02-13-2014 01:22 AM
While estimating my HCC p_values in SAS I noticed that they differ slightly from those I am estimating in Matlab. I thought I am doing a mistake in Matlab since I am writing the codes myself. However, when I have estimated my regressions on Eviews and obtained my HCC errors I have obtained the same results as those I am obtaining in Matlab. The error are mild like having a p-value of 0.0327 in SAS relative to having 0.0337 in my estimations but I still feel uneasy not to have consistent p-values across the board.
I have used the "HCCMETHOD" options (0 or 1) but I was not able to notice any changes on my SAS p_values. I though my SAS is using the "HCCMETHOD=0" default option when I am not including the "HCCMETHOD" in my options so I have re-estimated my Matlab p_values to be consistent with the original white robust standard errors but I am still not having matching p_values.
Any suggestions would be highly appreciated.