Lecture 5_Configurations Among Correlations

Question 1

What are the most important determinants of R²?

Answer 1

the correlations of the IVs with the DV

Question 2

What can also effect R²?

Answer 2

the correlation between the IVs

Question 3

What are the 4 configuration that can occur?

Answer 3

1. independence
2. redundancy
3. complementarity
4. suppression
   • redundancy, complementarity, and suppression can be either partial or complete.

Question 4

What is the independence configuration?

Answer 4

the two predictors do not correlate (overlap), but both add something to the prediction of Y (R² = rY1^2 + rY2^2)

Question 5

What is the redundancy configuration?

Answer 5

refers to the fact that the IVs are correlated with one another, thus they duplicate each other’s ability to predict the DV
•Partial redundancy is the standard configuration that we encounter in MR.
•complete redundancy X2 does not add any predictive value (adding X2 to regression does not increase R^2)

Question 6

What is the complementarity configuration?

Answer 6

any situation in which the unique contribution of the IVs exceeds the sum of their individual unique contributions (both IVs have positive zero-order correlations with the DV, but are negatively correlated with one another.)
• rare but interesting

Question 7

What is the suppression configuration?

Answer 7

• most complicated

one IV had 0 correlation with DV, but when added to regression with another IV R² increases

Question 8

The configuration of the correlations involved

in MR influences …

Answer 8

the size of R² for the model.

Question 9

What are the two counter-intuitive configuration results?

Answer 9

complementarity and suppression