Longitudinal factorial invariance of the Rosenberg Self-Esteem Scale: Determining the nature of method effects due to item wording

•Longitudinal models do not support the unidimensionality of the Rosenberg’s scale.•Wording effects are not ephemeral artifacts, but stable response styles.•The three-factor structure of the Rosenberg’s scale is free of any gender bias. The present study contributes to the debate on the nature of wording effects (i.e., ephemeral artifacts or stable response styles) associated with the Rosenberg Self-Esteem Scale (RSES) by analysing longitudinal data (5 waves over an 8-year time span) of a non-English version of this scale from a sample of 889 French elderly. Our findings showed, first, that the RSES consists of three factors: a self-esteem factor and two method factors. Second, the measurement properties of this solution of the RSES were consistent over time. Third, the structural parameters as well as individual differences were stable, indicating that the RSES taps three meaningful and enduring individual characteristics, i.e., like personality traits. Fourth, this solution is free of any gender bias.