association
Message posted by Olivier (via 212.190.123.113) on March 5, 2001 at 4:37 AM (ET)

I think my problem is very simple but I don't get the point.

I have a random sample and I want to test the association between a nominal variable and the age of the respondents (coded in age categories). So far there's no problem (I think): I use chi² and Cramers V.
But now I'm asked to test the significance level and association only within the oldest three age categories.
Do I have to split my data file for this and compare groups or are there other ways to do this?