Bayesian factor analysis to calculate a deprivation index and its uncertainty

Area of Science:

• Epidemiology
• Spatial statistics
• Public health

Background:

• Traditional deprivation index calculations often neglect spatial dependence and estimation uncertainty.
• These omissions can lead to inaccuracies in epidemiologic studies.
• Spatial factor Bayesian modeling presents a potential solution to these issues.

Purpose of the Study:

• To highlight problems in standard deprivation index calculations.
• To illustrate how spatial factor Bayesian modeling can address these problems.
• To compare Bayesian and non-Bayesian methods for deprivation index estimation.

Main Methods:

• Cross-sectional ecological design analyzing census tracts in three Spanish cities.
• Calculation of a deprivation index using five socioeconomic indicators from the MEDEA project.
• Estimation of the deprivation index via Bayesian factor analysis with hierarchical models accounting for spatial dependence.

Main Results:

• A strong correlation exists between Bayesian and non-Bayesian deprivation indices, but the relationship is non-linear.
• Discrepancies arise when grouping areas by quantiles using different methods.
• Failure to account for index uncertainty can cause misclassification bias in census tracts.

Conclusions:

• Ignoring uncertainty in deprivation index calculations can result in significant misclassification bias.
• This bias may impact subsequent analyses relying on deprivation indices.
• The proposed Bayesian approach offers an improved tool for identifying deprived areas and informing public policy.