More on ANOVA

Answers

1. Unequal sample sizes cause confounding.
   
   
2. Comparing unweighted means eliminates confounding.
   
   
3. Type III sums of squares test an effect while controlling for other effects, thus eliminating confounding.