I am very confused over something that should be very basic….my apologies but I really need some help.
eta square should tell you the percentage of variance in the dependent variable that is predicted by the independent variable. But, if there are 2 sets of variances(one for each of two independent variables—say high school education or college education) which variance is the one that has its variance predicted by the education level?
If you are talking about general linear model ANOVA, there will be an eta square for each IV. The eta square for education level should be listed near the p-value for that IV in the statistical output. It would, of course, be listed in the same row as the variable name representing education level. Don’t confuse it with the R-squared value for the entire model.
Yes, I am talking about GLM ANOVA. Yes, I know that an eta square number is for each IV. What I don’t get is in reading the descriptive material for the DV (say math scores on a placement test). There are means and standard deviations for each IV, so when eta square says that 24% of the variance of the DV is accounted for by education level (high school or college), which variance is it for: the mean and variance associated with the high school or the college group?
If education level is a binary factor (high school vs. college) then the eta square is for the entire education level variable, not just one classification like college or high school. Most GLM ANOVA programs will list the descriptives for both levels of the binary IV but only give one row of data for the p-value, B coefficient, and eta squared variable because the data is redundant – it applies to both levels of the IV. You can see this if you reverse the coding of your binary variable. The eta square will remain the same.