Air pollution exposure inequality across U.S. income and racial groups, 2000–2023: A hybrid monitor–satellite analysis with two new shape diagnostics

Estimates of exposure inequality depend both on where air pollution is measured and on how inequality is summarized from those measurements. Environmental-justice surveillance typically compresses a pollutant’s socioeconomic gradient into one regression slope, the slope index of inequality (SII), defined as the population-weighted slope of group-mean concentration on income rank. A single slope assumes that the concentration–rank profile is a straight line and that exposure and income move together within every income group. The SII still returns a number when either assumption fails, but it no longer flags the misfit. We characterize county-level exposure inequality across the contiguous United States for four criteria pollutants (PM 2.5 , NO 2 , O 3 , SO 2 ) over 2000–2023 (up to 44 states). We add two dimensionless shape diagnostics to the standard slope. The Middle-Group Residual (MGR) is the signed departure of the middle-income stratum from the SII line, and the Within-Stratum Slope Heterogeneity (WSSH) is the spread of within-stratum exposure–income correlations. PM 2.5 exposure uses a population-weighted hybrid of regulatory monitors and satellite-derived concentrations; the other pollutants use the monitor network. The mean income-stratified relative index of inequality (RII) is − 0.048 (PM 2.5 ), + 0.278 (NO 2 ), − 0.018 (O 3 ), and + 0.147 (SO 2 ). The PM 2.5 gradient reverses sign between the monitor-equipped county subset (RII + 0.074 , lower-income more exposed) and the population-representative full-county sample (RII − 0.048 ). This reversal directly reflects the siting of the monitoring network. Non-linear profiles dominate; | MGR | exceeds 0.02 in 50%–92% of state-years and rises monotonically at finer partitions. The SO 2 gradient reversed direction over the period (HAC p = 0.003 ). Its remaining burden along the racial-composition axis falls on lower-income but predominantly White rural counties, so the income and racial axes diverge. A single slope cannot reveal this divergence. Across these four pollutants, the slope alone is therefore an incomplete summary. Its sign can hinge on which counties host monitors, and its straight-line form fails in most state-years. The MGR and WSSH recover the missing shape and coverage information from the same county data the SII already uses.