PsychologyScale ConstructionUniversity
AQAAPOntarioNSWCBSEGCE O-LevelMoECAPS

Guttman Scale (Coefficient of Reproducibility)

Measures how well a set of items forms a cumulative (Guttman) scale, indicating unidimensionality.

Understand the formulaSee the free derivationOpen the full walkthrough
Open Full WalkthroughTry Calculator

This public page keeps the free explanation visible and leaves premium worked solving, advanced walkthroughs, and saved study tools inside the app.

Core idea

Overview

The Coefficient of Reproducibility (CR) is a key statistic used to evaluate the unidimensionality of a Guttman scale. A Guttman scale is a cumulative scale where agreement with a higher-level item implies agreement with all lower-level items. The CR quantifies the extent to which an individual's responses to a set of items can be reproduced from their total score, indicating how closely the observed data fit the perfect Guttman pattern. A high CR (typically 0.90 or above) suggests that the items form a strong cumulative scale.

When to use: Apply the Coefficient of Reproducibility when constructing or evaluating a Guttman scale to assess its unidimensionality and cumulative property. It is used to determine if a set of items measures a single underlying construct in a hierarchical manner.

Why it matters: Understanding CR is crucial for developing valid and reliable psychological scales. A high CR ensures that the scale accurately reflects a single, ordered dimension, which is fundamental for meaningful measurement in areas like attitude assessment, developmental stages, or clinical symptom severity.

Symbols

Variables

= Number of Errors, = Number of Responses, CR = Coefficient of Reproducibility

Number of Errors
Variable
Number of Responses
Variable
CR
Coefficient of Reproducibility
Variable

Walkthrough

Derivation

Formula: Guttman Scale (Coefficient of Reproducibility)

The Coefficient of Reproducibility measures the proportion of responses that fit the perfect Guttman scale pattern, indicating unidimensionality.

  • Items are dichotomous (e.g., agree/disagree, yes/no).
  • The scale is intended to be cumulative, meaning items can be ordered by difficulty or endorsement.
1

Define a Guttman Scale:

A perfect Guttman scale assumes that if a person agrees with a more 'difficult' item, they will also agree with all 'easier' items. Any deviation from this pattern is considered an 'error'.

2

Identify Errors:

Errors occur when a respondent answers a 'difficult' item correctly but an 'easier' item incorrectly, or vice-versa, violating the cumulative property. These are counted for each respondent across all items.

3

Calculate Total Responses:

This is the total number of individual item responses collected across all participants in the study.

4

Formulate Coefficient of Reproducibility (CR):

The CR is calculated by subtracting the proportion of errors from 1. A CR of 1 indicates a perfect Guttman scale with no errors, while lower values indicate more deviations from the ideal cumulative pattern.

Result

Source: Guttman, L. (1944). A basis for scaling qualitative data. American Sociological Review, 9(2), 139-150.

Free formulas

Rearrangements

Solve for

Guttman Scale: Make Number of Errors the subject

To make 'Number of Errors' the subject of the Coefficient of Reproducibility formula, rearrange the equation to isolate the error term.

Difficulty: 2/5

Solve for

Guttman Scale: Make Number of Responses the subject

To make 'Number of Responses' the subject of the Coefficient of Reproducibility formula, isolate the fraction containing 'Number of Responses' and then solve for it.

Difficulty: 3/5

The static page shows the finished rearrangements. The app keeps the full worked algebra walkthrough.

Visual intuition

Graph

The graph is a downward-sloping linear function where the coefficient of reproducibility decreases at a constant rate as the number of errors increases. For a psychology student, this means that a small number of errors indicates a high coefficient of reproducibility and strong unidimensionality, while a large number of errors suggests the scale fails to measure a single construct effectively. The most important feature of this linear relationship is that each additional error reduces the coefficient of reproducibility by a fixed amount relative to the total number of responses.

Graph type: linear

Why it behaves this way

Intuition

Imagine a set of items like rungs on a ladder, where ascending to a higher rung implies having stepped on all lower rungs. A Guttman scale aims for this perfect cumulative pattern, and 'errors' represent deviations where

CR
The Coefficient of Reproducibility; a measure of how closely observed responses align with the expected pattern of a perfect cumulative (Guttman) scale.
A higher CR value indicates that the scale items form a stronger cumulative hierarchy, meaning an individual's responses are more predictable from their total score.
Number of Errors
The total count of individual item responses that deviate from the expected pattern of a perfect Guttman scale.
More errors indicate a poorer fit to the Guttman model, suggesting the items do not form a strong cumulative scale and reducing the scale's unidimensionality.
Number of Responses
The total number of individual responses collected across all items and all respondents in the dataset.
This term normalizes the number of errors, providing a proportion of errors relative to the total data points, allowing for a standardized measure of reproducibility.

Signs and relationships

  • 1 - \frac{\text{Number of Errors}}{\text{Number of Responses}}: The subtraction from 1 ensures that a perfect scale (zero errors) yields a CR of 1, representing perfect reproducibility. As the proportion of errors increases, the CR decreases, reflecting a poorer fit to the Guttman

Free study cues

Insight

Canonical usage

The Coefficient of Reproducibility is a dimensionless statistic used to assess the unidimensionality and cumulative property of a Guttman scale, typically reported as a decimal or percentage.

Common confusion

Students sometimes incorrectly assign units to the Coefficient of Reproducibility or its components, or confuse it with other coefficients that may have specific units or ranges. It is a pure number (a proportion).

Dimension note

The Coefficient of Reproducibility is a dimensionless ratio of counts, representing a proportion. It does not have physical units.

Unit systems

CRdimensionless · Represents a proportion, indicating the degree to which responses fit the Guttman pattern. Values range from 0 to 1.
Number of Errorscount · The total count of individual responses that deviate from the perfect Guttman scale pattern.
Number of Responsescount · The total count of all possible item responses across all participants.

Ballpark figures

  • Quantity:

One free problem

Practice Problem

A researcher constructs a Guttman scale with 5 items administered to 40 participants, resulting in 20 errors. Calculate the Coefficient of Reproducibility (CR).

Number of Errors20
Number of Responses200

Solve for: CR

Hint: Total responses = number of items × number of participants. Then apply CR = 1 - (Errors / Total Responses).

The full worked solution stays in the interactive walkthrough.

Where it shows up

Real-World Context

Developing a scale to measure stages of moral development, where agreeing with a higher stage implies agreement with all preceding stages.

Study smarter

Tips

  • A CR of 0.90 or higher is generally considered acceptable for a Guttman scale.
  • The 'Number of Errors' refers to deviations from the perfect cumulative pattern.
  • The 'Number of Responses' is the total number of item responses across all participants.
  • CR is sensitive to the number of items and respondents; larger scales might naturally have more errors.

Avoid these traps

Common Mistakes

  • Incorrectly identifying or counting 'errors' in the Guttman scaling process.
  • Confusing CR with other reliability measures like Cronbach's Alpha, which assess internal consistency differently.
  • Applying Guttman scaling to data that are not inherently cumulative or ordinal.

Common questions

Frequently Asked Questions

The Coefficient of Reproducibility measures the proportion of responses that fit the perfect Guttman scale pattern, indicating unidimensionality.

Apply the Coefficient of Reproducibility when constructing or evaluating a Guttman scale to assess its unidimensionality and cumulative property. It is used to determine if a set of items measures a single underlying construct in a hierarchical manner.

Understanding CR is crucial for developing valid and reliable psychological scales. A high CR ensures that the scale accurately reflects a single, ordered dimension, which is fundamental for meaningful measurement in areas like attitude assessment, developmental stages, or clinical symptom severity.

Incorrectly identifying or counting 'errors' in the Guttman scaling process. Confusing CR with other reliability measures like Cronbach's Alpha, which assess internal consistency differently. Applying Guttman scaling to data that are not inherently cumulative or ordinal.

Developing a scale to measure stages of moral development, where agreeing with a higher stage implies agreement with all preceding stages.

A CR of 0.90 or higher is generally considered acceptable for a Guttman scale. The 'Number of Errors' refers to deviations from the perfect cumulative pattern. The 'Number of Responses' is the total number of item responses across all participants. CR is sensitive to the number of items and respondents; larger scales might naturally have more errors.

References

Sources

  1. Wikipedia: Guttman scale
  2. Psychometric Theory (3rd ed.) by Jum C. Nunnally and Ira H. Bernstein
  3. Psychometric Theory by Jum C. Nunnally and Ira H. Bernstein (3rd ed., 1994)
  4. Guttman, L. (1950). The basis for scalogram analysis. In Stouffer et al., Measurement and Prediction.
  5. Edwards, A. L. (1957). Techniques of Attitude Scale Construction.
  6. Guttman, L. (1944). A basis for scaling qualitative data. American Sociological Review, 9(2), 139-150.