d) Use the R-bar square to decide between models that include more of these additional characteristics. How many variables do you think should the multiple regression model, have to do the best possible job in explaining the variation in income as of 1999? e) Almost 3,000 of the respondents have a bachelor degree, how many have a PhD? What is the premium as of 1999 from having a PhD? 1) One of the great advantages of the NELS is that we have the test scores of the respondents as of the time they were 13. Using the math test scores in our multiple regression model explain the effect on income and the college premium. Is it what you expected? What about the other test scores, like reading, science, etc? Do the same using the income as of 1998. Do you think these tests are a good measure of ability