Pearson Chi-square Exact test

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

annac
Posts: 3
Joined: Mon Oct 15, 2012 10:16 pm

Pearson Chi-square Exact test

Postby annac » Mon Oct 15, 2012 10:25 pm

Hi,
I am trying to analyse data from a survey. I have a lot of nominal-nominal, and nominal-ordinal comparisons to do and was thinking to use chi-square. When performing this test however I almost always recieve the warning that xx% of expected values are under 5. Since my cross tabulation table is larger than 2x2 I assume I can't use Fisher's exact test. However, in the SPSS manual, and help function in regards to the exact test function, it states that if I use the exact option and obtain the exact p-value instead of the asymp. p-value, then I can ignore the warning given.. Does anyone know if that is true? Is it appropriate to use the exact version of the pearson chi-square to get around the warning? It says in the help function that the assumptions for the chi-square can be ignored, but apart from documents from SPSS I can't seem to find that much info about this test (except for on the fishers exact test). Or is there another approach out there that I could take?

If anyone can help me that would be much appreciated!

This is the link to a document on the exact test module: http://public.dhe.ibm.com/common/ssi/ec ... 18USEN.PDF
Penguin_Knight
Posts: 473
Joined: Thu Apr 05, 2012 5:58 pm

Re: Pearson Chi-square Exact test

Postby Penguin_Knight » Mon Oct 15, 2012 10:59 pm

Do you have the "Exact Tests" module installed? It is a separate advanced function that doesn't come with standard Base SPSS. If you do, then yes, you can use their test to derive exact test statistics, it can even handle Fisher's Exact with table bigger than 2x2, it'd just take a longer time to get the results.
annac
Posts: 3
Joined: Mon Oct 15, 2012 10:16 pm

Re: Pearson Chi-square Exact test

Postby annac » Mon Oct 15, 2012 11:11 pm

yes I got the exact module installed. Or at least I assume so as I am able to select the exact option and it does calculate an exact p-value when chosen. Sometimes it takes a long time though. For Fisher's test however, it only gives me "1.00" as a p-value in most cases, and I assume it does that because it didn't have enough time to calculate it?

Thanks for your reply! So hard to find anything in journals on anyone who has used this test.
annac
Posts: 3
Joined: Mon Oct 15, 2012 10:16 pm

Re: Pearson Chi-square Exact test

Postby annac » Mon Oct 15, 2012 11:13 pm

And that means that I can ignore the warning given by the chi-square in the first place with the expected values?

Who is online

Users browsing this forum: No registered users and 1 guest

cron