Hii all Could anyone please explain what covariance in the least square estimators we expect in regression model for the model to be a good fit? Should the estimators i.e. alpha and beta be uncorrelated to form a good model. If yes, why? I have seen a qus. April 2015, qus.11 (ifoa) part (i)(c) of this qus. tells that Cov(à,ß)=0, the estimators are uncorrelated and hence this would be a better model. I didn't get the reason. Please anyone explain. Thanks
The Core Reading does not explain this point. But I think the idea is that if they are uncorrelated then errors in the estimate of one parameter does not affect the other....