Experience of Using Bifactor Models to Reduce the Effects of Social Desirability on the Normative Questionnaire of Universal Competencies

Abstract

One of the significant lack of questionnaires is a scores distortion for the measured constructs, associated with the social desirability effects. An even greater threat to the validity of decisions is social desirability in high-stakes evaluation, such as selection for a position. Moreover the issue of the relationship between different components of social desirability and the most frequently measured personal constructs remains debatable. In the material of the author's normative questionnaire of universal competencies, an approach is considered for making adjustments to the final scores for measured constructs using the developed scales of egoistic and moralistic social desirability. Also discussed the prospect of using statement formulations that are neutral to social desirability or express the most positive degree of measured indicators.
 The empirical basis of this study is data gathered within a pilot conducted in the spring of 2022, during which data were obtained from 579 respondents in 49 measurable competencies. The analysis was aimed at assessing the quality of the developed scales of social desirability and modeling of each of the universal competencies scales was carried out with the inclusion of a scale of social desirability. The data were analyzed in the framework of structural modeling - confirmatory factor analysis (CFA) using bifactor models for each of the measured competencies.
 According to the results of this study, the use of the scale of egoistic social desirability as a measure for adjusting factor scores for the competencies has generally satisfactory psychometric statistics, but there is concern about the relatively large measurement error. The paper discusses the advantages and disadvantages of this approach and other practices that are most often used to reduce the effects of social desirability in the academic and business environment.