Violating the assumption of normality and homogeneity of variances. How to report?

Nivv

New Member
#1
I am trying to run One way ANOVA but after trying to test these two assumptions, I found that the data violated the assumptions.

I am now trying to use the non-parametric Kruskal wallis H test instead. But I have problems in reporting the violation of the assumptions in APA style.

Below are the results of the tests
Tests of Normality:
Shapiro-Wilk:
Group 1: df = 140 p = .096
Group 2: df = 182 p = .001
Group 3: df = 59 p = .124

Test of homogeneity of variances
Levene statistic = 3.02 df1 = 2 df2 = 378 p = .021


It will be nice if you guys can suggest other methods to deal with these violations of assumptions too!
Thanks a bunch!
 

Karabiner

TS Contributor
#2
Re: Violating the assumption of normality and homogeneity of variances. How to report

With n=384, violation of the normality assumption does not affect the
ANOVA results. Regarding the inhomogenous variances, there are
corrections for this (Brown-Forsythe, Welch).

With kind regards

K.
 

Nivv

New Member
#3
Re: Violating the assumption of normality and homogeneity of variances. How to report

I see.
So my Welch test gives significant results.
I have found that there are post hoc tests for unequal variances such as the Games-Howell post hoc test, so proceeding with that will be appropriate?
 
Last edited: