Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
Reporting t-test results
Message posted by Annette Hawkins on December 5, 1999 at 12:00 AM (ET)
I am comparing survey data from our 1995 and 1998 climate survey. My table shows if the t-test comes from equal variances (pooled) or unequal variances. Is this necessary?
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: Reporting t-test results
Message posted by Jack Tomsky on December 6, 1999 at 12:00 AM (ET)
The preliminary test for equality of variances is necessary, unless both sample sizes are large. If the variances are equal, then the t statistic is indeed distributed as Student's t. If the variances are unequal, the distribution of the statistic can be approximated by Student's t using Satterwaite's formula for the degrees of freedom. In the case of large sample sizes, the t is close to a normal distribution, irrespective of whether the variances are equal.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
