1

Depending on the textbook I use, sometimes the equality of error variances in a mixed ANOVA is considered an assumption of the model and sometimes not. I did a literature search, but found nothing.

Does the mixed ANOVA assume equal error variances, is it robust to a violation of this assumption or can this assumption be ignored all together?

amoeba
  • 93,463
  • 28
  • 275
  • 317
Firona
  • 77
  • 5
  • You should add references to the texts you used! – kjetil b halvorsen Sep 19 '16 at 08:54
  • Equality of variances is an assumption of ANOVA. However, it is said that for practical purposes differences of variances up to ten fold don't invalidate ANOVA results. I'm sorry that I can't point to any reference. – Pere Sep 19 '16 at 09:25
  • The answer to http://stats.stackexchange.com/questions/56971/alternative-to-one-way-anova-unequal-variance may be useful. – Pere Sep 19 '16 at 19:25

0 Answers0