Addressing misclassification in the assessment of self-rated health: insights from the older adult population in Italy

Self-rated health (SRH) is a widely used indicator for assessing the general health of populations, with renewed interest due to ageing societies. However, its subjective nature can introduce measurement errors, leading to biased estimates of health determinants. This study addresses the issue of misclassification in SRH assessments among the older adult population in Italy, using data from the 2019 European Health Interview Survey. We consider an extension of the ordered probit model that accounts for misclassification. A Monte Carlo simulation study shows a net improvement in parameter estimates, which may result in highly biased estimates when misclassification of the dependent variable is ignored. The real data analysis reveals that while the direction of estimated effects is consistent with traditional models, parameter estimates are more accurate and the effect sizes are significantly larger when misclassification is accounted for. In line with previous studies, the results indicate that age, marital status, education, and employment status are significant predictors of SRH, with geographical disparities also playing a role. Additionally, we identify sex-specific patterns in SRH reporting, with women more likely than men to misclassify their health status. The proposed model not only enhances the accuracy of SRH evaluations but also provides valuable insights for policymakers to better target health interventions, ensuring more efficient resource allocation and improved health outcomes. This study highlights the need to refine analytical models to address the complexities of subjective health assessments and suggests potential applications of this approach to other self-assessed health indicators.