# Test for the equality of two variances

• Aug 9th 2008, 12:05 AM
iamterribleatmaths
Test for the equality of two variances
Hi, my question is;

Which of the following is/are assumptions underlying the test for the equality of two variances?
1) The samples are independent
2) Both populations are normally distributed
3) Both populations have the same mean
4) 1 and 2
5) 1 and 3
6) 1,2 and 3

Thanks!
• Aug 9th 2008, 01:03 AM
CaptainBlack
Quote:

Originally Posted by iamterribleatmaths
Hi, my question is;

Which of the following is/are assumptions underlying the test for the equality of two variances?
1) The samples are independent
2) Both populations are normally distributed
3) Both populations have the same mean
4) 1 and 2
5) 1 and 3
6) 1,2 and 3

Thanks!

Depends on the test. If you are talking about the Variance Ratio or F test then the assumption is that both populations are normally distributed, and the samples have to be independent.

RonL
• Aug 9th 2008, 02:27 AM
mr fantastic
Quote:

Originally Posted by iamterribleatmaths
Hi, my question is;

Which of the following is/are assumptions underlying the test for the equality of two variances?
1) The samples are independent
2) Both populations are normally distributed
3) Both populations have the same mean
4) 1 and 2
5) 1 and 3
6) 1,2 and 3

Thanks!

Quote:

Originally Posted by CaptainBlack
Depends on the test. If you are talking about the Variance Ratio or F test then the assumption is that both populations are normally distributed, and the samples have to be independent.

RonL

Indeed. There are many tests.

There's also an F-test for two population variances with correlated observations where it's assumed that that the observations have been performed in pairs and that correlation exists between the paired observations. That the populations are normally distributed is also assumed.

In both F-tests it's not necessary that the populations should have the same mean.

And it should be noted that the Levene test is less dependent on the assumption of normality than most tests, although it is assumed that the populations under consideration are approximately normally distributed.