is there alternative to chi squared test?

kate_liverpool · **Joined:** Thu Mar 05, 2009 2:03 pm **Posts:** 19

I wanted to use the chi squared test on some data, but I have some categories which have a value less than 5. Also some of the expected values for some categories are zero. What can I use instead?

Thanks,

Kate

Fierce · **Posted:** Fri Mar 13, 2009 8:51 pm

You can use Fisher's exact test (ANALYZE>DESCRIPTIVES>CROSSTABS>EXACT).

statman · **Posted:** Sun Mar 15, 2009 6:40 pm

Fisher's Exact might work but only if this module is installed.

As far as alternatives to overcome the small case count, there isn't. Know that this is a warning and doesn't mean that the chi-square results is wrong, only that "use with caution."

kate_liverpool · **Joined:** Thu Mar 05, 2009 2:03 pm **Posts:** 19

what does that mean? I guess I can use the test but does it mean the answers aren't necessarily true or correct?

I tried the crosstabs, but exact wasn't there as an option, I guess that means I can't use Fishers exact.

Thanks for you help statman

Kate

statman · **Posted:** Mon Mar 16, 2009 12:46 pm

Right, if Exact isn't noted then not installed

And right again, you can use the results just note that there are case count issues

is there alternative to chi squared test?

Who is online