Consistency of aberrant response behavior: Are misfit persons consistent across two different questionnaires administered at the same time?

Person-fit statistics are important statistics in the modern test theory, which are used to identify respondents with aberrant response behaviors, where, based on statistical evidence, classified as misfit persons. This study aims to determine whether misfit persons are consistent across two different questionnaires that administered simultaneously. The respondents in this study were 723 students in Indonesia. Respondents completed two instruments, namely the 24-item Metacognitive Skills Assessment in Research-Proposal Writing (MSARPW) and the 10-item Rosenberg Self-Esteem Scale (RSES), where the MSARPW was administered in the same Google Form as the RSES. Both instruments were analyzed using the Multidimensional Partial Credit Model (MPCM). The MPCM produced two person fit statistics, which were the focus of this research, namely person Outfit z and Zh statistics. The results of the regression analysis show that the regression coefficients between Zh MSARPW and Zh RSES are 0.375 (p < .001) and 0.424 (p < .001) for Person Outfit z MSARPW and RSES, respectively. To conclude, with a significant regression coefficient, this finding shows that misfit persons in responding to the MSARPW tend to be misfit persons in responding to the RSES, which indicates that the misfit persons are consistent between different measuring instruments administered at the same administration time. Limitations and implications for future research were also noted.