摘要:Indices of socioeconomic deprivation, which combine a number of variables into a single measure, are often used in
public health and other fields to examine geographic disparities in health outcomes and quality of life. Much of the
research using these indices has been conducted outside the United States, and often focuses heavily on urban areas.
This study uses Principal Component Analysis (PCA) to combine a set of socioeconomic variables for more than
72,000 Census tracts in all 50 U.S. states to construct a set of deprivation indices for the year 2015. These measures
are highly correlated with one another and with measures that use a different weighting scheme. A comparison of our
main index with a simpler measure—tract-level poverty rates—show the two to be highly correlated, but that the
deprivation index value is higher than predicted by poverty alone. This is particularly true when spatial autocorrelation
is incorporated into the model. An analysis of only the 14,000 tracts within the largest cities shows less of a
discrepancy between these two measures, but that spatial autocorrelation is still an issue. Deprivation indices,
therefore, are shown to capture more than just poverty, particularly when geography is taken into account, for both
urban and rural areas.