# “Estimated effects may be unbalanced” message when running aov in R. What does it mean?

I am running a summary(aov(...)) in R and I got this message:

Estimated effects may be unbalanced


What does it mean? How may I solve this problem?

## 2 Answers

aov is designed for balanced data (link). Balanced design is: An experimental design where all cells (i.e. treatment combinations) have the same number of observations (link).

• I actually have faced the this is not necessarily true(Correct me if I am wrong). If you have an experiment with two factors. A<-factor(c(1,1,1,1,1,1,2,2,2,2,2,2)) B<-factor(c(1,1,2,2,3,3,1,1,2,2,3,3)) y<-rnorm(12) aov(y~A+B) gives the error prompt. We can see that all combination of levels A and B have equal replications , yet the "Estimated Effects may be unbalanced". I have a feeling that R compares the replication of levels of A with replications of levels of B. – tintinthong Aug 22 '16 at 14:46

Install package called car & activate it first and then calculate sum of squares using Anova(lm(y~x1*x2),type=2)

Doing this way will calculate type II SS which can be used for analysis when the interaction is not significant

and if the interaction is significant for the unbalanced data, you should calculate type III SS