Assuming normal distribution, one SD is equivalent to an effect size of 1.0, which is associated with a gain of 34 percentage points from the average. From a simple regression model, if past achievement, a score from E to A grade, predicted a change of one SD in current achievement, a standardized score (mean of 0), can I infer that an increase of one letter grade in past achievement (e.g., C to B) can result in a 34 percentage gain in current achievement? Or would it be more correct to say "can explain about 34 percentage of the variance"?
Any comments appreciated. Thanks!