Guttman Scale
Also known as: Cumulative scale, Scalogram analysis, Guttman scaling, Unidimensional cumulative scale
Guttman scaling is a methodology for constructing unidimensional scales with a cumulative property, developed by Louis Guttman in 1944. The method assumes that items form a perfect or near-perfect hierarchy: if a respondent endorses a harder item, they must endorse all easier items below it. This creates a reproducible scale structure useful for measuring constructs with ordinal properties such as difficulty, intensity, or severity.
Key highlights
- Provides a mathematically precise, reproducible scale in which scores perfectly or nearly perfectly predict response patterns
- Requires fewer items than Likert scales to measure a construct due to the informational efficiency of hierarchical ordering
- Clear interpretation: the score directly indicates the highest level of the construct endorsed
- Useful for detecting and quantifying scale structure violations through coefficients of reproducibility and scalability
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Guttman scaling is most appropriate for constructs with natural hierarchical or cumulative structure, such as ability levels, symptom severity, or intensity of experience. It is valuable in medical diagnostics, educational testing, and developmental assessments where ordering is intrinsic. However, it is less suited to constructs without clear ordering (e.g., personality traits) or where items do not relate hierarchically. Guttman scales are also useful for small samples or when efficient, parsimonious instruments are needed.
Strengths & limitations
- Provides a mathematically precise, reproducible scale in which scores perfectly or nearly perfectly predict response patterns
- Requires fewer items than Likert scales to measure a construct due to the informational efficiency of hierarchical ordering
- Clear interpretation: the score directly indicates the highest level of the construct endorsed
- Useful for detecting and quantifying scale structure violations through coefficients of reproducibility and scalability
- Requires items to follow a strict cumulative hierarchy, which is rare in practice; most real data violate perfect Guttman assumptions
- Restricted to dichotomous or ordinal responses; does not fully exploit rating scale options
- More demanding of sample size and item selection; finding items that scale perfectly is difficult
- Limited flexibility compared to multidimensional models (e.g., Rasch, IRT) that allow for probabilistic violations
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What is the difference between Guttman and Likert scales?
Likert scales assume items are parallel indicators of a construct with no necessary ordering; items are combined equally. Guttman scales assume items form a cumulative hierarchy where endorsing a 'harder' item implies endorsing all 'easier' items. Likert scales are more flexible and widely applicable; Guttman scales are more interpretable if the hierarchy holds.
What is a coefficient of reproducibility?
The coefficient of reproducibility (CRep) measures how accurately you can predict each respondent's complete item response pattern from their total score. CRep = 1 - (errors / total responses). A CRep > 0.90 indicates acceptable scale structure; values below 0.90 suggest the hierarchy is weak and the scale is less reproducible.
Can I use Guttman scaling with continuous responses?
Traditionally, Guttman scales use dichotomous responses, but the principle can extend to ordinal data (e.g., 0–4 rating). The cumulative property is harder to enforce with many response categories; modern approaches (Rasch, partial credit model) handle this better. For typical Guttman application, stick to yes/no or agree/disagree.
What should I do if my data don't fit the Guttman model?
If the coefficient of reproducibility is below 0.90, examine the error patterns. Are there items that violate the hierarchy? Reorder items or remove those with inconsistent patterns. Alternatively, acknowledge the violations and use a probabilistic model (Rasch, IRT) that permits some errors while estimating parameters.
Sources
- 1.Guttman, L. (1944). A basis for scaling qualitative data. American Sociological Review, 9(2), 139-150.
- 2.Guttman, L. (1950). The basis for scalogram analysis. In S. A. Stouffer et al. (Eds.), Measurement and Prediction. Princeton, NJ: Princeton University Press.
- 3.Menzel, H. (1953). A new coefficient for scalogram analysis. Public Opinion Quarterly, 17(2), 268-280.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Guttman Scale. ScholarGate. https://scholargate.app/psychometrics/guttman-scale