- Foundation
- About 15 min
- 5 practice questions
Measuring attitudes: how Likert, Thurstone, Guttman, semantic differential and social distance scales work
See how the main attitude scales are built and scored, work through a Likert score with reversed items, and understand why self-report scales can mislead and how implicit measures try to help.
If something looks wrong, please tell us.
An attitude cannot be seen. A psychologist, or an officer running a public survey, can only infer it from what people say or do. Over the twentieth century researchers designed scales that turn statements into numbers. This lesson goes beyond the names of the scales and explains how each one is built and scored, and where each one can go wrong.
Why measure at all?
A department that wants to know whether citizens favour a new service counter, or whether staff welcome a new procedure, needs more than a guess. A good measure gives a score that can be compared across groups and over time. But a score is only as good as the questions behind it, so an officer who commissions or reads a survey should know how it was built.
The main scales
| Scale | Developed by | How it works | Scoring |
|---|---|---|---|
| Likert | Rensis Likert | Respondent says how far they agree with each of several statements (strongly agree to strongly disagree) | Add the points; reverse the score of negatively worded items |
| Thurstone | L. L. Thurstone | Judges first sort many statements by how favourable they are; the respondent ticks those they agree with | Average the scale values of the statements ticked |
| Guttman | Louis Guttman | Statements run from mild to strong, so agreeing with a strong one implies agreeing with all milder ones | Count the number of statements accepted |
| Semantic differential | Charles Osgood | The object is rated on pairs of opposite adjectives, such as good-bad, on a seven-point line | Add or average the ratings |
| Social distance | Emory Bogardus | Respondent says how close a relationship they would accept with a group, in steps from close kinship to exclusion | The closest step accepted |
Worked example 1: scoring a Likert scale
A five-item scale on public libraries gives 5 points for 'strongly agree' down to 1 for 'strongly disagree'. Items 1, 2 and 4 are positive: 'Libraries are useful' (agree = 4), 'Libraries deserve more funding' (strongly agree = 5) and 'I would visit weekly' (neutral = 3). Items 3 and 5 are negative: 'Libraries waste space' (disagree = 2) and 'Books are better bought than borrowed' (disagree = 2). Because a negative item must be reversed, its score is 6 minus the raw score: 6 − 2 = 4 for each. The total is 4 + 5 + 4 + 3 + 4 = 20. The scale runs from 5 to 25, with a midpoint of 15, so 20 means a clearly favourable attitude.
Worked example 2: a Thurstone score and a Guttman pattern
In a Thurstone scale each statement has a scale value fixed by judges. A respondent ticks three statements with values 8, 9 and 7 on a scale where 11 is the most favourable. Their attitude score is the average, (8 + 9 + 7)/3 = 8, a favourable attitude. In a Guttman scale, suppose four statements about a group run from I 'I would talk to a member' to IV 'I would accept a member as a relative by marriage'. Respondent A accepts I, II and III and rejects IV: a consistent pattern with a score of 3. Respondent B accepts I and III but rejects II. That pattern is inconsistent with the cumulative idea, so B's answers contain an error or the scale is faulty.
Rule to remember: Likert adds agreement, Thurstone averages scale values, Guttman counts steps that must be cumulative, Osgood rates on opposite adjectives, Bogardus asks how close a relationship one would accept.
What can go wrong
- Social desirability: respondents give the answer that looks good, especially on sensitive topics such as caste or gender.
- Acquiescence: some people tend to agree with every statement. Mixing positive and negative items, as in Example 1, helps to detect this.
- Poor wording: questions that lead, or ask two things at once ('Are officers honest and friendly?'), give unclear answers.
- Words are not deeds: a person can truthfully report an attitude and still act differently, as LaPiere's study showed.
Because people may not wish, or may not be able, to report their attitudes, researchers also use implicit measures. The best known is the Implicit Association Test, developed by Anthony Greenwald and colleagues, which times how fast a person pairs categories with positive or negative words. Such measures are debated and are best treated as one source of evidence, alongside questionnaires and observed behaviour.
Key takeaways. Each scale turns agreement into a number in its own way. Reverse-score negative items, check for patterns of answering, and treat scores as evidence, not proof.
Check your understanding
-
1. On a five-point Likert item scored 5 to 1, a negatively worded statement receives the raw answer 'disagree' (2). What score is added to the total?
- 2
- 3
- 4
- 5
Show answer
Answer: C. Negative items are reverse-scored. With a 1 to 5 scale, the reversed score is 6 − 2 = 4.
-
2. In a Thurstone scale, how is a respondent's score obtained?
- By averaging the scale values of the statements they accepted
- By adding up their agreement ratings on all the statements
- By counting how many of the statements they rejected overall
- By finding the closest relationship that they would accept
Show answer
Answer: A. Thurstone's judges assign each statement a scale value in advance. The respondent's score is the average of the values of the statements they tick.
-
3. Which scale asks how close a relationship a person would accept with members of a group?
- Osgood semantic differential
- Guttman cumulative scale
- Likert summated scale
- Bogardus social distance scale
Show answer
Answer: D. The Bogardus scale lists steps from close kinship to exclusion and finds the closest step the respondent accepts.
-
4. Respondent B accepts the first and third statements of a four-step Guttman scale but rejects the second. What does this show?
- B holds a very strong attitude towards the group
- The pattern is not cumulative, so it is flawed
- B has a score of zero on the whole scale
- B holds a weaker attitude than one accepting only the first
Show answer
Answer: B. In a cumulative scale, accepting a stronger step implies accepting all the milder ones. Accepting the third but not the second breaks that rule.
-
5. A survey on caste attitudes finds that almost everyone reports tolerant views. Which problem should an analyst suspect first?
- Social desirability
- Acquiescence only
- A scale value error
- Guttman scoring
Show answer
Answer: A. On sensitive topics, people tend to give answers that look acceptable. Implicit measures and behavioural data can help to check for this.
Sign in to keep track of the lessons you finish.