Thread: ANOVA

Hey, my question is;
Consider the following statements
A. One way analysis of variance is used to compare whether c groups have equal variances
B. The one-way ANOVA F test is fairly robust against departures normality, particularly in large samples
C. There are n-1 degrees of freedom associated with the total sum of squares, as each observation is compared to the grand mean based on all observations.

I am pretty confident that A is wrong as it is used to comapre whether c groups have equal means.

I know the first part of C about the degrees of freedom is right, but don'y understand the last part.

Any help would be great!

Originally Posted by iamterribleatmaths

Hey, my question is;
Consider the following statements
A. One way analysis of variance is used to compare whether c groups have equal variances
B. The one-way ANOVA F test is fairly robust against departures normality, particularly in large samples
C. There are n-1 degrees of freedom associated with the total sum of squares, as each observation is compared to the grand mean based on all observations.

I am pretty confident that A is wrong as it is used to comapre whether c groups have equal means.

I know the first part of C about the degrees of freedom is right, but don'y understand the last part.

Any help would be great!

With regard to B., the Wikipedia article on ANOVA says:

"Some authors claim that the F-test is unreliable if there are deviations from normality (Lindman, 1974) while others claim that the F-test is robust (Ferguson & Takane, 2005, pp.261-2)."

so it would appear the answer is controversial. (Although maybe the phrase "especially in large samples" is a tip-off that you are expected to say it's robust.)