Eta square and effect size in General Linear Model

Question

I have an unbalanced data set that is categorized in 3 different topics. I wanted to see that how these different topics are effecting the data. So I used Generalized Linear Model to investigate this issue. For the significant results of ANOVA test, I dig more to see how is the effect size. So I managed to calculated the Eta square and then by measuring the value for power of two of the Eta square value, we can gain the effect size. My question is that, for example, for a effect size of 20%, we can say that the value of a particular category are dependent to each other like 20%?