Attitude Scales: Types, Examples, and How to Design One

Attitude scales turn opinions into numbers you can compare. They measure how favorable or unfavorable people feel about a product, policy, idea, or experience, using structured statements instead of a loose “what do you think?” question.

That structure matters. A yes-or-no answer hides the person who mostly agrees, the one who is unsure, and the one who feels strongly. A well-built scale shows all three.

In this blog, we’ll define attitude scales, compare the four classic types, and show you how to design, test, and score your own.

What are attitude scales?

An attitude scale is a set of statements or adjective pairs that turns a person’s favorable or unfavorable view of something into a score. In research, an attitude is a lasting positive or negative evaluation of an object, person, idea, or situation. Nobody can see it directly, so the scale infers it from structured answers.

Attitude scaling is assigning a number to a person’s attitude so you can compare it with other people’s. Some textbooks call these tools attitude measurement scales. The idea is the same.

Picture a university asking students how they feel about online classes. It shows this statement:

Online classes are just as effective as in-person classes.

Students then pick one answer:

• Strongly agree
• Agree
• Neutral
• Disagree
• Strongly disagree

Each answer maps to a number from 1 to 5. Add several statements on the same topic, and the total becomes the student’s attitude score. Psychology, education, marketing, health care, and public opinion research all use this approach.

How are attitude scales different from rating scales and other survey questions?

An attitude scale measures one underlying opinion across several items, while a rating scale is only a response format. People mix the two terms up often, and the mix-up leads to weak question choices. This table separates the five terms you will meet most.

Every attitude scale relies on a rating format, but not every rating scale measures an attitude. Rating a delivery driver from 1 to 5 scores performance, not a belief or feeling. Our guide to rating scale question types covers the formats in detail.

Why use attitude scales instead of yes-or-no questions?

Attitude scales show how strongly people feel, not just which side they pick. That detail changes what you can do with the results. A yes-or-no question lumps together the enthusiast, the lukewarm supporter, and the person who has not decided.

Scales fix this in four ways.

1. They capture intensity.
   A person who mildly agrees and a person who strongly agrees give the same answer to a yes-or-no question. On a five-point scale, they land on different points.
   
2. They produce numbers you can analyze.
   Coded responses give you averages, distributions, and group comparisons. Run the same scale next quarter, and you can see whether attitudes moved.
   
3. They stay easy for respondents.
   Agreement scales are familiar, so people finish them quickly. Short formats also keep respondents from rushing.
   
4. They support decisions.
   A mean of 2.4 out of 5 on “I trust this brand” tells a marketing team something a count of thumbs-up votes never will.

What are the main types of attitude scales?

Four classic attitude scales dominate research: Likert, Thurstone, Guttman, and semantic differential. Each came from a different researcher who wanted to turn an opinion into a number. Thurstone and Chave published their method in 1929. Likert followed in 1932, Guttman in 1944, and Osgood and colleagues introduced the semantic differential in 1957.

The four types differ mainly in what respondents do:

• Likert: Rate agreement with each statement
• Thurstone: Pick the statements they accept
• Guttman: Agree up a ladder of rising intensity
• Semantic differential: Place a concept between two opposite adjectives

Likert scale

A Likert scale is a summated rating scale. Respondents rate their agreement with several statements, and you add the item scores into one attitude score. It is the most common attitude scale in use, and you have almost certainly answered one. The classic format runs from strongly disagree to strongly agree on five or seven points.

Here is a three-item set that measures attitude toward remote work:

• Remote work makes me more productive.
• I would choose a remote job over an office job.
• Remote work weakens team relationships. (reverse-scored)

Every item uses the same five-point agreement scale. Reverse scoring flips a negatively worded item so a high number always means a favorable attitude. Our Likert scale examples show odd and even versions side by side.

Thurstone scale

A Thurstone scale uses statements that judges have already rated for how favorable or unfavorable they are. Respondents tick only the statements they accept, and their score is the average value of those statements. Researchers call the method equal-appearing intervals.

The judging step makes the scale precise, and it also makes it slow. A panel rates a large pool of statements on an 11-point favorability scale. Statements the judges disagree on get dropped. What remains is a set spread evenly from very negative to very positive.

Take attitude toward exercise. “Exercise is essential for a healthy life” would carry a high value, and “Exercise is a waste of time” a low one. Someone who accepts both ends up with a middling score, which often signals an inconsistent view. Our Thurstone scale guide walks through the full build.

Guttman scale

A Guttman scale is a cumulative scale. Agreeing with a stronger statement implies agreeing with every milder one before it. Picture a ladder where each rung asks for a little more commitment.

A climate policy ladder might look like this:

• I believe in recycling.
• I support stricter environmental laws.
• I am willing to pay higher taxes to fight climate change.
• I would join a climate protest.

Someone who agrees with the protest item should also agree with the three above it. Real data rarely fits that neatly, so researchers test the pattern with a coefficient of reproducibility, which shows how closely actual answers match the ladder. Our Guttman scale guide covers the step-by-step build.

Semantic differential scale

A semantic differential scale asks respondents to place a concept between two opposite adjectives, usually on a seven-point line. It measures the feeling attached to something rather than agreement with a claim. Marketers use it to map brand image.

A bank might ask customers to rate itself on pairs like these:

Average the scores for each pair, and you get a brand profile. Plot a rival’s profile beside it, and the gaps become easy to see.

Other attitude scales worth knowing

Three more scales appear often in textbooks and are worth a quick definition.

• Stapel scale: A single adjective rated from -5 to +5 for how well it describes the object. Researchers use it when bipolar opposites are hard to write.
• Bogardus social distance scale: A Guttman-type scale that measures how close a person will get to members of another group.
• Q-sort: Respondents sort a deck of statements into piles, from least to most like their own view.

How do you choose the right attitude scale?

Pick the scale that fits your attitude, your audience, and your time budget. Start with the construct, meaning the underlying idea you want to measure, and choose the technique second.

Three questions speed up the decision. Does the attitude sit on one line or several? Do you need precision or speed? Will your respondents tolerate a long list of statements?

What is the difference between unidimensional and multidimensional attitude scales?

A unidimensional attitude scale measures one attitude along one line, while a multidimensional scale measures several related attitudes at once. Thurstone, Likert, and Guttman scales are all unidimensional in design. They assume, for example, that support for online shopping sits on a single favorable-to-unfavorable line.

Many real attitudes have layers. Attitude toward surveys is a good example. Researchers built a nine-item survey attitude scale with three dimensions, survey enjoyment, survey value, and survey burden, using three items for each. One combined score would blur the fact that a person can find surveys valuable and still feel overloaded by them.

A 2024 test of that scale collected 6,235 usable responses from Idaho resident hunters. Burden scored lowest at an average of 2.8 on a five-point scale, enjoyment came in at 3.3, and value scored highest at 3.6. Three separate scores gave the researchers three separate stories.

What are some real-world attitude scale examples?

Attitude scales show up wherever someone needs opinions in numbers. These four examples cover the most common settings.

• Education: A teacher gives students five Likert items on motivation in online courses, then repeats them after changing the course format.
• Marketing: A brand team uses semantic differential pairs such as trustworthy to risky and compares its profile against two competitors.
• Public policy: A city agency builds a Guttman ladder on transit funding, from “I ride transit sometimes” to “I would vote for a tax to expand it.”
• Workplace research: An employer measures attitudes toward a diversity hiring policy with statements that judges rated in advance, as the Thurstone method requires.

How do you design an attitude scale, step by step?

Designing an attitude scale takes seven steps, from naming the construct to trimming the final items. Skipping the early steps is the usual reason a scale looks fine but measures nothing useful.

1. Define the attitude.
   Write one sentence that names the object and the direction, such as “attitude toward online learning.” Narrow topics beat broad ones. “Attitude toward school” is too wide to score.
   
2. Draft a large item pool.
   Write more statements than you plan to keep. Some will fail pretesting, and you will want spares.
   
3. Balance the wording.
   Mix favorable and unfavorable statements so agreement alone cannot produce a high score.
   
4. Pick one response format.
   Use the same anchors for every item.
   
5. Pretest with a small group.
   Pretesting is an essential step in questionnaire design. Ask testers which items were unclear, repetitive, or confusing.
   
6. Test reliability and validity. Run the checks below on a pilot sample.
   
7. Trim and lock.
   Drop weak items, keep the scale short, and freeze the wording before launch.

How should you write attitude statements?

Good attitude statements are short, neutral, and about one thing. Small wording changes can move answers. In a 2005 Pew survey, 51% of respondents favored one description of an end-of-life policy. Only 44% favored a harsher description of the same policy.

This table shows four common problems and fixes.

Spell out jargon and abbreviations too. A respondent who stops to decode a term is no longer reporting an attitude.

How many response options should an attitude scale have?

Five or seven points work for most attitude scales. Five is quicker to read, and seven allows finer shades of opinion. Pew Research Center generally keeps answer choices to four or five, partly because people struggle to hold more in mind at once, especially in phone surveys. Our comparison of the 5-point vs 7-point Likert scale weighs the trade-offs.

• Midpoint: An odd scale offers a neutral option. An even scale forces a lean. Use odd when neutrality is real, and even when you need a direction.
• Labels: Label every point with words, not only the two ends, so everyone reads the points the same way.
• Direction: Keep the scale running the same way across the whole questionnaire.
• Order: Present ordered options in order. Shuffling them would scramble the continuum.

How do you test whether an attitude scale works?

An attitude scale works when it is both reliable and valid. Reliability means it gives consistent results. Validity means it measures the attitude you intended. Test both on a pilot sample before the full launch.

The most common reliability check is Cronbach’s alpha, a statistic that shows how closely the items in a scale move together. Values above 0.7 are generally treated as acceptable. In a 2024 study of 6,235 Idaho hunters, the three survey attitude subscales scored alphas of .85, .79, and .73. The same study found that higher survey burden went with lower response, at a correlation of -.16, while enjoyment showed a small positive link of .08.

Validity needs more than one check. Content validity asks whether experts agree the items cover the attitude. Construct validity asks whether items group into the dimensions you expect, which researchers often test with factor analysis, a method that shows which items cluster together. Criterion validity asks whether scores predict something real, such as behavior.

How do you score and analyze attitude scale responses?

Add the item scores after reverse-scoring any negative items, then divide by the number of items to get a mean you can compare across groups. On a five-point scale, the reverse-scoring formula is six minus the response. Here is a worked example:

1. A student rates three statements on a five-point scale. “Online classes are as effective as in-person classes” gets a 4, “I would take another online course” gets a 5, and “Online classes make it hard to stay motivated” gets a 2.
2. Reverse the third item. Six minus 2 equals 4.
3. Add the three scores. 4 plus 5 plus 4 equals 13.
4. Divide by three items. 13 divided by 3 is 4.33, a clearly favorable attitude.

Many researchers debate whether Likert data is ordinal or interval. Ordinal data has a clear order but unequal gaps between points. Single items are ordinal, yet many analysts still average summed scales and report full response distributions as well. Our ordinal scale guide explains the difference.

Segment results by group, then repeat the survey with identical wording to track change over time.

What mistakes ruin attitude scales?

Most failures trace back to biased wording, mixed-up constructs, and respondent habits that nobody planned for.

• Acquiescence bias: This is the tendency to agree regardless of content. Pew notes that less educated and less informed respondents show it more, and that it grows when an interviewer is present. Mixing favorable and unfavorable items helps.
• Social desirability bias: People shade answers toward what looks acceptable, especially on sensitive topics. Self-administered surveys tend to reduce it, because the bias runs stronger when an interviewer is present.
• Straight-lining: Respondents tick the same column all the way down a grid. Keep grids short, and include a reverse-worded item to catch it.
• One item posing as a scale: A single item cannot be tested for reliability. Use several items per attitude.
• Flipped direction: Reversing the scale halfway through confuses people. Keep anchors consistent.
• Mixed constructs: Satisfaction, trust, and loyalty are different attitudes. One item set for all three blurs the results.

How do you add an attitude scale to a QuestionPro survey?

QuestionPro lets you build a multi-item attitude scale in one matrix block, then export the responses for reliability testing. Three question types fit attitude work: Likert Scale, Matrix Table, and Semantic Differential. A matrix table is a grid question that shows several statements against one shared set of answer choices.

1. Open or create a survey in QuestionPro survey software, then click Add Question where the scale belongs.
2. Choose Likert Scale or Matrix Table for agreement statements. Choose the semantic differential scale question for opposite adjective pairs.
3. Put every item for one attitude in a single matrix, so respondents see the same anchors on each row.
4. Preview the survey, then share it by email, link, or social media.
5. Check the built-in reports for response distributions, averages, and charts. Filter by demographics or behavior to compare groups.
6. Export the data to Excel, CSV, or SPSS, a statistics program, to run Cronbach’s alpha and factor analysis. Recurring surveys let you track attitudes over time.

Choose the construct first, then the scale

Most weak attitude surveys fail before anyone writes the first question. The team picks a familiar scale, then bends the topic to fit it. Reverse that order. Decide exactly which attitude you want to know. Decide whether it has one dimension or several. Then let that choice point you to Likert, Thurstone, Guttman, or semantic differential.

A scale is a measuring instrument, and the number it gives you is only as honest as the question behind it. Before you build, ask yourself three things:

• What exact attitude am I measuring, and toward what?
• Does it sit on one line or on several?
• Will I test the scale before I trust it?

Frequently Asked Questions (FAQs)