Comparison of Cluster Analysis and Latent Class Analysis for the Detection of Fake Responses on Personality Tests

Abstract

This study compared the effectiveness of cluster analysis and latent class analysis in detecting fake responses in personality tests. A post-test control group design was employed involving 543 11th-grade students from eight different high schools in Şanlıurfa province during the 2021–2022 academic year. The experimental group was instructed to portray a positive personality profile on the test and provide deceptive responses, as their admission to a university program was contingent on this. Conversely, the control group was asked to represent themselves truthfully and provide honest responses. In this study, the initial focus was on assessing the validity and reliability of the scores obtained from the personality test. Subsequently, a comparison was made between the scores of the participants in the experimental and control groups for each sub-dimension of the personality test to determine if there was a significant difference. The findings revealed a significant difference in mean scores between the two groups, favoring the experimental group. Moreover, the results obtained from Cluster Analysis and Latent Class Analysis demonstrated that Latent Class Analysis outperformed Cluster Analysis in detecting fake respondents, exhibiting a lower error rate.