What I want to do is take a correlation coefficient, which I have, and multiply by the average standard deviation to compute a predicted regression towards the mean. I have data points on some variable for both individuals, but I need them to come from two separate population with different means. To make the problem clear, let each score 130 on some measurement. One comes from a population with a mean of 100 and the other from a population with a mean of 85. Let both populations have an SD of 15. One individual scores +2 SD and the other +3. I want an average of how high they score above the mean..