SciencesPo (version 1.3.9)

calc.CV: Cramer's V Coefficient for Tables

Description

Computes the Cramer's V coefficient of association for tables. The Cramer's V is a measure of effect size for a chi-square goodness of fit test.

Usage

calc.CV(x, y = NULL, ...)
## S3 method for class 'default':
calc.CV(x, y = NULL, ...)

Arguments

x

A vector or a matrix.

y

A vector that is ignored if x is a matrix and required if x is a vector.

...

Extra parameters pass to the table function.

References

Agresti, Alan (1996) Introduction to categorical data analysis. NY: John Wiley and Sons.

Examples

# Consider an experiment with two conditions, each with 100 participants.
# Each participant chooses between one of following three parties.

cond1 <- c(40, 25, 35)
cond2 <- c(25, 35, 45)
mat <- cbind(cond1, cond2)
rownames(mat) <- c( 'party1', 'party2', 'party3')

# To test the null hypothesis that the distribution of preferences
# is identical in the two conditions, we run a chi-square test:
stats::chisq.test(mat) # still significant

# However, if we want to estimate the effect size, we then use Cramer's V:
calc.CV(mat)

# Agresti (2002), table 3.10, p. 104
# 1991 General Social Survey: The effect size of race on party identification.
gss <- data.frame(
   expand.grid(race=c("black", "white"),
   party=c("dem", "indep", "rep")),
   count=c(103,341,15,105,11,405))

GSS = untable(gss, freq = "count")

calc.CV(GSS$race, GSS$party)