Exploring the validity of scores from the Rosenberg Self-Esteem Scale (RSES) in Burundi: A multi-strategy approach

Fromont, Anne [ULB] Haddad, Slim [CHU de Québec] Heinmüller, Rolf [External consultant] Dujardin, Bruno [ULB] Casini, Annalisa [UCL]

The present study aimed to validate the Rosenberg Self-Esteem Scale (RSES) in Burundi, through a multi-strategy approach used in cross-cultural studies. Respondents were 906 health workers (men =56%; caregivers 60%). They responded to a bilingual version of RSES. We utilized Confirmatory Factorial Analysis (CFA) with structural equation modelling and a back translation test to explore the structure of the RSES and the reliability of scores from the scale. Data from an independent were analysed for the reliability of scores assessment. CFA results suggested that the global RSES factor was likely contaminated by a method-effect, mainly associated with negatively worded items. Internal consistencies and a back-translation test demonstrated that the negatively worded items were unsuitable in this context. The independent sample study confirmed poor reliability and internal consistency of scores from for both alternative language versions of the RSES. Our data suggested that an overall cultural effect, rather than a merely specific language effect, may undermine the cross-cultural transportability of the Western scale. Keywords: RSES, Self-Esteem, Negatively Worded Item, Cross-cultural equivalence, Structural Equation Modelling, Burundi