Normality test violations

Razza

New Member
#1
I have a small sample size (N=26) made up of three groups (n=11, 6 & 9).

Whilst running a mixed design ANOVA I have violated both Box's Test of Equality of Covariance (p = .025). I continued my analysis and found that the pre condition also violates Levene's test (p = .016).

Is this a problem?

I am new to stats and have searched high and low online without any luck... Any help is much appreciated!

Thank you

Damon
 

CB

Super Moderator
#2
It sounds like your concern is with the homogeneity of variance assumption, not the normality assumption....?

You mention a mixed design - what is the within-subjects factor?
 
#3
Yes it seems my problem is with the homogeneity of variance assumption, as I understand it that is what is being tested by Levenes Test of Equality of Error.

I made a mistake regarding Box's Test of Equality and Covariance, there is no violation to this test as the significance level is .001 and in my test p = .025.

My design is this:
3 IV = RR n=6, RX n=11, XX n=9
2 conditions = pre and post test

So am I correct in saying that the within-subjects factor is the testing time or stage (pre/post)?

Thank you!