![]() Then run a one-way ANOVA on this new table. To do Levene's test, first create a new table where each value is defined as the absolute value of the difference between the actual value and median of its group. Prism doesn't do this test (yet), but it isn't hard to do by Excel (combined with Prism). ![]() Some suggest using Levene's median test instead.So if the difference in variances is not huge, and especially if your sample sizes are equal (or nearly so), you might be safe just ignoring Barlett's test. It is too sensitive to minor differences that wouldn't really affect the overall variance. Some statisticians suggest never using Bartlett's test.You have strong evidence that the populations the data are sampled from are not identical. ![]() You haven't yet tested whether the means are distinct, but you already know that the variances are different.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |