Likert Scales: Everything You Ever Wanted to Know – But Never Dared to Ask

In this article, we cover the most important general considerations when constructing Likert scales.
However, if you’re looking for concrete examples and phrasing suggestions, check out our other blog post with a comprehensive overview of common Likert scales.

In this article:

About Likert Scales
Number of Response Options
Even or Odd Number of Options
Labeling the Scale Points
The Quirky History of Likert Scales

About Likert Scales

Named after its inventor Rensis Likert (1903–1981), the Likert scale is a widely used rating scale where respondents indicate the extent to which they agree or disagree with a given statement.

Typically, this is done using a 5- or 7-point scale ranging from one extreme to the other—for example, from “strongly disagree” to “strongly agree.”

Likert scales are popular because they are easy to understand, simple to construct, intuitive to answer, and straightforward to analyze. These qualities have made them a standard tool in today’s survey research—not just for measuring attitudes, but also opinions, perceptions, behaviors, and more.

Likert scales may appear simple at first glance—which is part of their appeal. But this simplicity conceals a need for thoughtful design. Anyone working with Likert scales should be aware of a few core principles to ensure reliable and meaningful results.

Below, we explore the key aspects you should consider when designing Likert scales:

Number of Response Options

Traditionally, Likert scales use 5 to 7 response options. In practice, however, you’ll also find scales with as few as two and up to eleven points. The ideal number of scale points depends on a variety of factors:

In general: The more response options, the finer the distinctions a scale can capture. However, respondents can only process a limited number of options meaningfully. To resolve this trade-off in your case, consider the following aspects:

Involvement and Knowledge of Respondents

The more your respondents know about the subject—or the more they care about it—the better they can identify and articulate subtle differences.

Example: In a survey of wine connoisseurs evaluating different wines, respondents can accurately assess nuanced taste differences—here, a scale with many response points is useful.

Conversely: If your audience has limited knowledge—because they’re not wine experts or the topic is a complex political issue—they likely lack the ability to make fine distinctions.

In such cases, a long scale may appear to provide more detail, but the results will likely reflect random variation or noise rather than true differences. This undermines data quality and complicates analysis.

In such cases, it’s better to use shorter scales—they produce more robust and reliable results.

Nature of the Rated Objects

Sometimes, the appropriate scale length depends on the nature of the object or concept being rated. Some items inherently involve subtle variations—like hotel room comfort or speaker sound quality. In other cases—such as basic everyday items like tissues or batteries—finer distinctions are less meaningful.

Survey Mode

How your survey is conducted also affects the ideal number of response options. If questions are read aloud by an interviewer—such as in phone interviews or at trade shows—respondents can typically process only a limited number of verbal options. In such cases, a shorter scale (max. 5 options) is advisable.

For online surveys, especially those on smartphones, make sure the full scale fits on the screen without scrolling. Otherwise, you risk introducing response bias.

Data Analysis

How do you plan to analyze your data? What’s the goal of your survey? These considerations are essential in choosing the right scale length.

If you’re planning only basic analysis—such as calculating averages, making general statements, or comparing groups—a longer scale offers little added value. In such cases, shorter scales are preferable.

If, however, you aim to analyze relationships or run advanced statistical models, longer scales are beneficial. For example, the correlation coefficient—a common measure of association—depends heavily on the number of response options. The fewer options, the lower the typical correlation. This can affect all analyses based on correlations, including regression models.

Even or Odd Number of Response Options

A scale with an odd number of options includes a clear midpoint, allowing for a neutral response. This neutral category enables respondents to express that they have no strong opinion—or to avoid engaging deeply with the topic.

You’ll need to decide whether to offer a neutral middle category or use a scale with an even number of options and no midpoint. Your decision should consider the following factors:

Respondents’ Knowledge and Topic Sensitivity

If you expect that some respondents lack a clear opinion—perhaps due to insufficient information or niche subject matter (e.g., “How do you rate the new EU data protection guidelines?”)—an odd-numbered scale is recommended. Without a neutral option, respondents may be forced into a position that doesn’t reflect their actual view, which can distort both the central tendency and the variance of your results.

In sensitive topics—like political or ethical questions (“What is your stance on assisted dying?”)—respondents should have the option to answer neutrally. Without it, they may feel uncomfortable, which could influence subsequent answers or lead to dropouts.

A potential compromise is to include an additional option like “Don’t know” or “No answer.” This way, respondents who have a clear opinion provide strong, contrasting answers—while those who feel unsure or uninformed can opt out meaningfully.

Research Goals

In other contexts, it may be desirable to avoid neutral responses and push respondents to take a position—especially if the survey will inform specific decisions (e.g., “Should the cafeteria serve only vegetarian meals?”). In such cases, neutral answers are of limited value. Here, a scale with an even number of options and no midpoint is preferable.

General Recommendation

The choice between even and odd scales significantly impacts your results and interpretations. As a rule of thumb: Most people have an opinion on familiar topics. Ensure your audience has enough background knowledge to provide honest and informed responses.

Labeling the Scale Points

Should every scale point be labeled, or are a few labels enough?

There is no clear evidence that labeling every single scale point is better than labeling only some of them. Research suggests that it doesn’t make a major difference whether all or just a few points are labeled.

In fact, using too many words or overly nuanced terms can sometimes confuse respondents—especially when labels are difficult to distinguish (e.g., “somewhat positive” vs. “fairly positive”).

What really matters is to avoid ambiguity or vagueness in your labeling. Respondents should always understand what dimension the scale is measuring. This is best achieved by clearly labeling the two poles of the scale—and, for odd-numbered scales, also the midpoint.

Minimal labeling is especially useful when space is limited—such as when using sliders or matrix questions, where labeling every point can quickly clutter the layout.

Likert Scales

Pointed vs. Flat Response Distributions

Another important aspect relates to how the scale endpoints are phrased—specifically, how extreme the wording is at either end.

Extremely worded endpoints (e.g., “completely dissatisfied” vs. “completely satisfied”) tend to produce a pointed distribution, where fewer respondents choose the extremes and more cluster around the middle.
Moderately worded endpoints (e.g., “satisfied” vs. “dissatisfied”) tend to yield a flatter distribution, as respondents are more comfortable choosing the extreme points—resulting in more nuanced responses across the scale.

Which version to choose depends on your research objective:

If your goal is to highlight subtle differences in opinion and encourage more expressive responses, use moderate wording (flatter distribution).
If you want to highlight only strong and clear positions, choose extreme wording (pointed distribution).

So be deliberate when designing your scale—choose a format that best aligns with your survey goals.

The Quirky History of Likert Scales

Originally developed to measure attitudes, the Likert scale emerged in the 1930s when psychologists began exploring how to quantify abstract constructs such as beliefs and attitudes.

The central challenge: Attitudes are invisible and not directly observable. People may like or dislike something for entirely different reasons—taste, looks, past experiences. To make these individual judgments comparable, a systematic approach was needed.

This is where American psychologist Rensis Likert came in with a simple yet brilliant idea:
He proposed breaking down an attitude into individual dimensions and phrasing one statement per dimension. Respondents were then asked to indicate how much they agreed or disagreed with each statement—typically using a five- or seven-point scale.

The individual responses were then aggregated into a total score, yielding a unified and comparable measure of a person’s attitude—regardless of which dimensions mattered most to that individual.

A simple example:
A person’s attitude toward apples might include dimensions such as taste, appearance, smell, juiciness, color, variety, shape, or size. Together, these aspects form a latent construct called “attitude toward apples.”

Likert’s real innovation wasn’t the rating scale itself—such scales already existed—but rather the multi-item approach to measuring attitudes and combining them into a single index. The core challenge then (and now) is choosing the best statements to capture the construct—this process is called scale construction and validation.

This approach paved the way for the measurement of latent constructs using multi-item scales, and inspired a wave of new methods for assessing everything from job satisfaction to brand image, trust, loyalty, engagement, personality traits, and social attitudes.

Over time, the concept of the Likert scale evolved to include not only agreement ratings but also measures like importance, likelihood, frequency, and preference.

Curiously, the term “Likert scale” in everyday use has shifted from Likert’s original concept—the multi-item agreement measurement—and now mostly refers to the rating format itself.

So today, we live with the odd reality that what we usually call a “Likert scale” isn’t quite what Rensis Likert had in mind.

By the way, Likert’s original paper was published in 1932:
Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology.