Re: When to use t-test?

desiree0311@xxxxxxxxx wrote:
Is there any rule/s stating that t-test is not applicable for samples
higher than 30? Please give me the site or the name of the
reference...Thank you very much

No, there is no such rule. Older textbooks often made a distinction between small- and large-sample tests (i.e., n =30+, or 50+ in some books), with the t-distribution being used for small samples, and the standard normal (z) being used for large samples. But note that the argument was that you *could* use the standard normal for large samples (because the approximation was good enough), not that you *had* to use it. That distinction was useful in the days of hand calculations, because tables of critical values could be smaller. But now that the tests are done using computers and stats packages that include the necessary probability density functions for computing p-values, there is little point in making that distinction.

--
Bruce Weaver
bweaver@xxxxxxxxxxxx
www.angelfire.com/wv/bwhomedir
"When all else fails, RTFM."
.