Comparison of U.S. states based on demographic, economic and social indicators

This thesis examines whether U.S. states can be grouped into meaningful clusters when demographic, economic, and social characteristics are analyzed simultaneously. Using data from the American Community Survey and the Centers for Disease Control and Prevention for 2022, a dataset of 30 state-level indicators was compiled and reduced to 23 non-redundant variables through correlation analysis. Principal Component Analysis was applied to the reduced dataset, retaining four components that explained 78.58% of the total variance. These components were used as inputs for K-means cluster analysis, which identified five distinct groups of states. The optimal number of clusters was determined using the silhouette score criterion, and the solution was validated through comparison with Ward's hierarchical clustering, showing 74% agreement after label matching. The five typologies identified are: aging states with moderate socioeconomic conditions; a small heterogeneous group with unique demographic and institutional characteristics; states with lower socioeconomic outcomes concentrated in the South and lower Midwest; high-income and highly urbanized states; and states with strong labor market performance located mainly in the Mountain West and Plains. Comparison with the U.S. Census Bureau's nine geographic divisions showed partial overlap, with some clusters displaying clear geographic concentration while others cut across regional boundaries. The findings suggest that a multidimensional approach can identify structural patterns among U.S. states that are not captured by single-indicator analyses or traditional geographic classifications.