SociologyQuantitative Research MethodsUniversity

AQAAPOntarioNSWCBSEGCE O-LevelMoECAPS

Cramer's V (Effect Size)

Measures the strength of association between two nominal variables in a contingency table.

Understand the formulaSee the free derivationOpen the full walkthrough

V = \frac{χ ^{2}}{n \cdot min ( k - 1 , r - 1 )}

Open Full Walkthrough Try 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

Cramer's V is a measure of association between two nominal variables, derived from the Chi-Square statistic. It ranges from 0 to 1, where 0 indicates no association and 1 indicates a perfect association. Unlike the Chi-Square statistic, Cramer's V is not affected by sample size or the number of categories, making it a useful measure for comparing the strength of relationships across different studies.

When to use: Applied after a significant Chi-Square test to quantify the practical strength of the association between two categorical variables. It is particularly useful when comparing the strength of relationships across different contingency tables or studies with varying sample sizes and dimensions.

Why it matters: Provides a standardized effect size for categorical data, complementing the Chi-Square test's p-value. In sociology, it helps assess the substantive importance of relationships between social categories (e.g., ethnicity and voting behavior, gender and occupation), moving beyond mere statistical significance to practical relevance.

Symbols

Variables

$χ^{2}$ = Chi-Square Statistic, n = Total Sample Size, $min$ (k-1, r-1) = Minimum of (Rows-1, Cols-1), V = Cramer's V

χ^{2}

Chi-Square Statistic

dimensionless

n

Total Sample Size

count

min (k - 1, r - 1)

Minimum of (Rows-1, Cols-1)

dimensionless

V

Cramer's V

dimensionless

Walkthrough

Derivation

Formula: Cramer's V (Effect Size)

Derives a standardized measure of association for nominal variables from the Chi-Square statistic.

Chi-Square test is appropriate for the data.
Variables are nominal or ordinal.
Expected frequencies are not too small.

Start with Phi Coefficient (for 2x2 tables):

The Phi coefficient is an effect size for 2x2 tables, but it can exceed 1 for larger tables.

ϕ = \frac{χ ^{2}}{n}

Generalize for larger tables (Cramer's V):

Cramer's V extends Phi by dividing by the minimum of (number of rows - 1) and (number of columns - 1), ensuring the value remains between 0 and 1 for any size contingency table.

V = \frac{χ ^{2}}{n \cdot min ( k - 1 , r - 1 )}

Result

V = \frac{χ ^{2}}{n \cdot min ( k - 1 , r - 1 )}

Source: Cramér, H. (1946). Mathematical Methods of Statistics. Princeton University Press.

Visual intuition

Graph

The graph follows a square root curve starting from the origin, where the value of V increases at a decreasing rate as the chi-square statistic grows larger. For a sociology student, this shape indicates that while a higher chi-square statistic suggests a stronger association between nominal variables, the effect size V eventually levels off, meaning that even very large statistical differences may represent diminishing returns in the strength of the relationship. The most important feature of this curve is that V is constrained by the square root relationship, which prevents the effect size from growing indefinitely even as the chi-square statistic increases.

Graph type: power_law

Why it behaves this way

Intuition

Imagine a grid of categories (a contingency table) where each cell holds a count. Cramer's V visualizes how much these counts cluster away from a perfectly uniform, random distribution across the grid, then scales this

V

A standardized measure of the strength of association between two nominal variables.

A higher V (closer to 1) means knowing the category of one variable gives you a much better idea of the category of the other variable; a V near 0 means they are largely independent.

χ^{2}

The Chi-Square test statistic, which quantifies the difference between observed frequencies and frequencies expected under the assumption of independence.

A larger

χ^{2}

value indicates a greater deviation from what would be expected if the variables were unrelated, suggesting a potential association.

n

The total number of observations or cases in the contingency table.

The Chi-Square statistic tends to increase with sample size, so 'n' is used in the denominator to adjust for this effect, making V comparable across studies with different sample sizes.

min (k - 1, r - 1)

The minimum of (number of columns - 1) and (number of rows - 1), representing the maximum possible degrees of freedom for the table's smaller dimension.

This term normalizes the Chi-Square statistic for the number of categories in the variables, ensuring V can range from 0 to 1 regardless of the table's dimensions.

Signs and relationships

\sqrt{}: The square root is applied to transform the Chi-Square-derived value, which is squared and variance-like, into a linear measure of association, making it more interpretable and comparable to other correlation
Denominator: This division normalizes the Chi-Square statistic. The 'n' accounts for sample size, preventing larger samples from artificially inflating the association strength.

Free study cues

Insight

Canonical usage

Cramer's V is a dimensionless statistical measure, reported as a value between 0 and 1, quantifying the strength of association between nominal variables.

Common confusion

Students sometimes confuse Cramer's V with a p-value. While both are dimensionless, Cramer's V measures the strength of an association (effect size), whereas a p-value indicates the statistical significance of the

Dimension note

Cramer's V is inherently dimensionless as it is a ratio derived from counts and a statistical test statistic (Chi-Square), which itself is dimensionless. It serves as an index of association, not a physical quantity.

Unit systems

$c h i^{2}$ dimensionless · The Chi-Square test statistic, which is a sum of squared differences between observed and expected frequencies divided by expected frequencies. It is inherently dimensionless.

$n$ dimensionless · The total sample size, representing the total number of observations in the contingency table. It is a count.

$k$ dimensionless · The number of columns in the contingency table. It is a count.

$r$ dimensionless · The number of rows in the contingency table. It is a count.

Ballpark figures

Quantity:

One free problem

Practice Problem

A Chi-Square test on the relationship between gender and preferred leisure activity yielded a Chi-Square statistic of 25. The total sample size was 100, and the minimum of (rows-1, cols-1) was 1. Calculate Cramer's V.

Chi-Square Statistic25 dimensionless

Total Sample Size100 count

Minimum of (Rows-1, Cols-1)1 dimensionless

Solve for: $V$

Hint: Take the square root of the Chi-Square statistic divided by the product of sample size and min(k-1, r-1).

The full worked solution stays in the interactive walkthrough.

Where it shows up

Real-World Context

After finding a significant Chi-Square association between educational attainment and political party preference, a sociologist calculates Cramer's V to determine the strength of this relationship.

Study smarter

Tips

Cramer's V values are typically interpreted as small (0.1), medium (0.3), or large (0.5) for effect size.
It is always non-negative, ranging from 0 to 1.
Ensure the Chi-Square statistic used is the total Chi-Square for the table.
The 'min(k-1, r-1)' term refers to the smaller of (number of columns - 1) or (number of rows - 1).

Avoid these traps

Common Mistakes

Interpreting Cramer's V as a measure of causation.
Using an incorrect Chi-Square value (e.g., from a single cell).
Miscalculating the degrees of freedom or the minimum of (rows-1, cols-1).

Common questions

Frequently Asked Questions

Derives a standardized measure of association for nominal variables from the Chi-Square statistic.

Applied after a significant Chi-Square test to quantify the practical strength of the association between two categorical variables. It is particularly useful when comparing the strength of relationships across different contingency tables or studies with varying sample sizes and dimensions.

Provides a standardized effect size for categorical data, complementing the Chi-Square test's p-value. In sociology, it helps assess the substantive importance of relationships between social categories (e.g., ethnicity and voting behavior, gender and occupation), moving beyond mere statistical significance to practical relevance.

Interpreting Cramer's V as a measure of causation. Using an incorrect Chi-Square value (e.g., from a single cell). Miscalculating the degrees of freedom or the minimum of (rows-1, cols-1).

After finding a significant Chi-Square association between educational attainment and political party preference, a sociologist calculates Cramer's V to determine the strength of this relationship.

Cramer's V values are typically interpreted as small (0.1), medium (0.3), or large (0.5) for effect size. It is always non-negative, ranging from 0 to 1. Ensure the Chi-Square statistic used is the total Chi-Square for the table. The 'min(k-1, r-1)' term refers to the smaller of (number of columns - 1) or (number of rows - 1).

References

Sources

Wikipedia: Cramer's V
Gravetter, F. J., & Wallnau, L. B. (2017). Statistics for the Behavioral Sciences (10th ed.). Cengage Learning.
Andy Field, Discovering Statistics Using IBM SPSS Statistics
J. Cohen, Statistical Power Analysis for the Behavioral Sciences
Andy Field Discovering Statistics Using IBM SPSS Statistics
Alan Agresti Categorical Data Analysis
Cramér, H. (1946). Mathematical Methods of Statistics. Princeton University Press.

Cramer's V (Effect Size)

Overview

Variables

Derivation

Start with Phi Coefficient (for 2x2 tables):

Generalize for larger tables (Cramer's V):

Graph

Intuition

Insight

Practice Problem

Real-World Context

Tips

Common Mistakes

Related Formulas

Pearson Correlation Coefficient

Frequently Asked Questions

Sources