There is no such thing as multiple R²; there is R² , and, in a multiple regression, there is multiple R , but there is no multiple R².
In a simple regression, there is a correlation coefficient between the (single) independent variable and the dependent variable: ρ. It can range from -1 to +1.
In a multiple regression, there is no correlation coefficient between the (multiple) independent variables (taken as a group) and the dependent variable: correlation is calculated between exactly two variables.
In a simple regression, R = √R², and ρ = ±R. In a multiple regression, multiple-R = √R², and that’s that; note that multiple-R is never negative.
If anyone tells you that for a multiple regression, ρ = ±√R², they’re wrong.
On the same item set, but Q18, can you tell me why they use 1.96 as the critical t-value in their calculations? You are asked to work out the 95pc CI, but it looks to me like a one sided test so I used a critical t-value of 1.645.
I’m bumping this because this question really threw me off and it’s still being used in the CFA EOC questions. Still question 20, no less.
I’ve combed through both CFAI regression chapters and this term is never presented… we either get Multiple R or R-squared. But yet in the ANOVA table, there it is… Multiple R-squared = 0.36 , and we’re asked to give the most appropriate interpretation of this! How ridicuous.
Edit: didn’t realize how old this was…but yeah, it’s a shame they haven’t fixed this typo (multiple R-squared isn’t a concept in any statistics/econometrics course)…