It’s used for testing for heteroskedasticity. Do we know how to do that?

…did a quick search, and it was mentioned that you do a Chi sq. test (n x R^2), not sure if there is a clear procedure for this or not.

yes. it is level I material, just a little more in depth. the BP chi^2 test detects conditional heterosked. and is a one tailed test. u gotta use the R^2 from a regression of the squared residuals against the independent variable BPstat = n*R^2 and then some other steps…too lazy to type it out

BP test: n*r^2 where r^2 corresponds to the regression of the squared residuals against the independent variables. Chi-Crit = Chi(k) where k=# of independent variables. It is a one-tailed test. If N*R^2 > Chi_Crit - reject the null hypothesis and accept that you have CONDITIONAL HETEROSKEDASTICITY. Correction: Use either White corrected Standard errors. You could also use the Hanson Corrected Std. Errors - which corrects for both Conditional Heteroskedasticity as well as Serial Correlation.

What is r^2? Is it the coefficient of determination (R2)? Or is it the sum of the squared residuals? Is n sample size? I guess I need to dig into CFAI for this one.

n is sample size. and r^2 is the coefficient of determination from the 2nd order regression. this is also present very much in Schweser… you do not need to break out the book for this one. (pg 198-199)

Thanks.