Objective. Burnout Assessment Tool (BAT) is gaining momentum as a psychometric instrument, it has a sounding theoretical framework and has already provided good validity evidence via several samples of different countries. BAT is starting to establish itself as the first-choice instrument of several researchers to measure burnout. Whenever a psychometric instrument is used it is desirable that its psychometric properties are assessed. However, when in the presence of small samples that task can be challenging to attain, particularly when using the frequentist framework (e.g., categorical models). Bayesian structural equation modeling (BSEM) represents an alternative to the mainstream (frequentist) structural equation modeling (SEM). BSEM is also particularly useful if the researcher has concerns about model convergence or problematic measured variable distributions. All these potential advantages are due to the specification of priors, priors are the heart of Bayesian statistics. Priors allow the user to incorporate prior knowledge and beliefs into the analysis and establish the hyperparameters. Such prior beliefs/knowledge will be updated by new information (the data), creating the posterior distribution for the model parameters. From the previous research with BAT researchers can extract information to establish their priors and — desirably — obtain more accurate estimates. The objective of this study was to analyze the use of different priors through BSEM in samples of workers from four different countries (Czechia, The Netherlands, South Africa, and Portugal). Method. A cross-sectional study was conducted with multiocupacional samples of workers from different countries (nCzechia = 162; nThe Netherlands = 800; nPortugal = 378; nSouth Africa = 660) from whom BAT’s scores and sociodemographic data was obtained. Prior were defined using information from previous research using BAT. Results and Conclusions. The BAT showed good validity evidence based on the internal (via Bayesian Confirmatory Factor Analysis). BAT’s dimensionality hold, the Bayesian estimates of internal consistency also provided satisfactory evidence. Sensitivity analyses of impact of the defined priors on posterior estimates provided insightful results. BSEM provides a more flexible representation of substantive theory, since it allows to successfully depart from previous research, updating the knowledge by the integration of the new data (even with small samples) and providing a degree of uncertainty for the estimates (i.e., credibility intervals).

--

burnout,measurement invariance,Bayesian structural equation modeling,psychometrics,validity,Portugal,Czechia,South Africa