Thanks for your answer. I guess I did not make it clear. I used the data file for refinement which is also used for phasing (peak_anomalous.hkl). Traditionally, people will reprocess the same set of data for refinement (rescale it in hkl2000 without using the option "anomalous", to yield the non_anomalous data set: peak_nonanomalous.hkl ).
So, my question is in refinement which data file should I use? And what's the difference.