Guttman Scale definition
A Guttman scale (also known as cumulative scaling or scalogram analysis) is an ordinal scale type where statements are arranged in a hierarchical order so that someone who agrees with one item will also agree with lower-order, easier, less extreme items. These statements should reflect an increasing intensity of attitude and form a continuum that is accepted by the respondents. The point at which the respondent disagrees with a statement reflects the respondent’s scale position.
A Guttman scale presents a number of items to which the person is requested to agree or not agree. This is typically done in a 'Yes/No' dichotomous format. It is also possible to use a Likert scale, although this is less commonly used.
Guttman scales are very commonly used in political science, anthropology, public opinion, research, and psychology.
Guttman scale examples
The ideal Guttman scale is such that if the respondent disagrees, for example, with statement 4 (having agreed with statements 1 to 3) then the respondent will disagree with statement 5 and higher as these represent more extreme expressions of the attitude being investigated.
For example, a series of items on attitude could be
- "I am willing to be near a cat"
- "I am willing to have a cat"
- "I love to have a cat"
- "I am willing to touch a cat"
Or a series of items on difficulty:
- counting from 1 to 50
- solving addition problems
- solving subtraction problems
- solving multiplication problems
- solving division problems
Guttman scale advantages
- It can be used to answer many questions in a short amount of space and/or time.
- It is intuitively appealing to most people.
- It provides ranked data.
- More one-dimensional than Likert scaling
Guttman scale disadvantages
- The rank order of the statements may not be interpreted in the same way by the researcher, the subject, or by independent judges.
- Difficult to construct
- Scalogram analysis may be too restrictive, only a narrow universe of content can be used
- Cornell technique questionable
- Results no better than summated Likert scales