Re: Chisquare and gamma coefficient

Posted by Julius Sim on Jan 23, 2011; 10:44am
URL: http://spssx-discussion.165.s1.nabble.com/SPSS-16-0-0-error-Two-step-cluster-Options-tp3353123p3353532.html

Hello,

If by chi-square you mean the Pearson chi-square, this test is designed
for nominal variables and takes no account of the ordinal ranking of the
categories in the variables, whereas gamma is a symmetric measure of
association for ordinal variables. Hence you are liable to get
contradictory conclusions regarding the null hypothesis of no association.
You need to use the chi-square test for trend (linear by linear
association in SPSS) to test for association in two ordinal variables.

Best wishes,

Julius

> I used chisquare to test the null hypothesis that the two ordinal
> variables with 4 categories each are uncorrelated.  The chisquare is
> non-significant (p>.05).  The gamma coefficient is .341 and highly
> significant (p<.05).  Why gamma is significant while chisquare is
> insignificant.  I confused about this.  I appreciate any comments.
>
> Thank you.
> Eins
>
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD