You can link to an explained variance applet that demonstrates how changes in the configuration of the scatterplot affect the amount of explained and unexplained variance, and the value of the correlation coefficient.
1. What happens to the value of the correlation coefficient and also, just a square away, the proportion of explained variance – that is, most generally, what happens to the degree of predictability -- as the value of the standard error of estimate increases? Answer
2. What happens to the degree of predictability as the value of the slope, whether positive or negative, approaches zero? Answer
3. What happens to the degree of predictability with decreases in the value of the standard deviation of X? Answer