This study aims was exploring model that explain faking responses on the psychological scale through item response theory modeling (IRT) and tested its goodness of fit. This research use quasi-experimental design that manipulates participant instructions to response Five Factors Personality Scale. In the first instruction, participants were asked to give honest response to given scale, while the second condition were asked to imagine as job applicants so they consider to manipulate their response to give positive impression. IRT based modeling was done on both types of
response by using partial credit model (PCM). Results of analysis suggest whether there is no difference in threshold parameters between the two types of responses. Additional analysiswith differential item functioning (DIF) test found inconsistent results. There are some items affected by DIF that indicated participants who manipulated their response have probability of getting a high score thanothers. Overall, this study shows that the nature invariance of IRT modeling interrupted by faking response.