[EMAIL PROTECTED] (Fedra) wrote in 
news:[EMAIL PROTECTED]:

> I have a little question. Recently I have read that Chi square and
> Cramers V are related. Therefore I would like to know if I may
> interpret the results of Cramers V (in significant cases) when the Chi
> Square conditions (expected count less then 5, or p > 0,05) are not
> met.

Cramer's V is a normalization of the Pearson chi-square statistic, but has 
nothing to do with the theoretical chi-square distribution and therefore 
its interpretation isn't affected by how quickly the Pearson statistic 
converges to a true chi-square under the null hypothesis.  It's a purely 
descriptive measure, though unless there are standards in your field for 
how to interpret it, I'd go with different measures of nominal-nominal 
association that have more-or-less standard interpretations.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to