What Is a Likert Scale?
A Likert scale is a rating format used in surveys and questionnaires that measures attitudes, opinions, or perceptions along a symmetric agree-disagree continuum. Respondents select from a range of ordered response options, typically from "Strongly Disagree" to "Strongly Agree", that represent varying degrees of intensity toward a statement. Named after psychologist Rensis Likert, who introduced the format in 1932, it remains one of the most widely used measurement tools in social science, market research, and customer experience programs. A single statement paired with its response options is called a Likert item. When multiple Likert items are grouped to measure a single construct, they form a Likert scale.
Why Likert Scales Matter in Research
Likert scales give researchers a standardized way to convert subjective opinions into numerical data that can be compared, aggregated, and analyzed statistically. They're used in roughly 80% of published social science survey research, making them the default approach for measuring attitudes at scale. Without a consistent rating format, comparing sentiment across segments, time periods, or product concepts becomes unreliable.
How Likert Scales Work
At their core, Likert scales ask respondents to evaluate a declarative statement by selecting the option that best matches their position. The researcher then assigns numerical values to each response option for analysis.
Anatomy of a Likert Item
Every Likert item has two parts: a stem (the statement) and a set of ordered response options. For example:
Stem: "I found the checkout process easy to complete."
Response options: Strongly Disagree (1). Disagree (2). Neither Agree nor Disagree (3). Agree (4). Strongly Agree (5)
The statement should be clear, specific, and focused on a single idea. Double-barreled statements like "The product was affordable and easy to use" force respondents to answer two questions at once, which produces unreliable data.
5-Point vs. 7-Point Scales
The two most common formats are the 5-point and 7-point scales. Each has trade-offs.
| Factor | 5-Point Scale | 7-Point Scale |
|---|---|---|
| Ease for respondents | Simpler, faster | Slightly more cognitive effort |
| Sensitivity | Adequate for most use cases | Better at detecting small differences |
| Statistical reliability | Good | Marginally higher |
| Mobile-friendly | Yes | Can feel cramped on small screens |
| Best for | Customer satisfaction, quick polls | Academic research, product testing |
Research published in the Journal of Marketing Research shows that 7-point scales produce slightly more variance, which improves the scale's ability to detect meaningful differences between groups. But 5-point scales generate fewer abandoned surveys, particularly on mobile devices.
If you're running a short customer feedback survey, five points will do. If you're measuring nuanced attitude shifts across an experimental study, seven points give you more room to work with.
Odd vs. Even Point Scales
Odd-numbered scales (5, 7) include a neutral midpoint. Even-numbered scales (4, 6) force respondents to lean one direction or the other.
Forced-choice (even) scales can be useful when you genuinely need people to take a side, for example, purchase intent studies where "neutral" doesn't help you forecast demand. But removing the midpoint can frustrate respondents who legitimately feel neutral, pushing them toward random selection or survey abandonment.
Most methodologists recommend keeping the midpoint unless you have a specific analytical reason to remove it.
Likert Scales vs. MaxDiff
Likert scales and MaxDiff (Maximum Difference Scaling) both measure preferences, but they work differently and answer different questions.
| Aspect | Likert Scale | MaxDiff |
|---|---|---|
| What it measures | Absolute agreement with statements | Relative importance among items |
| Output | Individual item scores | Ranked preference scores |
| Scale bias risk | High (acquiescence, extreme response) | Low (forced trade-offs) |
| Best for | Measuring satisfaction, attitudes | Prioritizing features, messages, concepts |
| Respondent task | Rate each item independently | Choose best/worst from sets |
Likert scales let every item score high, which is why satisfaction surveys often show ceiling effects where most items cluster near the top. MaxDiff forces trade-offs, producing a clear rank order. When you need to know which features matter most rather than whether people like them, MaxDiff is the better tool.
Analyzing Likert Data
Likert data sits in a gray zone between ordinal and interval measurement, and how you treat it affects which statistical tests are valid.
Conservative approach (ordinal): Use medians, mode, and non-parametric tests like Mann-Whitney U or Kruskal-Wallis. This is technically correct since the distance between "Agree" and "Strongly Agree" isn't necessarily equal to the distance between "Disagree" and "Strongly Disagree."
Practical approach (interval): Calculate means, standard deviations, and run parametric tests like t-tests or ANOVA. Many researchers treat Likert data as interval when scales have five or more points, responses approximate a normal distribution, and sample sizes are adequate (n > 30).
Both approaches are defensible. The key is being consistent within a study and transparent about your assumptions.
When to Use a Likert Scale
- Customer satisfaction surveys where you need to track sentiment over time using a consistent benchmark
- Employee engagement studies measuring agreement with workplace culture statements
- Product concept testing to gauge initial reactions to features or positioning
- Brand perception research comparing how different audience segments view your brand attributes
- Pre/post studies measuring attitude change after an intervention, training, or campaign
Common Mistakes to Avoid
- Writing double-barreled statements that combine two ideas in one item, making responses impossible to interpret
- Using too many scale points (9 or 10) without a clear analytical need, which increases respondent fatigue without improving data quality
- Mixing scale directions within a survey without clear labeling, respondents develop a rhythm and may not notice when anchors flip
- Reporting only top-box scores (% selecting the highest option) instead of examining the full distribution, which hides important variation in responses
- Treating a single Likert item as a reliable measure of a complex construct, use multiple items and check internal consistency with Cronbach's alpha
How Quali-Fi Supports Likert Scales
Quali-Fi's survey platform includes pre-built Likert scale templates in 5-point, 7-point, and custom configurations across all plan tiers starting at $89/month. The platform's real-time dashboards display response distributions, mean scores, and cross-tabulated breakdowns by segment, so you can spot patterns without exporting to a spreadsheet. For teams that need to go beyond simple rating scales, Quali-Fi also offers MaxDiff and Conjoint analysis within the Research and Intelligence plans.
Frequently Asked Questions
Is a Likert scale ordinal or interval?
Technically, Likert data is ordinal because the spacing between response options isn't guaranteed to be equal. In practice, most researchers treat scales with five or more points as approximately interval for statistical analysis. The debate continues in academic circles, but treating the data as interval is widely accepted when sample sizes are large enough.
How many items should a Likert scale have?
A single construct should be measured by at least three to five Likert items to establish reliability. You can then average those items into a composite score. A lone Likert item can work for quick pulse checks, but it won't give you enough measurement stability for serious analytical work.
Should I label every point or just the endpoints?
Labeling every point (fully anchored scales) produces more reliable data than labeling only the endpoints. Research by Krosnick and Fabrigar found that fully anchored scales reduce respondent confusion and improve measurement consistency, especially in self-administered surveys where there's no interviewer to clarify.
Can I use a Likert scale for NPS-style questions?
Net Promoter Score uses an 11-point (0-10) numeric scale, which is structurally different from a Likert agree-disagree format. You can measure recommendation intent with a Likert item ("I would recommend this product to a friend"), but the scoring methodology and benchmarks won't be comparable to standard NPS data.
What's the difference between a Likert scale and a Likert item?
A Likert item is a single statement with its response options. A Likert scale is a collection of items measuring the same underlying construct, scored together. The distinction matters because reliability statistics like Cronbach's alpha apply to scales (multiple items), not individual items.