“It’s a Trap!” – Faking in multidimensional forced choice response formats: Latent change score modeling of faking across MFC and RS formats

Focus of the present study is the faking susceptibility of rating scale (RS) and multidimensional forced choice (MFC) questionnaires in high-stakes assessments as opposed to low-stakes assessments. Following the transformation of a situation based Big-Five questionnaire (B5PS; Ziegler, 2014) from a RS response format into a MFC questionnaire and a successful Thurstonian IRT (TIRT) calibration, the MFC format version is put to the test in a lab-simulated high-stakes assessment situation. The vulnerability of MFC and RS formats towards intentional response distortion was assessed in a within-design study (N = 187). Test takers were asked to act as being in a fictitious job application process and were encouraged to fake accordingly on both the aforementioned MFC questionnaire and a standard RS Big Five inventory. In the faking condition, participants were asked to fake both response formats according to a predefined job requirement profile. They were elicited to present themselves as ideal candidates in selected Big Five domains, namely Conscientiousness, Emotional Stability, and Agreeableness. Latent Change Score Modelling (LCSM) was then applied to model faking behavior within individuals across honest and faking conditions on both MFC and RS questionnaires. Analyses showed a strong effect of faking in the RS format while the LCSM did not fit the data for the MFC format on any Big-Five domain but Agreeableness (A). Results demonstrate the reduced faking susceptibility of the MFC format compared to the RS format. A dedicated faking variable did not constitute itself in the MFC format while homogenous faking behavior was evident in the RS format. This is discussed in reference to current empirical findings on the susceptibility of MFC formats to faking in high-stakes assessment situations, comparing these results to low-stakes assessment contexts. The specific role of Emotional Stability (N) in MFC questionnaires’ faking susceptibility is highlighted, corroborating findings from previous studies.Keywords: multidimensional forced choice, Latent Change Score modelling, faking, ability to identify criteria