Document Type : Research Paper
Authors
1
Department of Educational Measurement, Faculty of Psychology and Educational Sciences, Allameh Tabataba’i University,Tehran, Iran.
2
Associate Professor, Department of Educational Measurement, Faculty of Psychology and Educational Sciences, Allameh Tabataba’i University, Tehran, Iran.
3
Department of Educational Measurement, Faculty of Psychology and Educational Sciences, Allameh Tabataba’i University, Tehran , Iran
4
Department of Educational Measurement, Faculty of Psychology and Educational Sciences, Allameh Tabataba’i University, Tehran , Iran.
Abstract
The purpose of the present study is to identify the questions in the depression component of the Minnesota Multifaceted Personality Questionnaire-2, which are prejudicial to a specific demographic of job applicants on the basis of gender and age. This research is a survey, and the statistical population consisted of all recruitment candidates in Isfahan city during the 2022 and spring 2023. The one-step cluster method was used to select 5997 individuals, with 57.31% being male and 42.69% being female. The tool used was MMPI-2 questionnaire with 370 questions. Winsteps software version 3.64 was used to analyze the data. The fit of the Items with the model was confirmed using outfit MnSq and infit MnSq, and DIF according to gender and age was checked and compared with Mantel-Haenszel and Rasch model by Winsteps version 3.64. After confirming the fit of the data with the model, Differential Item Functioning was performed. The findings indicated that over half of the items (64.9%) were biased against women and young people. In addition to Rasch’s model, Mantel-Haenszel also found a comparable result, with the exception of two items. The use of depression component questions simultaneously can result in non-bias. However, the use of Harris-Lingoes subscales or selective questions will result in severe DIF that is heavily biased toward women and young people. Additionally, the Rasch model exhibits a relative advantage over the Mantel-Haenszel model in detecting DIF.
Keywords