POL222H1 Lecture 9: Explaining the Intricacies Behind the Multiple Linear Regression Model
Document Summary
The multiple linear regression model (continued from lecture 8) When confounding variables (z) arise, the multiple linear regression (mlr) model is needed to control for them: to control for z: Divide data by different values of x and z. Calculate the conditional average (average of y) for each value of x and z. The result is how much y varies on average as x changes while holding z constant. The mlr formula: y(hat sign) = + x + z: y(hat sign) = values of y along linear regression lines, note that and z are added/subtracted together before calculating x, z = confounding variable. As z changes, the intercept shifts upwards/downwards. If the coef cients of x and z are positive ( 1 > 0/ >0), a positive z value will raise the intercept, while a negative z value will lower it.
