[Richard Williams]

Sui, if all your variables are binary (and that is how you made it sound before, but perhaps that isn't true) and if you are on a fishing expedition for vars that are significantly correlated with the dependent variable, then just computing the correlations and significance levels (using pwcorr) will probably lead you to about the same conclusions that running a bunch of 2 by 2 tables would, and it will be much quicker. Like Joseph says, whether you should be engaging in any kind of fishing expedition is questionable.

[gokhan ekonomi]

hi there, i saw a explanation in an article which says "chi2 test more accurate than Fisher's exact test on large samples" so i m confused. how can i decide to my test in terms of sample numbers? in other studies says fisher test is useful both on small and large samples.

Kind regards

[gokhan ekonomi]

i would be very glad if someone could help me

[Maarten Buis]

The exact test, even in moderate sample sizes, requires an unreasonable amount of computations. The test is a permutation test, that computes a statistic for every possible table given the margins. Here is combinatorics at work: even with moderate sample sizes you get a humongous number of possible tables. Stata does use a smart algorithm to cut this down a bit, but there is only so much it can do.

Calling a test an exact test is good marketing, but that does not mean it is any good. The exact test tends to be overly conservative. Many approximate tests have much better coverage than the exact test, and are substantially faster. See:
Alan Agresti & Brent A. Coull (1998) Approximate is Better than “Exact” for Interval Estimation of Binomial Proportions, The American Statistician, 52:2, 119-126. https://doi.org/10.1080/00031305.1998.10480550

[Maarten Buis]

In #19 I have given arguments I found convincing against the over-use of the exact test. That does not mean that there aren't special cases where it is useful, nore does that mean you have to find them convincing. Welcome to doing science: there is no longer the easy answer.