Hi Everyone,
I used regression with HC2 standard errors and am now interested in producing diagnostics. I am specifically interested in predicting Cook's D and DFBETAS.
But Stata's documentation states that:
rstandard, rstudent, cooksd, leverage, dfbeta(), stdf, stdr, covratio, dfits, and welsch are not available if any vce() other than vce(ols) was specified with regress.
Generally, when Stata has this type of restriction it is because the specified estimate is no long valid, and I am assuming that this is the case here.
1. Are Cook's distances invalid when using, say vce(HC2)? If so, why? B-values don't change with VCE options and Cook's distances don't rely on standard error in their calculations.
2. Are the estimates produced with vce(ols) acceptable to use for diagnostics for models that used vce(HC2)?
Cheers,
David.
I used regression with HC2 standard errors and am now interested in producing diagnostics. I am specifically interested in predicting Cook's D and DFBETAS.
But Stata's documentation states that:
rstandard, rstudent, cooksd, leverage, dfbeta(), stdf, stdr, covratio, dfits, and welsch are not available if any vce() other than vce(ols) was specified with regress.
Generally, when Stata has this type of restriction it is because the specified estimate is no long valid, and I am assuming that this is the case here.
1. Are Cook's distances invalid when using, say vce(HC2)? If so, why? B-values don't change with VCE options and Cook's distances don't rely on standard error in their calculations.
2. Are the estimates produced with vce(ols) acceptable to use for diagnostics for models that used vce(HC2)?
Cheers,
David.
Comment