# BP Test

How do you calculate R2 (square) for the Breusch Pagan test? For a single variabe R2 = Square of (co-relation) but what would it be in the case of multiple regression?

R2 for the BP test is the R2 for the regression of the risduals. You calculate it the same way. I doubt you will have to calculate it, most likely it will be given to you.

don’t think you need to know the details of how to calculate… keep in mind this is NOT R^2 of the independent variables from the original regression for the BP test - you regress the residuals from your original regression against the independent variables and use THAT R^2 very likely to be tested…small point - could be worth 1.7% on the exam!

mumkada, regress the residuals from your original regression against the independent variables and use THAT R^2 it still doesnt click thanks !!

You make a NEW regression from the residuals of the original regression. That regression produces its own R^2. Now if that R^2 is HIGH, the residuals are clearly correlated. To make sure you scientifically prove this fact, they’ve made up the BP test.

You know what I just realized? I just realized that BP is a one-tailed test for hetero~~~. I suppose that is a dumb discovery. I assume that you guys all have known this way back… ok, back to books. you know, sometimes looking up the critical value on various test confuses me…

mcpass Wrote: ------------------------------------------------------- > Now if that R^2 is HIGH, the residuals are clearly > correlated. To make sure you scientifically prove > this fact, they’ve made up the BP test. The R^2 obtained in the BP test actually doesnt represent correlation between the residuals, it represents correl between the residuals^2 and the independent variables. I would not worry too much about how to calculate the R^2 because you need a stats program like Stata to predict the residuals. And if you are interested, the test statistic for BP is Chi^2 with k degrees of freedom.