Two widely used psychological scales designed to measure ego-resilience have shown considerable shortcomings when tested in a South African student population, raising questions about their suitability for research and clinical use in the region. The findings have implications for how resilience is assessed in diverse, multilingual, and socioeconomically unequal contexts across Africa and beyond.
The study, published in Psychology International, examined the psychometric properties of the Ego-Resilience Scale (ER89) and its shorter revised version (ER89-R) among 337 undergraduate students at a university in the Western Cape province of South Africa. Researchers used a range of analytical techniques, including classical test theory, confirmatory factor analysis, bifactor modelling, Mokken scale analysis, and Rasch analysis, to evaluate how well the instruments performed.
On the positive side, both scales showed acceptable internal consistency and correlated in the expected directions with related constructs. Higher ego-resilience scores were associated with greater sense of coherence and lower levels of depression and anxiety, suggesting the scales do capture something meaningful about psychological resilience.
However, the broader psychometric picture was far less encouraging. Item correlations were generally low, and most items did not meet the threshold for strong item-total relationships. The average variance extracted fell below recommended levels for both scales, and the pattern of results raised concerns about both convergent and discriminant validity.
The study also tested whether the scales measured a single, unified construct or multiple distinct dimensions. While bifactor models fitted the data better than one-factor models, the specific subscales showed unstable loadings, including some that were negative or nonsignificant. Parallel analysis and Rasch analysis both pointed to the same conclusion: both scales are best understood as measuring a broad general factor of ego-resilience, and their proposed subscales do not hold up as independent, reliable dimensions.
Mokken scale analysis added further cause for concern. Scalability coefficients were weak across both versions, and several items showed serious violations of invariant item ordering, meaning the items did not consistently rank individuals in the same way across the range of the trait being measured. Person separation was also below recommended thresholds, particularly for the revised scale, limiting the instruments’ ability to distinguish between individuals with different levels of ego-resilience.
The researchers noted that the ER89-R did not demonstrate a clear advantage over the longer original version, despite having fewer items. The two scales were very highly correlated with each other, yet shortening the instrument did not resolve the underlying measurement problems.
The study adds to a growing body of evidence calling for local validation of psychological tools imported from Western research traditions before they are used in South African or other non-Western contexts. Constructs such as resilience may be shaped by cultural, linguistic, and social factors that differ substantially from the settings in which these scales were originally developed and validated.
The authors concluded that both scales should be treated as provisional research indicators of general ego-resilience rather than reliable tools for individual-level assessment, screening, or clinical decision-making. Further refinement, including qualitative item review and cross-validation in independent South African samples, is recommended before either instrument can be used with confidence.
