Construct validity and invariance assessment of the social impacts of occupational heat stress scale (SIOHSS) among Ghanaian mining workers

Heat exposure studies over the last decade have shown little attention in assessing and reporting the psychometric properties of the various scales used to measure impacts of occupational heat stress on workers. A descriptive cross-sectional survey including 320 small- and large-scale mining workers was employed to assess the construct validity of the social impacts of occupational heat stress scale (SIOHSS) in the Western Region of Ghana in 2017. A confirmatory factor analysis (CFA) and invariance analysis were carried out using AMOS version 25 and statistical product and service solutions (SPSS) version 26 to examine the model fit and establish consistency correspondingly across multiple groups (gender, age, working hours, type of mining and workplace setting). Empirically, our results depicted that effects on health and safety, behavioural, productivity and social well-being were each found to be reliable, with Cronbach's α of 0.722, 0.807, 0.852 and 0.900 respectively. Though there were issues of insufficient discriminant validity as some average variance extract (AVE) were smaller than the corresponding maximum shared variance (MSV), the CFA showed good model fit indices (CFI = 0.856, GFI = 0.890, TLI = 0.863, SRMR = 0.08, RMSEA = 0.08). Also, the model was variant for all constrained aspects of the structural model indicating a requirement for an adaptation of the instrument across groups. The good to moderate internal consistency and construct validity of the SIOHSS is adequate evidence for the confidence required for its reliability and accuracy in measuring the social impacts of occupational heat stress on workers. [Display omitted] •Psychometric analysis of a perceptual scale for measuring workers' heat stress impact•CFA and invariance test were used to assess the model fit.•The Cronbach's α for the reliability of the instrument was excellent.•The four-domain model recorded good overall fit indices.•The model was variant across gender, age, working hours, work milieu and mining type.