That is, Cramer’s V for any given set of data can only be into interpreted in the light of those for other sets.
As a result of these problems of interpretation, in recent times measures of association based on chi-square have become less popular. In their place other measures of association, called proportional reduction in error measures (PRE), are calculated. Cramer’s V is still useful in determining whether a significant chi-square result is due to sample size, but is the condition for the use of PRE measures hold, these are preferred as actual measures of association.
All PRE measures follow a similar logic: we predict the dependent variable while ignoring the information provided by the independent variable, and then we predict the dependent variable using knowledge of the independent, and see if we make fewer mistakes. The PRE measures for nominal data is lambda.