STATS 3 REGRESSION

Question 1

What is the least squares criterion? What does it mean in terms of regression residuals?

Answer 1

The method for finding the line of best fit. Means finding the solution with the least regression residuals.

Question 2

How do you interpret B and beta coefficients?

Answer 2

B coefficients (B0 and B1) are unstandardised and depend on the scales of X and Y
Beta coefficients are standardised (B0 is 0 and B1 is 1)

Question 3

When do you use B and beta coefficients?

Answer 3

B coefficients are used to predict Y from X

Beta coefficients are used to predict Y from X and generalise findings across measurement scales

Question 4

What is the intercept in a standard regression equation? Why?

Answer 4

0 - because measurements are standardised (turned into Z scores), 0 becomes the origin for both variables

Question 5

What kind of variables will any regression equation involve?

Answer 5

One DV (outcome)
A number p of independent variables (predictors)

Question 6

What is multiple R?

Answer 6

The correlation between observed Y and predicted Y

Question 7

What is R square?

Answer 7

the proportion of variance in observed Y accounted for by the multiple regression equation (predicted Y)

Question 8

What is the point of multiple regression?

Answer 8

To find out how much variance in the outcome all predictors explain together (R square) and how much variance each predictor explains alone

Question 9

What is part correlation?

Answer 9

Part correlation looks at the relationship between Y and X1 removing the effects of X2 from X1 and Y

Question 10

What is partial correlation?

Answer 10

Partial correlation looks at the relationship between Y and X1 removing the effects of X2 from X1 but not Y

Question 11

What is the meaning of the squared partial correlation in multiple regression?

Answer 11

Part correlation squared is the amount by which R-square would decrease if the predictor was removed from the regression