they are different way to handle the reg model misspecification.  hc3 may be more conservative than robust claim to be better by the authors.  In small sample, it seems better to use hc3.  for large sample, not sure, not sure if make much difference between thes two.

I don't know what do you mean missing p-value.  I saw missing F value before, it is because the F value is too big to display by stata.

thank you!!!太感谢了。