[en] The two parameters one response model of the generalized linear least squares parameter adjustment methodology was used in order to analyze the procedure of improving cross sections agreement with integral experimental results. Two extreme adjustment cases were considered. One case is using extremely accurate measured integral responses, such as k_eff of Godiva or Jezebel for instance. The other one is modifying only one parameter, for instance the modification of ν-bar in ENDF/B-VII, in order to improve agreement of calculated responses with corresponding integral measurements results. In both cases new posterior parameters correlations are generated. The conclusion is that it is not advised to use integral measurements information in the evaluation process of a general purpose nuclear data file. (authors)