Lecture 4, 5, 6&7 notes

Question 1

Regression with No Correlation Between IVs

Conclusions & Interpretations

(simulated data)

Answer 1

When IVs Don’t Correlate, β=r

So

The overall regression is statistically significant (F3, 905 = 55.45, p < .01). The results suggest that for every one standard score of increase in time spent on homework out of school, the student’s grades are predicted to increase by .256 standard scores (holding constant parental highest level of education and time spent on homework in school). Also, for every one standard score of increase in parents’ level of education, the student’s grades are predicted to increase by .234 standard scores

(holding constant all other variables in the regression). The results also suggest that two types of homework hold different predictions, with homework completed in school being not a statistically significant predictor of grades. In contrast, time spent on homework outside of school (presumably at home) being a relatively strong (?) positive predictor of grades (after controlling for all other variables in the regression). This prediction was of a similar size to the prediction that parental education had on grades. …….

Question 2

How do you calculate Rsqu?

Answer 2

SS regression/ SS total

OR