# statistic question (chi-square) - (Feb/21/2006 )

Dear all: got a catergorical analysis question and was wondering if someone can help me with it.

I have a 2x3 contigency table. Chi-square analysis from "Prism 4" shows a significance for the table. However is there a way I can determine whether the differences in the columns (1, 2, 3) are significant (for all possible pairs: 1-2, 1-3, 2-3)?

for example: lets say row1=death row2=recovered and column 1= control column 2-3=drugs are different drugs: chisquare P<.05 suggest that death or recovery is dependent on drugs BUT here it does not say which drug is it dependent on. Is there a way to do this?

An easy way I can see out of this is to do the various permutations in multiple fisher exact tests, but that just doesn't seem right since there is not any post-hoc correction.

-simplitia-

I think what you want to do is a Two way ANOVA

you can do Tukey's compare all possible matches to see whether A or B or C is the one having major effect...

I need a little refresher, but from what i remember this should do it.. Prism does two way ANOVAs and has the Tukey test too... You set the death recovery and drug as the two parameters

Something like that....

You could check the prism guide to be sure this is what you want, anyway, it is probably worth a shot... Hope this helps you...

-beccaf22-

QUOTE (beccaf22 @ Feb 21 2006, 07:16 PM)
I think what you want to do is a Two way ANOVA

you can do Tukey's compare all possible matches to see whether A or B or C is the one having major effect...

I need a little refresher, but from what i remember this should do it.. Prism does two way ANOVAs and has the Tukey test too... You set the death recovery and drug as the two parameters

Something like that....

You could check the prism guide to be sure this is what you want, anyway, it is probably worth a shot... Hope this helps you...

Hi, thanks for you input ...oh if only if it was that easy. Unfortunently these type of non-parametric techniques would not work for me - the reason is because my values are either Yes or No, recovered or not - there are no numerical values in between - thanks for your help though.

al

-simplitia-