City Research Online

Abstract

This thesis concerns the development of a practical and theoretical framework for adapting of questionnaires building on van de Vijver and Leung's (1997) Theory of Equivalence and Bias. In contrast to extant research which has largely concentrated on the adaptation of ability measures the present research was operationalised through adapting and translating Orpheus, a work-based Big Five personality questionnaire, into English, Arabic, Chinese (Mandarin) and Spanish. The first phase, 'Quality Control', used a mixed method technique in two studies. Study 1 (Translation and Monitoring) was qualitative and usedforward and back translation followed by dyads and triads. Results from this study (n = 10) reflected the importance of qualitative judgment techniques in test adaptation and showed the emergence of three main types of bias (linguistic, psychological, and conceptual), which were discussed in the literature review but do not constitute part of the Theory of Equivalence and Bias (van de Vijver & Leung, 1997). Study 2 (n = 185) (Pre-Testing) and Study 3 (n = 12) (Cognitive interview) combined quantitative (pre-test) and qualitative techniques (cognitive interviews). Results were inconclusive as to what extent p values or Cohen's d is better at detecting potential problems in adaptation of items. Cognitive interviews were shown to be effective for interpreting statistically significant results as they unravelled many linguistic, psychological, and cultural problems that went unnoticed in back translation dyads/triads. The second phase (,Field Pilot') was laid out over two studies that used the same data but focused on different statistical investigations. Study 4 (n=815) centred on item bias analysis using Logistic Regression as well as ANOVA and showed that 12 items in Arabic, 11 in Chinese and 3 in Spanish were functioning differently than the English version of the items. Study 4 examined the metric equivalence between the four groups using EF A and MG-CF A. Results showed that no model fits the data as it was. Intrinsic test problems and using criterion-related validity as a sole method of validation were identified as two potential causes of model failure.

Publication Type:	Thesis (Doctoral)
Subjects:	B Philosophy. Psychology. Religion > BF Psychology
Departments:	School of Health & Psychological Sciences > Psychology Doctoral Theses School of Health & Psychological Sciences > School of Health & Psychological Sciences Doctoral Theses

Export

Downloads