Effect of Item Order on Certain Psychometric Properties: A Demonstration on a Cyberloafing Scale

Creative Commons License

ŞAHİN M. D.

FRONTIERS IN PSYCHOLOGY, cilt.12, 2021 (SSCI) identifier identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 12
  • Basım Tarihi: 2021
  • Doi Numarası: 10.3389/fpsyg.2021.590545
  • Dergi Adı: FRONTIERS IN PSYCHOLOGY
  • Derginin Tarandığı İndeksler: Social Sciences Citation Index (SSCI), Scopus, IBZ Online, Linguistic Bibliography, MLA - Modern Language Association Database, Psycinfo, Directory of Open Access Journals
  • Anahtar Kelimeler: item order effect, self-report questionnaire, modification indices, measurement invariance, cyberloafing, SOCIAL DESIRABILITY BIAS, QUESTION-ORDER, BULLYING VICTIMIZATION, COGNITIVE-PROCESSES, FIT INDEXES, CONTEXT, INVARIANCE, RESPONSES, INTERNET, STUDENTS
  • Anadolu Üniversitesi Adresli: Evet

Özet

Many studies have been conducted on the effect of item order in self-report questionnaires on mean scores. This research aims to study the effect of item order on measurement invariance in addition to mean scores. To this end, two groups randomly obtained from the same sample were presented a fixed order form in which all items belonging to the same dimension were adjacent to each other, and a random order form in which the items were randomly sequenced respectively. The results obtained revealed a statistically significant difference between the mean scores of the two forms. In the next stage of the study, the fit indices obtained from the confirmatory factor analysis (CFA) applied to the two separate forms and the modification indices (MI) suggested by the software were compared. Both forms returned high modification suggestions for adjacent items or items presented near each other. Additionally, it was found that high chi(2) reductions suggested by the MIs in one form resulted in low chi(2) reductions in the other. Lastly, multiple group CFA (mg-CFA) was conducted to determine whether or not measurement invariance was achieved through different item order presentations of the scale. The findings indicate that measurement invariance could not be achieved even at the first stage of analysis. It may specifically be stated that presenting respondents items under the same dimension together ensures empirical findings congruent with theoretical structure.