Chapter 13.6

Air pollution and the vote

Tracts with more diesel exposure voted more Democratic in 2020, by 24.8 points per standard deviation of exposure, but only 4.9 points once race, income, education and density are held fixed.

Author: Vishal Singh
Affiliation: NYU Stern
Published: 30 September 2026

DraftFigures and numbers may change as the analysis is rebuilt.what is still being checked

Still being checked: the release years of the deprivation and asthma values and of the American Community Survey values in the tract atlas, and which archived copy of EJScreen version 2.3 the atlas was built from. The atlas file records none of them.

The tenth of census tracts with the most diesel exhaust in the air had a Democratic two-party margin of +49.1 (95% CI 42.4 to 55.8) points in the 2020 presidential election, against −29.8 (95% CI −32.2 to −27.4) points in the tenth with the least, a gap of 78.9 points. The margin difference per standard deviation of diesel exposure shrinks when tracts are compared on who lives in them: 42% (95% CI 36% to 48%) of the no-controls coefficient remains with race, income and education held fixed, and 20% (95% CI 12% to 27%) once population density is added. The four air measures in the chapter move together, so the data cannot say which one carries the difference that remains. The chapter also compares neighborhood deprivation, modeled asthma prevalence and survey support for EPA enforcement. The asthma figures are modeled estimates and not measured health outcomes, a distinction that the section on asthma explains.

The margin is the Democratic vote minus the Republican vote as a share of the two, so a positive value is a Democratic lead. The votes come from precinct returns that a land-cover model allocated to 2020 census tracts [1], so each tract's margin is an estimate and not a count. Pollution comes from EJSCREEN. Diesel exposure is its modeled annual average concentration of diesel particulate matter in the tract's air, in micrograms per cubic meter, and the other measures are modeled annual averages of fine particles and nitrogen dioxide and a traffic proximity score (vehicles per day on nearby major roads, weighted by distance).The EPA removed EJSCREEN from its website in February 2025. Public Environmental Data Partners hosts a reconstruction of version 2.3, released in July 2024 with air data modeled for 2020 [2]. Of the 82,765 tracts in the atlas, 81,617 have more than 100 estimated votes, and 81,134 of those have every pollution measure, control and land area that the models need. They lie in 47 states and carry 155.1 million votes, and each tract counts in proportion to its votes. The unit is a neighborhood, and the last section returns to what that allows.

The margin rises with diesel exposure

Figure 1 sorts the tracts into ten equal groups by diesel exposure and shows the margin in each, with each row labeled by its decile and median exposure. The point estimates rise with each step. The deciles below the fifth lean Republican and the deciles above it lean Democratic, and the fifth decile, with a median exposure of 0.14 micrograms per cubic meter of diesel particulate matter, is an even split within its interval (+1.4 (95% CI −1.8 to 4.7) points). The intervals are widest in the top two deciles. The tenth-decile tracts sit in 112 counties, against 1,824 counties for the first decile, and the intervals allow for tracts in one county resembling each other.

Figure 1

The Democratic margin rises from −29.8 points in the tenth of tracts with the least diesel exposure to +49.1 points in the tenth with the most

Democratic minus Republican share of the two-party 2020 presidential vote, percentage points; vote-weighted mean by decile of tract diesel particulate exposure, each row labeled with the decile and its median exposure in micrograms per cubic meter; 95% intervals clustered by county

Show data

Data for The Democratic margin rises from −29.8 points in the tenth of tracts with the least diesel exposure to +49.1 points in the tenth with the most
conf_high	conf_low	decile	decile_label	diesel_median	margin	n
−27.42	−32.23	1	1 (0.04 µg/m³)	0.042	−29.82	8,114
−26.24	−31.31	2	2 (0.07)	0.068	−28.78	8,113
−19.39	−24.18	3	3 (0.09)	0.092	−21.78	8,113
−7.48	−13.25	4	4 (0.12)	0.118	−10.36	8,114
4.68	−1.83	5	5 (0.14)	0.144	1.42	8,113
12.45	6.15	6	6 (0.17)	0.171	9.30	8,113
21.73	14.99	7	7 (0.20)	0.203	18.36	8,114
30.07	22.62	8	8 (0.24)	0.244	26.34	8,113
40.11	28.24	9	9 (0.31)	0.311	34.18	8,113
55.84	42.39	10	10 (0.48)	0.477	49.12	8,114

The margin rises with each step from the tenth of tracts with the least diesel exposure to the tenth with the most, and the interval of the fifth decile includes zero. Dots are vote-weighted mean margins and whiskers are 95% intervals that allow tracts in one county to resemble each other.Note: Deciles of diesel exposure hold equal numbers of tracts; ties are broken by census tract code.Note: Tracts with 100 or fewer votes are left out. Intervals treat counties as the units that vary. They do not include error in the allocation of precinct votes to tracts or in the modeled pollution values.Source: Precinct votes allocated to census tracts (Fekrazad 2025), 2020 presidential election; EPA EJScreen version 2.3 (2024 release, 2020 air data); American Community Survey tract data (vintage not recorded in the atlas); vote-weighted, full counts of tracts, no sampling weights

The tenth decile differs from the first in more than the air (Figure 2). The white-alone share of residents is 36% in the tenth decile against 79% in the first, the Hispanic share is 33% against 9%, and the median tract holds 5,167 people per square kilometer against 14. Income peaks in decile 6 and the college share is nearly flat from there. Diesel exposure moves with density, and across tracts the correlation of exposure with the log of density is +0.68 (95% CI 0.66 to 0.71). Holding these traits fixed means comparing tracts that resemble each other on them.

The Area Deprivation Index ranks neighborhoods from the least to the most deprived [3], and its mean national rank falls from 62.5 (95% CI 60.9 to 64.2) in the first decile of diesel exposure to 22.2 (95% CI 16.2 to 28.2) in the tenth, against 44.8 (95% CI 42.2 to 47.4) for all tracts weighted by votes. The index is built from income, education, employment and housing measures, so it overlaps with the controls in the vote models. Chapter 13.3, Deprivation and mobility (planned) asks how deprivation itself relates to the vote, and Chapter 13.1, Measures of hardship compared sets several hardship measures side by side.

Figure 2

Higher-diesel tracts are denser, less white and more Hispanic, and income peaks at decile 6 while college share stays nearly flat after it

Characteristics of the tracts in each decile of diesel particulate exposure (8,113 or 8,114 tracts per decile): shares of residents (percent), income (dollars) and deprivation rank averaged by votes, density and exposure as medians

Decile of diesel exposure	Diesel PM, median (µg/m³)	Democratic margin (points)	White alone (%)	Black alone (%)	Hispanic (%)	College degree (%)	Median household income	Deprivation rank (100 = most)	People per km², median
1 (least)	0.04	−29.8	79	5	9	25	$70k	62.5	14
2	0.07	−28.8	78	7	9	28	$78k	59.1	44
3	0.09	−21.8	77	7	10	32	$87k	52.2	152
4	0.12	−10.4	72	8	11	36	$95k	47.1	427
5	0.14	+1.4	67	11	13	40	$98k	45.0	821
6	0.17	+9.3	62	13	15	41	$102k	42.7	1,080
7	0.20	+18.4	57	14	17	42	$101k	41.4	1,394
8	0.24	+26.3	53	15	19	42	$100k	39.4	1,669
9	0.31	+34.2	47	14	25	42	$98k	34.4	2,354
10 (most)	0.48	+49.1	36	13	33	42	$97k	22.2	5,167

Show data

Data for Higher-diesel tracts are denser, less white and more Hispanic, and income peaks at decile 6 while college share stays nearly flat after it
adi	black	college	decile_label	density_median	diesel_median	hispanic	income	margin	n	white
62.53	5.35	25.50	1 (least)	13.95	0.042	9.39	69.75	−29.82	8,114	78.60
59.12	7.10	27.58	2	44.19	0.068	8.89	78.14	−28.78	8,113	78.46
52.19	7.03	32.12	3	152.3	0.092	10.05	87.46	−21.78	8,113	76.60
47.10	8.38	36.50	4	426.8	0.118	11.08	94.75	−10.36	8,114	72.38
45.03	10.72	39.56	5	821.2	0.144	12.88	98.50	1.42	8,113	66.65
42.66	12.75	41.49	6	1079.7	0.171	14.56	102.1	9.30	8,113	61.82
41.40	13.51	41.93	7	1394.1	0.203	17.16	100.6	18.36	8,114	56.66
39.38	15.18	42.40	8	1669.1	0.244	18.69	99.85	26.34	8,113	52.93
34.44	14.46	42.20	9	2354.4	0.311	24.51	97.56	34.18	8,113	47.12
22.21	12.75	42.07	10 (most)	5167.4	0.477	33.43	96.62	49.12	8,114	35.99

Density and the Democratic margin rise and the white share falls from each decile to the next, while income peaks in decile 6 and the college share changes little after it. Each row is a decile of tracts by diesel exposure; shares, income and deprivation rank are vote-weighted means of tract values, and density and exposure are medians.Note: Deciles of diesel exposure hold equal numbers of tracts; ties are broken by census tract code.Note: Shares, incomes and ranks are means of tract values weighted by votes. They describe the tracts in each decile, so the table shows no interval.Note: The atlas labels the white share as white alone. Together the white, Black and Hispanic shares never exceed a tract's whole population, so the white share appears to leave out Hispanic residents, but the atlas file does not say so.Note: The Area Deprivation Index has no rank for a small number of tracts, which are left out of that column.Source: Precinct votes allocated to census tracts (Fekrazad 2025), 2020 presidential election; EPA EJScreen version 2.3 (2024 release, 2020 air data); American Community Survey tract data (vintage not recorded in the atlas); vote-weighted, full counts of tracts, no sampling weights; Area Deprivation Index national rank (Neighborhood Atlas; release year not recorded in the atlas file)

Controls shrink every gap, and fine particles end near zero

Figure 3 regresses the tract margin on each exposure. Each exposure is standardized, so a coefficient is the difference in the margin, in points, between tracts one standard deviation apart in exposure. One standard deviation of diesel exposure is 0.14 micrograms per cubic meter. For the other exposures it is 3.5 parts per billion of nitrogen dioxide, 2.1 micrograms per cubic meter of fine particles and 2.4 million on the traffic proximity score. Tracts count in proportion to their votes, and the intervals are clustered by county, so they allow tracts in one county to resemble each other.

The diesel coefficient is 24.8 (95% CI 21.5 to 28.0) points with no controls and 10.3 (95% CI 8.1 to 12.6) points with the white, Black and Hispanic shares of residents, the college share and median household income held fixed. Adding the log of population density lowers it further, to 4.9 (95% CI 2.6 to 7.2) points. Traffic proximity goes from 24.7 (95% CI 19.2 to 30.2) points to 12.2 (95% CI 9.0 to 15.3) with the demographic controls and 8.7 (95% CI 6.1 to 11.3) with density added. Nitrogen dioxide goes from 24.3 (95% CI 22.3 to 26.3) points to 10.1 (95% CI 8.3 to 11.9) and then 3.6 (95% CI 1.5 to 5.8). Fine particles are the exception, with a coefficient of 8.8 (95% CI 5.3 to 12.2) points with no controls, 0.9 (95% CI −1.4 to 3.1) with the demographic controls and −0.7 (95% CI −2.8 to 1.4) with density, so once tracts are compared on who lives in them the fine-particle interval includes zero.

The four exposures are correlated, so the first four rows of each panel do not say which one matters. Diesel exposure moves with traffic proximity and nitrogen dioxide (correlations of +0.68 and +0.72 across tracts). In the joint row, where all four enter one model with the controls and density, only traffic proximity has an interval above zero: it keeps 8.7 (95% CI 5.6 to 11.7) points while diesel falls to 1.3, an interval that includes zero, so the data cannot say which exposure carries the difference.

Figure 3

Controls shrink the Democratic gap for every pollutant, and with all four exposures in one model traffic proximity keeps 8.7 points while diesel falls to 1.3

Difference in the Democratic two-party margin (percentage points) per standard deviation of tract exposure, by pollutant and set of controls; 81,134 tracts; 95% intervals clustered by county

No controls
Race, income, education
Plus density
All four together
County fixed effects

Diesel particulate matter

Traffic proximity

Nitrogen dioxide

Fine particles (PM2.5)

Margin per SD of exposure (points)

Show data

Data for Controls shrink the Democratic gap for every pollutant, and with all four exposures in one model traffic proximity keeps 8.7 points while diesel falls to 1.3
conf_high	conf_low	estimate	n	pollutant	specification
27.99	21.52	24.75	81,134	Diesel particulate matter	No controls
12.56	8.11	10.34	81,134	Diesel particulate matter	Race, income, education
7.21	2.60	4.90	81,134	Diesel particulate matter	Plus density
4.64	−2.05	1.29	81,134	Diesel particulate matter	All four together
8.89	5.26	7.07	81,134	Diesel particulate matter	County fixed effects
30.22	19.19	24.70	81,134	Traffic proximity	No controls
15.35	9.04	12.20	81,134	Traffic proximity	Race, income, education
11.29	6.08	8.69	81,134	Traffic proximity	Plus density
11.72	5.59	8.65	81,134	Traffic proximity	All four together
8.56	4.64	6.60	81,134	Traffic proximity	County fixed effects
26.28	22.35	24.31	81,134	Nitrogen dioxide	No controls
11.86	8.29	10.07	81,134	Nitrogen dioxide	Race, income, education
5.80	1.46	3.63	81,134	Nitrogen dioxide	Plus density
1.85	−3.40	−0.777	81,134	Nitrogen dioxide	All four together
7.80	5.16	6.48	81,134	Nitrogen dioxide	County fixed effects
12.24	5.26	8.75	81,134	Fine particles (PM2.5)	No controls
3.09	−1.37	0.859	81,134	Fine particles (PM2.5)	Race, income, education
1.36	−2.80	−0.721	81,134	Fine particles (PM2.5)	Plus density
0.309	−3.76	−1.72	81,134	Fine particles (PM2.5)	All four together
2.53	−2.43	0.050	81,134	Fine particles (PM2.5)	County fixed effects

Diesel, traffic proximity and nitrogen dioxide have intervals above zero in every row except the joint model, and fine particles have an interval that includes zero once controls are added. In the joint row only traffic proximity has an interval above zero. Each panel is one pollutant and each dot is the coefficient from a vote-weighted regression of the tract margin on the standardized exposure, with the controls named on the row; in the joint row the coefficient is that pollutant's in a model with all four. One standard deviation of diesel exposure is 0.14 micrograms per cubic meter.Note: Controls are the white, Black and Hispanic shares of residents, the college share and median household income; density is the log of people per square kilometer. The joint row enters all four exposures in one model with the controls and density.Note: The county fixed effects row is a different comparison and not a further control: it uses the same controls and density and compares tracts only within their own county.Note: Intervals treat counties as the units that vary. They do not include error in the allocation of precinct votes to tracts or in the modeled pollution values.Source: Precinct votes allocated to census tracts (Fekrazad 2025), 2020 presidential election; EPA EJScreen version 2.3 (2024 release, 2020 air data); American Community Survey tract data (vintage not recorded in the atlas); vote-weighted, full counts of tracts, no sampling weights

Population density is closely tied to the Democratic vote across American places [4]. Entering density as 20 equal-size bins in place of the log term gives a diesel coefficient of 3.4 (95% CI 1.1 to 5.8) points. Clustering the intervals by state in place of county gives 4.9 (95% CI 0.2 to 9.6) points for the diesel coefficient with density, and the lower bound is near zero.

Comparing tracts inside the same county, with a fixed effect for each county and the same controls and density, gives 7.1 (95% CI 5.3 to 8.9) points for diesel, 6.5 (95% CI 5.2 to 7.8) points for nitrogen dioxide and 6.6 (95% CI 4.6 to 8.6) points for traffic. The point estimates for diesel and nitrogen dioxide are larger than the pooled ones with density, and the point estimate for traffic is smaller.The within-county coefficient minus the pooled coefficient with density is +2.2 (95% CI −0.4 to 5.3) points for diesel, an interval that includes zero, +2.8 (95% CI 0.5 to 5.3) points for nitrogen dioxide and −2.1 (95% CI −4.4 to −0.1) points for traffic. A within-county comparison removes every difference between counties, including the urban-rural contrast, but it is not a comparison of neighbors. Tracts in one county range from dense cores to suburbs, which is why density stays in the model. Chapter 12.6, Suburbs and density (planned) takes up the density gradient in the vote directly.

In counties, the link depends on weighting

Figure 4 moves the question to counties. A tract model of the margin on race, income and education, with no pollution term, predicts each tract's margin, and the residual margin is the actual margin minus that prediction. The map averages the residual over each county's tracts by votes, which smooths the noise in single tract estimates and lets the map use county shapes. Diesel exposure for a county is the same vote-weighted average of its tract values.

Figure 4

The residual margin after race, income and education averages −15.9 points in southern counties and +14.5 points in northeastern counties, weighted by votes

Residual margin: tract Democratic two-party margin minus the margin predicted from tract race, income and education, averaged over each county's tracts by votes, percentage points; county diesel exposure in micrograms per cubic meter

Loading county values

Residual margin after race, income and education (percentage points)

More RepublicanMore Democratic

Values outside −40 to +40 take the end colors.

No data (3,143 counties)

Select a class to show only those places.

Find a county

Hover or tap a county for its value, or search by name. Zoom with the buttons on the map, then drag to move it.

Alaska draws as no data. Its election returns are reported by election district, and the district codes cannot be matched to boroughs and census areas.

Red counties voted more Republican than race, income and education predict, and blue counties more Democratic, across 3,096 counties. Switch views to hold density fixed as well, or to see diesel exposure, which is highest in New York, Kings, Bronx and Queens Counties in New York City and in Hudson County, New Jersey, where county means run from 0.56 to 0.75 micrograms per cubic meter.Note: Hawaii, Connecticut and the District of Columbia draw as no data (Hawaii has no tract vote estimates, the tract atlas has no rows for Connecticut, and it has no white share for DC tracts), and so does any county without a tract of more than 100 votes.Note: County values are averages of modeled tract values, so they carry the errors of both models. No interval is shown on the map.Source: Precinct votes allocated to census tracts (Fekrazad 2025), 2020 presidential election; EPA EJScreen version 2.3 (2024 release, 2020 air data); American Community Survey tract data (vintage not recorded in the atlas); vote-weighted county means of tract values, full counts, no sampling weights

The residual margin has a strong regional component. Counties in the South average −15.9 points and counties in the Northeast +14.5 points, and 61% of the variance of the residual, weighted by county votes, lies between states.

Figure 5 plots each county's diesel exposure against its residual margin. Across counties weighted by their votes, the residual correlates with county diesel exposure at +0.26 (95% CI 0.17 to 0.34), so counties with more exposure vote more Democratic than the demographic model predicts. Counting each county once, the correlation is +0.27 (95% CI 0.25 to 0.30). When density also enters the model that produces the residual, the vote-weighted correlation is +0.08 (95% CI −0.02 to 0.17), which includes zero, and the unweighted correlation is −0.06 (95% CI −0.09 to −0.03), which does not, so the sign of the density-adjusted correlation depends on whether large counties count more. Within states, the vote-weighted correlation is +0.18 (95% CI 0.01 to 0.41) net of race, income and education and −0.07 (95% CI −0.21 to 0.16) with density added, with intervals that resample the states.

Figure 5

County diesel exposure correlates at +0.26 with the residual margin after race, income and education, and at +0.08 (interval includes zero) with density added

Each dot is a county: its mean diesel exposure (averaged over tracts by 2020 votes) against its residual margin, the Democratic two-party margin minus the margin predicted from race, income and education, in percentage points; correlations across 3,096 counties weighted by county votes

Dot area: Votes for president, 2020430,0001,500,0004,300,000

Show data

Data for County diesel exposure correlates at +0.26 with the residual margin after race, income and education, and at +0.08 (interval includes zero) with density added
County	Diesel exposure (µg/m³)	Residual margin (points)	Residual margin, density held fixed (points)	Votes for president, 2020
Autauga, AL	0.099	−37.1	−34.3	27,770
Baldwin, AL	0.079	−39.1	−36.2	109,679
Barbour, AL	0.048	−21.7	−4.6	10,518
Bibb, AL	0.061	−31.6	−21.9	9,595
Blount, AL	0.078	−38.5	−33.6	27,588
Bullock, AL	0.040	+10.2	+37.9	4,613
Butler, AL	0.059	−25.1	−7.6	9,488
Calhoun, AL	0.091	−25.8	−24.3	50,980
Chambers, AL	0.080	−22.0	−11.1	15,284
Cherokee, AL	0.073	−29.5	−22.0	12,301
Chilton, AL	0.075	−36.1	−29.0	19,310
Choctaw, AL	0.034	−15.7	+7.6	7,464
Clarke, AL	0.041	−22.4	−1.5	13,135
Clay, AL	0.052	−32.7	−17.4	6,930
Cleburne, AL	0.077	−38.4	−26.5	7,227
Coffee, AL	0.048	−37.5	−31.8	22,275
Colbert, AL	0.107	−17.4	−17.3	27,886
Conecuh, AL	0.047	−16.9	+7.3	6,441
Coosa, AL	0.048	−21.2	−1.1	5,479
Covington, AL	0.039	−36.8	−27.5	17,430
Crenshaw, AL	0.035	−33.3	−14.9	6,617
Cullman, AL	0.090	−36.4	−34.6	41,851
Dale, AL	0.044	−33.3	−25.6	19,738
Dallas, AL	0.067	−10.5	+4.6	17,864
DeKalb, AL	0.073	−34.6	−27.6	29,356
Elmore, AL	0.072	−34.4	−29.9	41,030
Escambia, AL	0.054	−29.5	−16.8	15,910
Etowah, AL	0.097	−24.6	−25.1	47,728
Fayette, AL	0.051	−34.1	−20.8	8,766
Franklin, AL	0.057	−36.0	−25.9	12,578
Geneva, AL	0.043	−39.6	−29.5	12,545
Greene, AL	0.059	+5.7	+37.6	4,775
Hale, AL	0.051	−19.7	+3.5	6,772
Henry, AL	0.040	−34.9	−19.5	9,298
Houston, AL	0.070	−38.1	−36.1	46,173
Jackson, AL	0.073	−25.9	−19.6	23,636
Jefferson, AL	0.194	−18.3	−15.8	325,762
Lamar, AL	0.047	−31.6	−18.7	7,193
Lauderdale, AL	0.085	−22.8	−21.7	44,339
Lawrence, AL	0.073	−27.2	−17.2	16,031
Lee, AL	0.087	−37.8	−31.5	71,065
Limestone, AL	0.112	−29.6	−25.6	49,235
Lowndes, AL	0.057	−2.4	+27.9	6,835
Macon, AL	0.064	+0.7	+25.9	8,723
Madison, AL	0.118	−26.4	−23.0	194,767
Marengo, AL	0.048	−24.2	−0.6	10,900
Marion, AL	0.061	−31.4	−23.7	13,807
Marshall, AL	0.076	−41.1	−37.6	39,655
Mobile, AL	0.150	−24.0	−24.2	181,166
Monroe, AL	0.040	−17.7	−0.4	10,668
Montgomery, AL	0.145	−24.7	−18.6	98,448
Morgan, AL	0.124	−31.2	−30.8	53,722
Perry, AL	0.043	+0.8	+31.8	5,230
Pickens, AL	0.046	−21.4	−0.2	9,668
Pike, AL	0.047	−36.5	−20.2	13,841
Randolph, AL	0.059	−42.5	−27.8	10,837
Russell, AL	0.090	−10.3	−3.5	21,329
St. Clair, AL	0.109	−30.2	−29.8	44,441
Shelby, AL	0.128	−46.2	−42.5	114,950
Sumter, AL	0.058	−3.8	+28.8	6,291
Talladega, AL	0.080	−19.6	−13.1	35,663
Tallapoosa, AL	0.054	−31.2	−19.7	20,991
Tuscaloosa, AL	0.118	−25.9	−22.1	89,347
Walker, AL	0.087	−28.7	−25.0	31,170
Washington, AL	0.038	−34.1	−13.0	8,876
Wilcox, AL	0.039	−2.4	+26.4	5,903
Winston, AL	0.056	−34.7	−26.9	11,284
Apache, AZ	0.016	+16.7	+54.6	35,171
Cochise, AZ	0.049	−19.7	−5.7	60,391
Coconino, AZ	0.072	+9.6	+27.7	73,274
Gila, AZ	0.038	−11.5	+4.0	27,663
Graham, AZ	0.039	−28.7	−12.5	14,995
Greenlee, AZ	0.019	−35.2	+1.5	3,685
La Paz, AZ	0.071	−17.9	+5.1	6,546
Maricopa, AZ	0.256	−3.0	−7.5	2,067,902
Mohave, AZ	0.074	−18.1	−20.4	104,667
Navajo, AZ	0.040	−6.5	+17.2	51,767
Pima, AZ	0.159	+7.1	+10.1	519,220
Pinal, AZ	0.131	−7.2	−6.6	184,398
Santa Cruz, AZ	0.048	+4.7	+30.6	19,556
Yavapai, AZ	0.073	−10.5	−6.0	143,219
Yuma, AZ	0.162	−14.3	−10.5	69,651
Arkansas, AR	0.074	−21.5	−10.7	6,292
Ashley, AR	0.050	−29.0	−16.4	7,926
Baxter, AR	0.061	−14.0	−11.2	21,007
Benton, AR	0.090	−25.8	−22.0	119,903
Boone, AR	0.051	−16.3	−13.4	17,114
Bradley, AR	0.043	−31.9	−9.7	3,654
Calhoun, AR	0.040	−32.4	−6.3	2,182
Carroll, AR	0.042	−5.9	+7.9	11,797
Chicot, AR	0.062	−22.2	−0.3	4,103
Clark, AR	0.086	−14.0	+0.5	8,395
Clay, AR	0.091	−14.6	−8.8	5,183
Cleburne, AR	0.052	−24.0	−17.7	12,680
Cleveland, AR	0.045	−31.3	−12.4	3,600
Columbia, AR	0.056	−36.9	−22.8	8,616
Conway, AR	0.100	−10.1	+2.0	8,685
Craighead, AR	0.125	−26.5	−24.9	38,511
Crawford, AR	0.117	−25.3	−22.8	24,093
Crittenden, AR	0.180	−13.8	−8.3	16,367
Cross, AR	0.106	−24.5	−16.8	6,942
Dallas, AR	0.051	−23.5	−4.5	2,649
Desha, AR	0.070	−19.0	+4.3	4,164
Drew, AR	0.055	−30.4	−12.4	6,908
Faulkner, AR	0.106	−21.5	−19.0	54,428
Franklin, AR	0.073	−20.0	−9.8	7,129
Fulton, AR	0.053	−14.3	+0.2	5,119
Garland, AR	0.062	−13.9	−11.6	44,199
Grant, AR	0.054	−26.7	−18.2	8,191
Greene, AR	0.103	−17.5	−19.6	16,100
Hempstead, AR	0.078	−32.9	−16.5	6,845
Hot Spring, AR	0.090	−16.7	−9.9	12,557
Howard, AR	0.059	−34.2	−17.6	4,837
Independence, AR	0.072	−22.4	−14.5	14,512
Izard, AR	0.042	−27.7	−12.4	5,810
Jackson, AR	0.100	−18.4	−7.5	5,091
Jefferson, AR	0.093	−10.8	−1.1	24,796
Johnson, AR	0.084	−20.2	−9.5	9,541
Lafayette, AR	0.047	−31.1	−10.8	2,679
Lawrence, AR	0.101	−14.6	−8.4	5,857
Lee, AR	0.076	−17.7	+1.1	2,848
Lincoln, AR	0.054	−21.9	−6.7	3,875
Little River, AR	0.066	−35.4	−18.2	5,177
Logan, AR	0.048	−17.5	−9.5	8,182
Lonoke, AR	0.098	−22.0	−21.4	30,699
Madison, AR	0.037	−22.6	−5.7	7,351
Marion, AR	0.041	−12.4	−2.0	7,503
Miller, AR	0.118	−30.3	−28.4	16,527
Mississippi, AR	0.102	−19.7	−10.7	12,342
Monroe, AR	0.091	−23.3	+1.7	2,816
Montgomery, AR	0.035	−19.8	−0.8	3,873
Nevada, AR	0.065	−24.2	−3.2	3,358
Newton, AR	0.030	−15.7	+2.9	3,998
Ouachita, AR	0.109	−19.0	−5.1	9,629
Perry, AR	0.054	−11.5	+1.7	4,627
Phillips, AR	0.076	−12.0	+6.2	6,242
Pike, AR	0.046	−36.5	−18.0	4,246
Poinsett, AR	0.101	−20.0	−15.3	7,521
Polk, AR	0.039	−29.8	−17.6	8,490
Pope, AR	0.100	−25.2	−21.2	24,432
Prairie, AR	0.085	−28.7	−9.5	3,495
Pulaski, AR	0.154	−6.9	−4.6	169,926
Randolph, AR	0.076	−17.9	−10.8	6,812
St. Francis, AR	0.123	−14.4	−0.3	7,108
Saline, AR	0.100	−21.7	−21.5	56,960
Scott, AR	0.036	−28.1	−13.0	3,551
Searcy, AR	0.036	−27.0	−9.3	4,019
Sebastian, AR	0.092	−20.8	−20.8	47,138
Sevier, AR	0.058	−35.9	−19.2	5,202
Sharp, AR	0.059	−13.6	−5.3	7,566
Stone, AR	0.039	−17.1	−3.5	5,938
Union, AR	0.057	−29.6	−18.1	16,607
Van Buren, AR	0.042	−14.1	−1.9	7,807
Washington, AR	0.092	−3.1	+0.5	94,275
White, AR	0.083	−26.5	−22.1	30,885
Woodruff, AR	0.086	−16.2	+7.1	2,476
Yell, AR	0.056	−27.7	−13.0	6,741
Alameda, CA	0.250	+21.4	+21.4	767,729
Alpine, CA	0.012	+26.8	+71.0	741
Amador, CA	0.052	+2.6	+11.5	22,308
Butte, CA	0.133	+8.5	+10.7	102,023
Calaveras, CA	0.049	+4.6	+16.4	27,159
Colusa, CA	0.078	−20.1	+8.3	7,962
Contra Costa, CA	0.161	+22.9	+23.3	581,271
Del Norte, CA	0.024	−0.7	+7.4	11,452
El Dorado, CA	0.067	+3.4	+6.8	115,597
Fresno, CA	0.369	−9.6	−6.9	364,089
Glenn, CA	0.100	−16.9	−7.7	11,296
Humboldt, CA	0.054	+42.2	+48.7	68,824
Imperial, CA	0.180	−4.7	+7.2	55,690
Inyo, CA	0.021	+0.6	+25.6	9,480
Kern, CA	0.318	−17.1	−15.6	305,129
Kings, CA	0.268	−19.2	−16.6	43,863
Lake, CA	0.031	+28.5	+35.5	28,808
Lassen, CA	0.020	−25.9	−9.4	11,980
Los Angeles, CA	0.401	+12.5	+10.1	4,256,459
Madera, CA	0.225	−13.6	−1.8	53,736
Marin, CA	0.077	+54.1	+54.9	155,764
Mariposa, CA	0.030	−0.5	+24.0	10,277
Mendocino, CA	0.031	+47.4	+61.9	43,333
Merced, CA	0.302	−0.9	+5.1	90,569
Modoc, CA	0.015	−23.4	+2.3	4,338
Mono, CA	0.014	+27.3	+57.3	6,737
Monterey, CA	0.093	+25.7	+30.6	163,789
Napa, CA	0.101	+32.4	+34.7	72,042
Nevada, CA	0.058	+28.7	+34.7	64,806
Orange, CA	0.316	−13.3	−16.2	1,521,670
Placer, CA	0.211	−3.9	−5.4	234,940
Plumas, CA	0.022	+17.6	+37.8	11,256
Riverside, CA	0.271	−4.1	−3.9	997,214
Sacramento, CA	0.234	+10.4	+6.0	716,929
San Benito, CA	0.100	+19.1	+25.4	28,876
San Bernardino, CA	0.353	−7.1	−5.6	840,486
San Diego, CA	0.234	+2.9	+1.3	1,601,471
San Francisco, CA	0.307	+33.3	+25.2	439,813
San Joaquin, CA	0.265	+3.3	+1.6	288,024
San Luis Obispo, CA	0.070	+13.8	+17.3	159,755
San Mateo, CA	0.125	+30.3	+29.4	373,511
Santa Barbara, CA	0.116	+20.0	+20.6	200,147
Santa Clara, CA	0.202	+10.1	+11.3	849,568
Santa Cruz, CA	0.078	+54.1	+54.6	143,698
Shasta, CA	0.090	−7.5	−8.4	92,846
Sierra, CA	0.020	+11.3	+42.4	1,930
Siskiyou, CA	0.029	+5.2	+25.2	23,439
Solano, CA	0.151	+15.3	+15.2	205,888
Sonoma, CA	0.095	+52.5	+53.1	268,261
Stanislaus, CA	0.269	−0.2	−2.9	214,859
Sutter, CA	0.151	−15.4	−13.3	42,604
Tehama, CA	0.098	−16.0	−4.3	28,787
Trinity, CA	0.010	+26.5	+52.9	6,262
Tulare, CA	0.312	−16.4	−11.5	146,843
Tuolumne, CA	0.040	+8.3	+15.4	30,395
Ventura, CA	0.165	+11.7	+10.8	422,731
Yolo, CA	0.190	+12.7	+18.2	96,830
Yuba, CA	0.143	−9.6	−5.7	29,816
Adams, CO	0.312	+15.0	+11.1	236,740
Alamosa, CO	0.050	−13.7	+6.6	7,808
Arapahoe, CO	0.268	+9.9	+5.7	350,172
Archuleta, CO	0.015	−10.8	+12.4	9,144
Baca, CO	0.022	−45.1	−8.9	2,225
Bent, CO	0.036	−21.5	+13.9	2,274
Boulder, CO	0.122	+38.6	+41.4	206,111
Broomfield, CO	0.238	+18.7	+15.8	46,632
Chaffee, CO	0.016	+10.8	+29.8	13,718
Cheyenne, CO	0.020	−51.0	−10.4	1,136
Clear Creek, CO	0.025	+13.0	+37.8	6,548
Conejos, CO	0.014	−14.2	+24.9	4,333
Costilla, CO	0.010	+5.3	+54.4	2,094
Crowley, CO	0.030	−34.5	−8.2	1,750
Custer, CO	0.009	−14.1	+13.6	3,635
Delta, CO	0.029	−13.8	+2.5	19,370
Denver, CO	0.462	+32.4	+28.7	393,776
Dolores, CO	0.011	−26.3	+8.8	1,449
Douglas, CO	0.166	−12.9	−13.0	231,605
Eagle, CO	0.029	+14.9	+38.6	26,745
Elbert, CO	0.048	−22.8	−9.3	18,983
El Paso, CO	0.117	−13.2	−15.9	378,813
Fremont, CO	0.026	−11.6	−5.1	25,557
Garfield, CO	0.041	+5.8	+24.5	30,904
Gilpin, CO	0.021	+15.1	+36.1	4,186
Grand, CO	0.017	+7.0	+33.4	9,870
Gunnison, CO	0.018	+20.3	+46.2	11,190
Hinsdale, CO	0.005	−13.8	+33.5	632
Huerfano, CO	0.022	+5.5	+35.4	4,406
Jackson, CO	0.013	−37.8	+4.6	876
Jefferson, CO	0.185	+18.0	+14.2	376,131
Kiowa, CO	0.022	−41.1	−2.2	903
Kit Carson, CO	0.026	−35.8	−4.6	3,871
Lake, CO	0.015	+5.3	+25.4	3,961
La Plata, CO	0.030	+15.7	+32.3	35,667
Larimer, CO	0.146	+13.7	+14.1	224,264
Las Animas, CO	0.028	−0.8	+26.4	7,961
Lincoln, CO	0.028	−46.7	−12.3	2,651
Logan, CO	0.070	−27.8	−17.5	10,532
Mesa, CO	0.093	−8.8	−11.9	90,623
Mineral, CO	0.006	−19.7	+29.4	756
Moffat, CO	0.034	−35.2	−30.9	7,026
Montezuma, CO	0.021	−9.2	+9.4	15,500
Montrose, CO	0.043	−20.2	−10.1	24,922
Morgan, CO	0.109	−26.5	−14.0	13,744
Otero, CO	0.049	−15.0	+0.7	9,576
Ouray, CO	0.009	+20.3	+52.5	4,020
Park, CO	0.013	+3.1	+24.7	12,285
Phillips, CO	0.033	−43.4	−12.1	2,485
Pitkin, CO	0.014	+34.5	+60.7	11,957
Prowers, CO	0.044	−35.8	−19.4	5,561
Pueblo, CO	0.101	+2.0	+5.1	88,239
Rio Blanco, CO	0.015	−42.7	−7.1	3,691
Rio Grande, CO	0.022	−22.1	+1.3	6,305
Routt, CO	0.028	+28.3	+47.8	16,876
Saguache, CO	0.011	+18.8	+61.5	3,389
San Juan, CO	0.007	+23.9	+68.5	562
San Miguel, CO	0.008	+36.5	+69.2	5,147
Sedgwick, CO	0.058	−41.9	−8.0	1,446
Summit, CO	0.026	+36.8	+52.4	18,479
Teller, CO	0.016	−20.1	−8.9	16,962
Washington, CO	0.031	−45.5	−11.7	3,008
Weld, CO	0.190	−9.6	−7.3	166,255
Yuma, CO	0.033	−54.2	−18.9	4,981
Kent, DE	0.108	−0.5	+1.0	87,021
New Castle, DE	0.183	+18.2	+16.2	287,633
Sussex, DE	0.079	−0.8	+2.3	129,356
Alachua, FL	0.146	−1.4	+5.7	142,988
Baker, FL	0.080	−37.4	−31.1	14,082
Bay, FL	0.092	−30.7	−33.8	93,213
Bradford, FL	0.073	−29.0	−22.6	13,704
Brevard, FL	0.120	−9.8	−15.5	360,580
Broward, FL	0.285	−7.9	−9.9	959,445
Calhoun, FL	0.036	−29.1	−14.4	6,537
Charlotte, FL	0.092	+0.2	−6.7	116,560
Citrus, FL	0.080	−7.0	−10.1	93,403
Clay, FL	0.124	−26.3	−29.9	124,589
Collier, FL	0.115	−26.4	−26.5	208,285
Columbia, FL	0.086	−24.8	−19.8	33,092
DeSoto, FL	0.076	−18.1	−7.4	12,676
Dixie, FL	0.043	−17.7	−8.7	8,173
Duval, FL	0.201	−16.9	−19.5	494,212
Escambia, FL	0.120	−17.9	−20.5	170,856
Flagler, FL	0.093	−9.6	−12.1	71,871
Franklin, FL	0.036	−16.3	+0.5	6,874
Gadsden, FL	0.058	−2.6	+16.2	23,763
Gilchrist, FL	0.051	−21.7	−13.3	9,702
Glades, FL	0.056	−34.8	−10.9	5,212
Gulf, FL	0.036	−32.4	−17.0	8,172
Hamilton, FL	0.065	−25.6	−5.7	5,567
Hardee, FL	0.070	−38.5	−21.2	8,500
Hendry, FL	0.069	−31.6	−14.0	12,943
Hernando, FL	0.105	−5.6	−10.9	109,147
Highlands, FL	0.072	−20.7	−17.3	52,232
Hillsborough, FL	0.216	−15.2	−15.4	713,279
Holmes, FL	0.044	−35.5	−26.1	9,078
Indian River, FL	0.098	−11.8	−15.0	97,740
Jackson, FL	0.051	−27.7	−15.3	22,456
Jefferson, FL	0.049	−2.2	+15.9	8,468
Lafayette, FL	0.039	−45.1	−26.9	3,661
Lake, FL	0.133	−11.1	−11.9	211,284
Lee, FL	0.127	−11.7	−16.1	393,618
Leon, FL	0.120	−6.5	−1.2	161,120
Levy, FL	0.057	−16.7	−5.8	23,196
Liberty, FL	0.033	−26.1	−3.4	3,566
Madison, FL	0.052	−19.5	−2.6	9,393
Manatee, FL	0.147	−10.5	−15.1	217,471
Marion, FL	0.119	−12.3	−12.6	204,736
Martin, FL	0.129	−13.7	−16.6	98,969
Miami-Dade, FL	0.286	−42.6	−39.1	1,154,200
Monroe, FL	0.059	−9.7	−8.9	48,122
Nassau, FL	0.101	−27.0	−24.8	58,920
Okaloosa, FL	0.082	−34.2	−37.1	116,748
Okeechobee, FL	0.075	−26.7	−20.2	15,984
Orange, FL	0.255	−9.1	−6.7	649,086
Osceola, FL	0.167	−12.8	−5.7	172,793
Palm Beach, FL	0.192	−4.5	−7.5	774,179
Pasco, FL	0.137	−8.6	−13.4	302,673
Pinellas, FL	0.184	+3.9	−5.9	556,668
Polk, FL	0.146	−15.3	−15.7	344,106
Putnam, FL	0.069	−19.8	−13.8	36,403
St. Johns, FL	0.126	−25.3	−24.5	177,026
St. Lucie, FL	0.136	−5.5	−8.9	172,340
Santa Rosa, FL	0.078	−28.1	−30.3	107,193
Sarasota, FL	0.132	+0.2	−5.6	271,167
Seminole, FL	0.212	−14.4	−14.6	261,618
Sumter, FL	0.170	−17.7	−22.0	92,538
Suwannee, FL	0.059	−31.6	−20.7	21,057
Taylor, FL	0.050	−33.6	−20.8	10,135
Union, FL	0.062	−32.5	−24.7	6,251
Volusia, FL	0.135	−2.7	−7.3	308,100
Wakulla, FL	0.048	−15.6	−7.8	18,133
Walton, FL	0.062	−34.3	−28.0	43,795
Washington, FL	0.046	−29.4	−19.3	12,335
Appling, GA	0.049	−33.5	−21.6	8,341
Atkinson, GA	0.042	−36.7	−15.0	3,155
Bacon, GA	0.052	−44.8	−32.8	4,668
Baker, GA	0.036	−28.7	+1.2	1,555
Baldwin, GA	0.072	−18.7	−10.4	18,283
Banks, GA	0.097	−37.1	−30.9	8,784
Barrow, GA	0.175	−29.0	−29.6	37,921
Bartow, GA	0.179	−28.0	−27.9	50,434
Ben Hill, GA	0.073	−27.3	−21.2	6,560
Berrien, GA	0.048	−33.0	−22.2	7,743
Bibb, GA	0.170	−14.0	−10.3	69,327
Bleckley, GA	0.063	−40.5	−30.7	5,706
Brantley, GA	0.046	−32.1	−24.2	7,746
Brooks, GA	0.051	−20.3	−3.6	7,100
Bryan, GA	0.099	−29.6	−23.4	21,340
Bulloch, GA	0.068	−31.9	−22.1	30,084
Burke, GA	0.053	−13.7	+6.0	10,684
Butts, GA	0.130	−28.3	−23.6	11,771
Calhoun, GA	0.038	−16.0	+9.1	2,194
Camden, GA	0.099	−18.3	−13.8	23,688
Candler, GA	0.051	−31.7	−16.0	4,431
Carroll, GA	0.104	−27.0	−25.1	54,450
Catoosa, GA	0.165	−21.8	−27.3	32,593
Charlton, GA	0.055	−45.2	−26.7	4,566
Chatham, GA	0.140	−12.2	−9.9	133,420
Chattahoochee, GA	0.051	−21.1	−5.2	1,582
Chattooga, GA	0.089	−23.1	−19.8	10,050
Cherokee, GA	0.156	−30.7	−33.2	144,830
Clarke, GA	0.168	+2.0	+8.1	51,333
Clay, GA	0.035	−26.5	+3.0	1,434
Clayton, GA	0.184	+5.2	+7.8	111,190
Clinch, GA	0.043	−34.1	−11.5	2,864
Cobb, GA	0.230	−20.7	−18.8	393,747
Coffee, GA	0.064	−34.4	−23.5	15,214
Colquitt, GA	0.062	−44.3	−33.8	16,083
Columbia, GA	0.127	−30.9	−30.7	80,581
Cook, GA	0.081	−36.0	−24.1	7,035
Coweta, GA	0.117	−33.1	−30.9	76,798
Crawford, GA	0.067	−23.1	−10.9	6,102
Crisp, GA	0.108	−40.8	−28.2	8,039
Dade, GA	0.096	−23.4	−18.7	7,434
Dawson, GA	0.091	−44.4	−38.5	16,081
Decatur, GA	0.050	−25.9	−14.0	11,627
DeKalb, GA	0.255	+3.2	+7.0	370,006
Dodge, GA	0.053	−38.7	−25.2	8,070
Dooly, GA	0.080	−26.3	−3.6	4,105
Dougherty, GA	0.090	−17.1	−9.9	35,296
Douglas, GA	0.159	−13.8	−8.3	69,121
Early, GA	0.039	−29.6	−5.8	5,187
Echols, GA	0.045	−59.6	−33.6	1,441
Effingham, GA	0.092	−34.0	−29.9	31,570
Elbert, GA	0.073	−26.4	−15.2	9,171
Emanuel, GA	0.048	−33.0	−16.9	9,501
Evans, GA	0.052	−35.6	−20.7	4,247
Fannin, GA	0.049	−27.5	−20.8	14,865
Fayette, GA	0.121	−33.3	−26.7	71,994
Floyd, GA	0.134	−27.8	−25.4	41,648
Forsyth, GA	0.163	−48.4	−45.0	129,176
Franklin, GA	0.108	−38.2	−32.2	10,765
Fulton, GA	0.273	−14.0	−9.0	518,228
Gilmer, GA	0.054	−31.4	−23.6	16,510
Glascock, GA	0.048	−42.4	−28.0	1,566
Glynn, GA	0.089	−23.0	−19.3	41,984
Gordon, GA	0.156	−31.5	−29.4	24,066
Grady, GA	0.051	−29.3	−16.4	10,707
Greene, GA	0.074	−38.0	−20.8	11,247
Gwinnett, GA	0.231	−21.8	−18.5	413,865
Habersham, GA	0.069	−42.1	−37.1	20,453
Hall, GA	0.157	−34.9	−33.5	90,517
Hancock, GA	0.049	+5.5	+32.1	4,133
Haralson, GA	0.097	−31.9	−30.3	14,248
Harris, GA	0.068	−37.5	−24.8	19,991
Hart, GA	0.087	−27.5	−22.2	12,727
Heard, GA	0.059	−25.0	−16.1	5,391
Henry, GA	0.179	−14.8	−10.8	122,762
Houston, GA	0.117	−29.7	−29.4	74,819
Irwin, GA	0.054	−42.7	−24.0	4,168
Jackson, GA	0.172	−37.7	−34.8	37,673
Jasper, GA	0.081	−36.5	−23.5	7,644
Jeff Davis, GA	0.051	−40.8	−28.8	5,771
Jefferson, GA	0.057	−11.4	+9.4	7,642
Jenkins, GA	0.048	−21.4	−3.7	3,455
Johnson, GA	0.046	−25.0	−11.4	4,100
Jones, GA	0.083	−17.6	−11.2	14,966
Lamar, GA	0.088	−31.0	−23.3	9,039
Lanier, GA	0.048	−40.8	−27.4	2,844
Laurens, GA	0.064	−32.0	−22.5	22,729
Lee, GA	0.062	−46.9	−38.6	16,729
Liberty, GA	0.083	−0.9	+5.4	21,257
Lincoln, GA	0.054	−25.7	−10.6	4,650
Long, GA	0.059	−20.6	−5.6	5,656
Lowndes, GA	0.104	−30.0	−25.2	46,355
Lumpkin, GA	0.064	−32.1	−26.0	15,531
McDuffie, GA	0.085	−27.4	−15.9	10,455
McIntosh, GA	0.077	−18.0	0.0	6,696
Macon, GA	0.069	−8.2	+14.8	4,662
Madison, GA	0.090	−27.9	−21.3	14,936
Marion, GA	0.045	−22.1	−0.9	3,624
Meriwether, GA	0.065	−19.9	−5.3	10,878
Miller, GA	0.038	−45.3	−24.0	2,835
Mitchell, GA	0.045	−22.4	−5.7	8,963
Monroe, GA	0.126	−34.4	−22.9	15,492
Montgomery, GA	0.042	−38.8	−21.9	3,966
Morgan, GA	0.101	−34.0	−20.3	11,707
Murray, GA	0.073	−29.7	−27.5	15,389
Muscogee, GA	0.133	−14.7	−13.9	80,509
Newton, GA	0.158	−13.8	−9.7	54,239
Oconee, GA	0.125	−34.4	−25.7	25,168
Oglethorpe, GA	0.071	−25.2	−11.2	8,131
Paulding, GA	0.144	−24.2	−26.1	85,385
Peach, GA	0.117	−25.3	−15.2	12,549
Pickens, GA	0.071	−28.9	−26.7	17,116
Pierce, GA	0.058	−44.8	−36.9	9,048
Pike, GA	0.071	−35.2	−29.8	10,720
Polk, GA	0.101	−30.8	−26.0	17,399
Pulaski, GA	0.060	−38.0	−30.7	4,059
Putnam, GA	0.070	−41.2	−27.0	11,855
Quitman, GA	0.033	−20.2	+3.9	1,106
Rabun, GA	0.036	−33.0	−18.8	9,568
Randolph, GA	0.038	−36.4	−5.9	3,074
Richmond, GA	0.150	−3.5	−1.9	87,014
Rockdale, GA	0.179	−9.5	−3.0	44,686
Schley, GA	0.045	−35.2	−19.7	2,275
Screven, GA	0.050	−22.0	−2.2	6,628
Seminole, GA	0.041	−30.7	−16.6	3,884
Spalding, GA	0.123	−21.5	−20.1	30,116
Stephens, GA	0.088	−25.6	−24.6	11,884
Stewart, GA	0.035	−21.3	+11.4	1,990
Sumter, GA	0.072	−27.2	−10.9	12,150
Talbot, GA	0.057	−4.3	+22.5	3,522
Taliaferro, GA	0.059	+8.0	+37.2	928
Tattnall, GA	0.048	−45.2	−30.7	8,183
Taylor, GA	0.061	−32.1	−11.3	3,839
Telfair, GA	0.046	−30.8	−11.6	4,333
Terrell, GA	0.056	−13.0	+6.9	4,416
Thomas, GA	0.061	−29.5	−19.7	21,853
Tift, GA	0.106	−37.4	−31.2	16,283
Toombs, GA	0.057	−44.2	−33.7	10,914
Towns, GA	0.039	−35.0	−25.1	7,983
Treutlen, GA	0.048	−26.5	−11.4	3,077
Troup, GA	0.106	−28.8	−21.9	30,049
Turner, GA	0.074	−29.7	−11.4	3,792
Twiggs, GA	0.067	−15.5	+5.5	4,444
Union, GA	0.044	−27.6	−21.0	15,556
Upson, GA	0.069	−18.8	−14.6	12,905
Walker, GA	0.112	−23.3	−25.0	29,354
Walton, GA	0.153	−36.3	−35.4	51,095
Ware, GA	0.096	−32.2	−27.2	14,192
Warren, GA	0.062	−13.8	+10.8	2,651
Washington, GA	0.061	−25.9	−4.7	9,459
Wayne, GA	0.064	−36.3	−29.0	12,778

Showing the first 500 of 3096 rows.

Counties with similar diesel exposure differ by tens of points in residual margin. Each dot is a county, and the lines are binned means over tenths of counties ordered by exposure, weighted by votes and counting each county once, with 95% intervals from resampling states. Net of race, income and education the binned means counting each county once rise from the third tenth to the tenth. With density also held fixed, the tenth with the least exposure has the highest means under both weightings. Within states, the correlation is +0.18 net of race, income and education and −0.07 with density added.Note: Binned means group counties into tenths by diesel exposure. The number of counties in each bin is in the tooltip.Note: Hawaii, Connecticut and the District of Columbia, and counties without a tract of more than 100 votes, are left out. County values are averages of modeled tract values, so they carry the errors of both models.Note: Intervals on the binned means resample the states. They do not include error in the allocation of precinct votes to tracts or in the modeled pollution values.Source: Precinct votes allocated to census tracts (Fekrazad 2025), 2020 presidential election; EPA EJScreen version 2.3 (2024 release, 2020 air data); American Community Survey tract data (vintage not recorded in the atlas); vote-weighted county means of tract values, full counts, no sampling weights

Modeled asthma cannot register local air

A modeled health estimate is computed from other data by a formula. A measured health outcome is a count or an answer recorded for the people who live in the place. The asthma figures in this chapter are of the first kind: the CDC's PLACES estimates of current asthma among adults [5] are modeled and not measured, so they cannot show an effect of local air. PLACES fits a multilevel model to survey answers, with the age, sex, race and education of respondents and the poverty rate of the county as inputs, and applies it to each tract's population [5]. One tract's estimate differs from another's through its demographic mix and its county, and not through its local air.

Modeled prevalence is 11.0% (95% CI 11.0% to 11.1%) in the first decile of diesel exposure and 9.9% (95% CI 9.6% to 10.3%) in the tenth. Per standard deviation of diesel exposure, prevalence differs by −0.25 (95% CI −0.35 to −0.14) points with no controls, by 0.04 (95% CI −0.04 to 0.12) points with race, income and education held fixed and by −0.02 (95% CI −0.11 to 0.07) points with density added. The difference is negative with no controls and its interval includes zero once race, income and education are held fixed, as expected when the model is built from demographic inputs. The pattern describes how the modeled values line up with diesel exposure and does not test whether exposure affects asthma. Chapter 13.4, Health outcomes and the vote (planned) looks at small-area health outcomes and the vote more broadly.

The support gap shrinks within party

The Cooperative Election Study asked respondents in 2020 whether they support strengthening the Environmental Protection Agency's enforcement of the Clean Air Act and the Clean Water Act even if it costs U.S. jobs. The item names two statutes and a cost in jobs, so it measures more than support for the agency, and it does not ask about local pollution or health. Among the 57,411 respondents in 2,630 counties who gave an answer and a party ID and live in a county with tracts in the sample, weighted support is 59.7% (95% CI 58.5% to 60.8%). Respondents in the fifth whose counties have the most diesel exposure are more supportive, 69.5% (95% CI 67.2% to 71.8%) against 51.2% (95% CI 49.8% to 52.7%) in the fifth whose counties have the least, and they are more often Democrats, 59.0% (95% CI 56.0% to 62.0%) against 32.3% (95% CI 30.9% to 33.7%). Support is 86.1% (95% CI 85.3% to 86.9%) among Democrats, 59.7% (95% CI 58.0% to 61.4%) among independents and 30.6% (95% CI 29.5% to 31.7%) among Republicans, with leaners counted in their party.

Figure 6 fits a linear probability model of support on standardized county diesel exposure, with survey weights and errors clustered by county. For all respondents the coefficient is 6.4 (95% CI 5.3 to 7.6) points per standard deviation with no controls, 1.3 (95% CI 0.9 to 1.7) points with party ID and demographics held fixed and −0.4 (95% CI −1.1 to 0.4) with the density of the county added. Among Democrats it is 0.0 (95% CI −0.8 to 0.8) with no controls and 0.2 (95% CI −0.6 to 1.0) with controls. Among Republicans it is 4.3 (95% CI 2.4 to 6.2) with no controls, 3.0 (95% CI 1.2 to 4.8) with controls and 1.0 (95% CI −1.4 to 3.4) once county density is added. With county density added every group's interval includes zero. The fifth of respondents in the most exposed counties live in only 54 counties, so large urban counties make up that end of the exposure range.

Figure 6

Support for EPA enforcement rises with county diesel exposure, but the rise shrinks within party and its interval includes zero with county density held fixed

Percentage-point difference in support for strengthening EPA enforcement of the Clean Air Act and Clean Water Act even if it costs U.S. jobs, per standard deviation of county diesel exposure; 57,411 respondents in 2,630 counties, 2020; 95% intervals clustered by county

All respondents
Democrats
Independents
Republicans

No controls

Party ID and demographics

Plus county density

Support per SD of diesel (points)

Show data

Data for Support for EPA enforcement rises with county diesel exposure, but the rise shrinks within party and its interval includes zero with county density held fixed
conf_high	conf_low	estimate	group	n	specification
7.57	5.30	6.43	All respondents	57,411	No controls
1.74	0.861	1.30	All respondents	57,411	Party ID and demographics
0.369	−1.12	−0.377	All respondents	57,411	Plus county density
0.829	−0.773	0.028	Democrats	28,671	No controls
0.955	−0.620	0.168	Democrats	28,671	Party ID and demographics
0.568	−1.77	−0.603	Democrats	28,671	Plus county density
4.81	1.34	3.08	Independents	8,643	No controls
3.51	−0.177	1.66	Independents	8,643	Party ID and demographics
1.18	−4.11	−1.47	Independents	8,643	Plus county density
6.16	2.39	4.27	Republicans	20,097	No controls
4.78	1.18	2.98	Republicans	20,097	Party ID and demographics
3.40	−1.42	0.991	Republicans	20,097	Plus county density

For all respondents the coefficient falls from 6.4 points with no controls to 1.3 with party ID and demographics held fixed and −0.4 with county density added, and every group's interval includes zero at that last step. Each dot is a weighted linear probability model coefficient. The first panel uses no controls, the second holds fixed the seven-point party ID, race, education, income, age and gender, and the third adds the log density of the county. One standard deviation of county diesel exposure across respondents is 0.12 micrograms per cubic meter. Democrats, independents and Republicans include leaners, and each row is estimated on that group alone.Note: The CES is an opt-in panel, so its intervals are model based. County exposure is a county value assigned to each respondent, not an exposure the respondent was measured to have.Note: It is a different CES item from the one on giving the EPA power to regulate carbon dioxide.Source: Cooperative Election Study 2020 wave, item on strengthening EPA enforcement of the Clean Air Act and Clean Water Act, weighted with the CES survey weight; county diesel exposure from EPA EJScreen version 2.3 averaged over tracts by 2020 votes

What these data cannot show

The estimates describe places in one election. They do not say how the residents of a tract voted, because a pattern across areas can differ from the pattern among the people in them [6] (ecological inference). The vote in each tract is an allocation of precinct returns by a land-cover model and each pollution value is modeled, so the estimates carry errors that the intervals leave out. The exposures and the vote are both from 2020, so the data cannot say whether pollution changes how people vote, whether people who vote a certain way settle near sources, or whether the siting of sources follows political power, a claim the environmental justice literature makes [7].

Holding traits fixed is partial. The controls come from the American Community Survey and carry sampling error, and density enters as one summary of urban form. The four exposures move together, so the data cannot say which one the coefficients reflect, and the county residual varies strongly between states, so a correlation across counties can reflect regional differences that the data cannot separate from exposure. Coverage is 47 states, because Alaska, Hawaii and Connecticut have no tract vote estimates and the District of Columbia lacks the white share that the controls need. The survey comparison assigns a county exposure to each respondent and cannot see how much air pollution a respondent breathes. The vote is the presidential margin in one election, so the data say nothing about turnout, down-ballot races or votes on local environmental measures, and they hold no local political context, so they cannot say why tracts with more exposure vote as they do.

Data sources

Precinct votes on census geographiesPrecinct votes allocated to 2020 census tracts (Fekrazad 2025) with the ACS demographics of the tract atlas (ACS years not recorded), 2020 presidential electionYears: 2016 and 2020 (contiguous United States only) · Weights: votes cast in the tract (pres_tot); full counts of tracts, no sampling weights · License: CC0 (data doi:10.7910/DVN/Z8TSH3), as stated in the project README.
EJScreenEPA EJScreen version 2.3 (July 2024; modeled air data for 2020), tract values in the atlas; EPA removed the tool in February 2025 and the archived copy is not recordedYears: Version as distributed with the tract atlas file. · Weights: none (tract values) · License: US government data. The EPA removed EJScreen from its website in February 2025, so the archived copy is the source to cite.
Area Deprivation IndexArea Deprivation Index national rank, tract values in the atlas; release year not recordedYears: Vintage as distributed with the tract and ZCTA atlas files. · Weights: none (tract values) · License: The Neighborhood Atlas may restrict commercial use and redistribution; read its terms before publishing derived tract values.
CDC PLACESCurrent asthma among adults, model-based tract estimates in the atlas; release year not recordedYears: Release as distributed with the tract and ZCTA atlas files. · Weights: none (tract values) · License: US government data, open access.
Cooperative Election StudyCooperative Election Study 2020 wave; support for strengthening EPA enforcement of the Clean Air Act and Clean Water Act even if it costs U.S. jobsYears: 2006-2024 (every year; self-reported turnout is available in the 10 even years) · Weights: weight (single wave) · License: See the provider's terms of use.

All data sources

Methods

Sample: census tracts in the tract atlas (82,765 tracts in 47 states and the District of Columbia; Alaska and Hawaii have no tract vote estimates, and the tract atlas has no rows for Connecticut) with more than 100 estimated presidential votes in 2020, a two-party margin, all four pollution measures, the five controls and a land area above zero: 81,134 tracts in 47 states and 3,096 counties, carrying 155.1 million votes. The District of Columbia drops out because the atlas has no white share for its tracts. Vote: precinct returns allocated to 2020 tracts by Regionalized Land Cover Regression (Fekrazad 2025), fractional votes; margin = 100 (D minus R) over (D plus R). Pollution: tract values of EPA EJScreen version 2.3 (July 2024) in the atlas: diesel particulate matter (micrograms per cubic meter; 2020 AirToxScreen), traffic proximity (annual average daily traffic on major roads within 10 kilometers divided by distance; 2020 highway counts), nitrogen dioxide (parts per billion; NASA satellite model, 2020) and fine particles (micrograms per cubic meter; 2020). EPA removed EJScreen from its website on 5 February 2025. The atlas does not record the version or the archived copy; version 2.3 is identified from the values (the fine-particle range and the nitrogen dioxide maximum match the 2.3 documentation, and the median diesel value is close to the documented median, which is a block-group value while the atlas holds tract values, so the atlas diesel range is narrower than the documented one). Controls: white, Black and Hispanic shares of residents, college share and median household income are American Community Survey tract values; the atlas does not record the survey years. Estimands: the vote-weighted mean margin by decile of diesel exposure (equal numbers of tracts); the coefficient of the margin on each exposure standardized with the unweighted standard deviation across tracts; the share of the raw diesel coefficient that remains after controls (a ratio of point estimates); the within-county coefficient minus the pooled coefficient with density; and, across counties, the correlation of county diesel exposure with the county residual margin (the tract margin minus the margin predicted by a weighted least squares model with no pollution term, on the demographic controls alone or with log density added, averaged over the county's tracts by votes). Weights: tract presidential votes. Estimator: weighted least squares; controls are the shares of white, Black and Hispanic residents, the college share and median household income; density is the log of people per square kilometer; the within-county estimate removes the vote-weighted county mean from every variable before fitting. Interval method: 95%, clustered by county (3,096 clusters); the retained shares and the within-county differences use a county bootstrap (2,000 replicates, seed 12345, the same resampled counties for every model); correlations across tracts and counties use a county bootstrap (2,000 replicates, seed 12345); county correlations are reported weighted by county votes and unweighted, because the weighting decides the sign of the density-adjusted one; within-state correlations (county diesel exposure and the county residual margin each measured from its vote-weighted state mean, correlated with county votes as weights) and the binned means of the county residual margin over tenths of counties ordered by diesel exposure (weighted by county votes, or counting each county once) resample the 47 states (2,000 replicates, seed 12345); the share of the variance of the county residual margin that lies between states is one minus the ratio of the vote-weighted within-state variance to the total variance and has no interval; the state-clustered interval on the diesel coefficient uses 47 clusters. Intervals are model based and do not include error in the vote allocation or in the modeled pollution. The atlas does not record how EJScreen block-group values were turned into tract values (diesel and fine particles are tract values in the source). Deprivation: Area Deprivation Index national rank. Asthma: CDC PLACES current asthma among adults, a model-based estimate that cannot vary with local air (its inputs are demographic and geographic). The release years of the ADI and PLACES values are not recorded in the atlas. CES: 2020 wave, respondents with an answer on the item on strengthening EPA enforcement of the Clean Air Act and Clean Water Act even if it costs U.S. jobs (support_epa_airwater, from the source item enviro_airwateracts, coded 1 for support), a seven-point party ID and a county in the tract sample (57,411 respondents in 2,630 counties); weighted least squares on a 0 or 100 outcome with the survey weight; county exposure is the vote-weighted county mean of tract diesel exposure. Suppression: tracts with 100 or fewer votes are dropped; no survey cell rests on fewer than 30 respondents; tract deciles and county values are full counts of tracts and counties.

References

Fekrazad, Amir (2025). A dataset of US precinct votes allocated to Census geographies with precision. Scientific Data 12: 794. doi:10.1038/s41597-025-05140-3
US Environmental Protection Agency (2024). EJScreen Technical Documentation for Version 2.3 (revised July 31, 2024). Office of Environmental Justice and External Civil Rights, Washington, DC. epa.gov/system/files/documents/2024-07/ejscreen-tech-doc-version-2-3.pdf
Kind, Amy J. H.; Buckingham, William R. (2018). Making Neighborhood-Disadvantage Metrics Accessible: The Neighborhood Atlas. New England Journal of Medicine 378: 2456-2458. doi:10.1056/NEJMp1802313
Rodden, Jonathan A. (2019). Why Cities Lose: The Deep Roots of the Urban-Rural Political Divide. Basic Books, New York.
Greenlund, Kurt J.; Lu, Hua; Wang, Yan; Matthews, Kevin A.; LeClercq, Jennifer M.; Lee, Benjamin; Carlson, Susan A. (2022). PLACES: Local Data for Better Health. Preventing Chronic Disease 19: 210459. doi:10.5888/pcd19.210459
Robinson, W. S. (1950). Ecological Correlations and the Behavior of Individuals. American Sociological Review 15(3): 351-357.
Mohai, Paul; Pellow, David; Roberts, J. Timmons (2009). Environmental Justice. Annual Review of Environment and Resources 34: 405-430. doi:10.1146/annurev-environ-082508-094348

How to cite

@misc{singh2026airpollutionandthevote,
  author = {Singh, Vishal},
  title = {Air pollution and the vote},
  year = {2026},
  url = {https://vishalsingh.org/politics/air-pollution-and-the-vote},
  note = {In The Public Mind. Published 30 September 2026.}
}

Reading paths

A tour of the book, one chapter per part