the numbering for negatively worded items (1 becomes 5, and so on) before combining them with the rest, so higher numbers mean the same thing throughout.
The "correct" scoring approach depends entirely on how the instrument was designed — there's no universal rule that applies to every study.
How to Analyze Likert Scale Data
Individual Likert Items
A single Likert-type item is generally treated as ordinal data: the categories have a clear order, but the psychological distance between "agree" and "strongly agree" can't be assumed to equal the distance between "disagree" and "neutral." Frequency counts, percentages, mode, and median are the most defensible starting point here, since they don't require assuming equal intervals.
Multiple-Item Likert Scales
When several items combine into a composite score measuring one construct, many researchers treat that summed or averaged score as approximately continuous, particularly with five or more response points and enough items. This is a widely used convention, but still a convention — some methodologists urge caution, and reviewers may expect you to justify the choice. Before combining items this way, it's standard to check internal consistency (commonly reported as Cronbach's alpha) to confirm the items measure the same underlying construct.
Likert Scale Analysis Methods
| Method | Purpose | Type of Result | Suitable Use / Consideration |
|---|---|---|---|
| Frequency & percentage | Show response distribution | Descriptive | Appropriate for any single item; good for reporting and visualization |
| Median & mode | Central tendency for ordered data | Descriptive | Preferred over the mean for individual ordinal items |
| Mean & standard deviation | Central tendency and spread | Descriptive | Common for composite scale scores; use cautiously for single items |
| Mann–Whitney U / Kruskal–Wallis | Compare groups on ordinal data | Inferential (non-parametric) | Use when comparing two or more groups without assuming interval data |
| Spearman correlation | Relationship between ordinal variables | Inferential | Preferred over Pearson for individual ordinal items |
| Chi-square | Association between categorical variables | Inferential | Useful when comparing response category frequencies across groups |
| Cronbach's alpha | Internal consistency reliability | Diagnostic | Run before treating multiple items as one composite scale |
The right test depends on your research design, whether you're analyzing single items or a composite score, your sample size, and the distribution of your data — not on a single rule that applies to every Likert-based study.
How to Interpret Likert Scale Results
Interpreting Likert results means going beyond the raw average. A mean of 3.8 on a 5-point agreement scale is only meaningful once you know how responses are distributed — a tight cluster around 4 tells a different story than a bimodal split between 2 and 5, even with a similar mean.
Some sources publish fixed interpretation ranges (e.g., treating 1.00–1.80 as "strongly disagree"). Treat these cautiously — such cutoffs aren't universal statistical rules, and using them without justification can misrepresent your data. Any threshold you use should be explained within your own methodology, based on your scale design, sample, and research question — not borrowed uncritically from another study.
Common Mistakes When Using Likert Scales in Research
Confusing a Likert item with a full Likert scale in your write-up.
Using vague or ambiguous wording that respondents interpret differently.
Asking double-barrelled questions that combine two ideas in one item.
Using unbalanced response options that skew results in one direction.
Applying inconsistent coding without reverse-scoring where needed.
Forcing (or removing) a neutral midpoint without a clear rationale.
Adding response categories beyond what the construct or sample needs.
Choosing statistical tests without checking whether items or a composite scale are being analyzed.
Reporting only the mean without showing the underlying distribution.
Drawing causal conclusions from descriptive, self-reported data.
Likert Scale in Research: Example
Research topic: Student attitudes toward flipped-classroom instruction.Construct: Perceived engagement with course material Sample item: "I actively participate more in class discussions under the flipped-classroom format." Response scale: 5-point, strongly disagree to strongly agree, plus three related engagement items Coding: 1–5, with one negatively worded item reverse-coded Analysis: Cronbach's alpha confirms the four items form a reliable composite; composite scores are compared between two course sections with a Mann–Whitney U test Interpretation: Differences are discussed alongside each item's response distribution, not just the overall mean, with claims limited to association rather than causation
This end-to-end pipeline — construct, item, coding, analysis, cautious interpretation — is what a methods reviewer expects to see spelled out.
Frequently asked questions
A research tool that combines multiple ordered response items — typically strongly disagree to strongly agree — into a single score measuring attitudes, opinions, or perceptions that can't be observed directly.
Strongly disagree, Disagree, Neither agree nor disagree, Agree, and Strongly agree — usually numbered 1 to 5 for scoring.
It collects responses to subjective, qualitative questions but converts them into numbers, making the resulting data quantitative, even though the underlying attitude is qualitative in nature.
Individual items are ordinal, since equal distance between categories can't be assumed. Composite scores from multiple items are sometimes treated as approximately interval by convention, though this remains debated among methodologists.
A Likert item is a single rated statement. A Likert scale, strictly, is a set of multiple related items whose scores are combined into one composite measure of a construct.
A 5-point scale is simpler and works for most studies; a 7-point scale adds sensitivity for finer differences but can be harder for respondents to navigate. Match the choice to your construct and required precision.
Individual items are usually summarized with frequencies, percentages, median, and mode. Composite scores may use means and standard deviations, plus non-parametric tests like Mann–Whitney U or Kruskal–Wallis when comparing groups.