Chapter 10.2 · Weighing the ingredients

The anatomy of a deprivation index

The Area Deprivation Index ranks census tracts almost as median home value does, with a rank correlation of −0.988 across 75,717 tracts (negative because the index rises as home values fall), and the home value term alone accounts for 74.3% of the variance of the score that the formula produces.

Author: Vishal Singh
Affiliation: NYU Stern
Published: 9 October 2026

Across 75,717 census tracts, the Area Deprivation Index (ADI) of the Neighborhood Atlas has a population-weighted rank correlation, a correlation between the positions of tracts in two rankings, of −0.988 (95% CI −0.989 to −0.986) with median home value. The sign is negative because the index rises as home values fall, and the size means that the index orders tracts almost as home prices do.[1, 2] A replica built from the index's published formula and American Community Survey (ACS) estimates for 2019 to 2023 reproduces the published tract rank with a rank correlation of +0.991 (95% CI 0.990 to 0.992), and in the replica's score the home value term alone accounts for 74.3% (95% CI 70.5% to 77.0%) of the variance. The formula multiplies each of 17 census measures by a fixed coefficient and adds the products. 4 of the measures are dollar amounts and the rest are proportions or a log ratio, so the dollar amounts dominate the sum.

The same arithmetic shapes who the index counts as less deprived. Of the 17.2 million residents of the 5,100 tracts of the 50 states and DC where 30% or more of people are below the poverty line, the published ADI places 16.1% (95% CI 11.3% to 21.4%) in tracts ranked at or below the 50th national percentile, the less deprived end of its ranking. The same measures standardized before the same coefficients are applied place 0.34% (95% CI 0.15% to 0.55%) of them there. This chapter rebuilds the index, shows where its weight goes, compares it with other rankings of the same tracts and tests how closely each ranking tracks life expectancy and upward mobility. Chapter 10.1, Gradients, slopes and variance covers the variance of outcomes across states, counties and tracts and the slopes of outcomes on the index. Chapter 10.3, A deprivation index at two zooms repeats the index for counties, and Chapter 10.4, Families of places sorts tracts into groups instead of ranking them.

How the Neighborhood Atlas builds the index

Gopal Singh chose the census measures and estimated their coefficients as factor scores from 1990 census tracts, for a study of mortality in the United States.[3] The Neighborhood Atlas at the University of Wisconsin applies his recipe to American Community Survey data and publishes national percentile ranks of census block groups, with the lowest ranks for the least deprived and the highest for the most deprived.[4] A block group's score is the sum of the 17 measures in Figure 1, each multiplied by Singh's coefficient, and the Atlas publishes the ranks of the scores and not the scores.[2, 5] The ADI used here is the 2023 release, version 4.0.1, rolled up from block groups to 2020 tracts. The dataset registry records it as built on ACS five-year estimates for 2019 to 2023, the period of the ACS measures used here. Since the 2020 release the Atlas has counted households without internet where Singh counted households without a telephone.[5]

Figure 1

The Atlas score sums 17 measures with fixed coefficients, and 4 of them are dollar amounts

The measures in Singh's deprivation index, the unit each enters the sum in, Singh's factor score coefficient, and whether this chapter rebuilds it

No.	Measure	Unit in the sum	Coefficient	Rebuilt here
1	Adults 25 and older with less than 9 years of education	Proportion	+0.0849	Yes, as adults who stopped at 8th grade
2	Adults 25 and older with at least a high school diploma	Proportion	−0.0970	No: not in the staged ACS files; the SVI 2022 file has it for ACS 2018 to 2022 (sensitivity check)
3	Employed people 16 and older in white-collar occupations	Proportion	−0.0874	Yes, as management, business, science and arts occupations
4	Median family income	Dollars	−0.0977	Yes
5	Income disparity: log of 100 times households under $10,000 over households of $50,000 or more	Log ratio	+0.0936	Yes, with a continuity constant
6	Median home value	Dollars	−0.0688	Yes
7	Median gross rent	Dollars	−0.0781	Yes
8	Median monthly mortgage	Dollars	−0.0770	Yes, as monthly owner costs of homes with a mortgage
9	Owner-occupied housing units	Proportion	−0.0615	Yes
10	Civilian labor force 16 and older that is unemployed	Proportion	+0.0806	Yes
11	Families below the poverty level	Proportion	+0.0977	Yes, as persons below the poverty line
12	Persons below 150 percent of the poverty threshold	Proportion	+0.1037	Yes
13	Single-parent households with children under 18	Proportion	+0.0719	Yes, as single-parent households
14	Households without a motor vehicle	Proportion	+0.0694	Yes
15	Households without a telephone (the Atlas now counts households without internet)	Proportion	+0.0877	No: not in the staged ACS files; the SVI 2022 file has households without internet for ACS 2018 to 2022 (sensitivity check)
16	Occupied housing units without complete plumbing	Proportion	+0.0510	Yes
17	Households with more than one person per room	Proportion	+0.0556	Yes

Show data

Data for The Atlas score sums 17 measures with fixed coefficients, and 4 of them are dollar amounts
coefficient	measure	order	rebuilt	unit
0.085	Adults 25 and older with less than 9 years of education	1	Yes, as adults who stopped at 8th grade	Proportion
−0.097	Adults 25 and older with at least a high school diploma	2	No: not in the staged ACS files; the SVI 2022 file has it for ACS 2018 to 2022 (sensitivity check)	Proportion
−0.087	Employed people 16 and older in white-collar occupations	3	Yes, as management, business, science and arts occupations	Proportion
−0.098	Median family income	4	Yes	Dollars
0.094	Income disparity: log of 100 times households under $10,000 over households of $50,000 or more	5	Yes, with a continuity constant	Log ratio
−0.069	Median home value	6	Yes	Dollars
−0.078	Median gross rent	7	Yes	Dollars
−0.077	Median monthly mortgage	8	Yes, as monthly owner costs of homes with a mortgage	Dollars
−0.061	Owner-occupied housing units	9	Yes	Proportion
0.081	Civilian labor force 16 and older that is unemployed	10	Yes	Proportion
0.098	Families below the poverty level	11	Yes, as persons below the poverty line	Proportion
0.104	Persons below 150 percent of the poverty threshold	12	Yes	Proportion
0.072	Single-parent households with children under 18	13	Yes, as single-parent households	Proportion
0.069	Households without a motor vehicle	14	Yes	Proportion
0.088	Households without a telephone (the Atlas now counts households without internet)	15	No: not in the staged ACS files; the SVI 2022 file has households without internet for ACS 2018 to 2022 (sensitivity check)	Proportion
0.051	Occupied housing units without complete plumbing	16	Yes	Proportion
0.056	Households with more than one person per room	17	Yes	Proportion

Dollar amounts, a log ratio and proportions enter one sum, and the coefficients are all of similar size, so the unit of a measure rather than its coefficient decides how much it moves the score. Rows follow the order in Singh (2003); a negative coefficient lowers the score and a positive one raises it, and the last column says which measures the chapter rebuilds for 15 of the 17.Note: The Atlas applies Singh's coefficients to proportions between zero and one, to dollars, and to the log ratio, and ranks the sums; it does not publish the sums. Since the 2020 release it counts households without internet where Singh counted households without a telephone.Note: Singh estimated the coefficients as factor scores in 1990 census tract data; they are not re-estimated here.Source: Singh (2003) factor score coefficients for 1990 census tracts, as printed in Petterson (2023); Neighborhood Atlas change log, 2022; published coefficients, not weighted

The measures cover education, work, income and housing. 4 are dollar amounts (median family income, home value, gross rent and the monthly mortgage cost), one is a log ratio of low-income to high-income households, and the rest are proportions of people, households or homes. The Atlas sums them in these units, with proportions entered as fractions of one and not as percentages.[2] The coefficients are close in size, so the unit of a measure more than its coefficient sets how much the measure moves the sum, and a dollar amount in the hundreds of thousands varies over a wider range than a proportion.

Most of the measures can be rebuilt from ACS data for 2020 tracts

The staged ACS tract file and a Census API pull of housing, occupation, income and schooling tables supply 15 of the 17 measures for 2020 census tracts. The 2 missing are the share of adults with at least a high school diploma and the share of households without internet. Some rebuilt measures are substitutes. The poverty rate counts persons where Singh counted families, white-collar work is the share of workers in management, business, science and arts occupations, and the mortgage cost is the median monthly owner cost of homes with a mortgage. Income disparity is Singh's ratio of low-income to high-income households, defined in the table, logged, with a continuity constant added to each count so that a tract with no low-income household keeps a value.

The replica works with tracts, and the Atlas ranks block groups. The staged ADI is the Atlas's block-group ranks averaged over each tract's block groups, without population weights. The frame covers the 50 states and DC. Puerto Rico's tracts are left out because the SVI and USALEEP have none, although the Atlas ranks them with the states. Of the 84,415 tracts, 82,784 have a published rank and residents, and 75,717 of those have all 15 measures. The frame holds 93.0% of the residents of tracts with a rank. Of the tracts lost, 90.0% lack a median rent, home value or mortgage cost, which the ACS leaves out for tracts with almost no renters or almost no owners. The Atlas imputes missing values from the surrounding geography, and the replica does not.[5]

Figure 2 shows how the rebuilt measures move together, from the measures of advantage (income, home value, rent, mortgage cost, white-collar work and homeownership) to the measures of hardship. The dollar measures move together, with home value correlating +0.86 (95% CI 0.85 to 0.87) with the mortgage cost and +0.71 (95% CI 0.68 to 0.74) with rent. They are a loose guide to hardship. Home value correlates −0.31 (95% CI −0.36 to −0.27) with the poverty rate across tracts, a weaker link than the −0.61 (95% CI −0.62 to −0.59) of family income, so a score dominated by home value follows the housing market more closely than it follows poverty.

Figure 2

Home value correlates −0.31 with the poverty rate across tracts, a weaker link than family income has at −0.61

Population-weighted Pearson correlations among the 15 rebuilt measures in their own units, 75,717 tracts, ACS 2019 to 2023

−1.000.00+1.00

A fifteen by fifteen grid of correlations among the rebuilt measures, ordered with the six measures of advantage first and the nine measures of hardship after. Dollar measures correlate positively with each other, hardship measures correlate positively with each other, and the two groups correlate negatively; home value's correlations with the hardship measures are small to moderate.
Measure	Family income	Home value	Gross rent	Mortgage cost	White-collar jobs	Homeownership	Poverty	Below 150% of poverty	Income disparity	Unemployment	Single parents	No vehicle	Crowding	No plumbing	Under 9 years of school
Family income	+1.00	+0.68, interval +0.64 to +0.73, n = 75,717	+0.64, interval +0.61 to +0.67, n = 75,717	+0.74, interval +0.70 to +0.78, n = 75,717	+0.79, interval +0.77 to +0.80, n = 75,717	+0.33, interval +0.28 to +0.37, n = 75,717	−0.61, interval −0.62 to −0.59, n = 75,717	−0.70, interval −0.71 to −0.70, n = 75,717	−0.44, interval −0.46 to −0.42, n = 75,717	−0.29, interval −0.30 to −0.27, n = 75,717	−0.49, interval −0.50 to −0.48, n = 75,717	−0.14, interval −0.26 to −0.03, n = 75,717	−0.25, interval −0.28 to −0.22, n = 75,717	−0.28, interval −0.30 to −0.27, n = 75,717	−0.36, interval −0.38 to −0.34, n = 75,717
Home value	+0.68, interval +0.64 to +0.73, n = 75,717	+1.00	+0.71, interval +0.68 to +0.74, n = 75,717	+0.86, interval +0.85 to +0.87, n = 75,717	+0.55, interval +0.52 to +0.59, n = 75,717	−0.05, interval −0.12 to +0.04, n = 75,717	−0.31, interval −0.36 to −0.27, n = 75,717	−0.38, interval −0.43 to −0.34, n = 75,717	−0.25, interval −0.29 to −0.22, n = 75,717	−0.08, interval −0.13 to −0.04, n = 75,717	−0.27, interval −0.30 to −0.25, n = 75,717	+0.11, interval −0.07 to +0.26, n = 75,717	+0.10, interval +0.02 to +0.15, n = 75,717	−0.26, interval −0.28 to −0.25, n = 75,717	−0.07, interval −0.14 to −0.01, n = 75,717
Gross rent	+0.64, interval +0.61 to +0.67, n = 75,717	+0.71, interval +0.68 to +0.74, n = 75,717	+1.00	+0.73, interval +0.71 to +0.76, n = 75,717	+0.52, interval +0.49 to +0.55, n = 75,717	+0.06, interval +0.02 to +0.11, n = 75,717	−0.39, interval −0.41 to −0.37, n = 75,717	−0.45, interval −0.47 to −0.43, n = 75,717	−0.36, interval −0.38 to −0.34, n = 75,717	−0.09, interval −0.12 to −0.06, n = 75,717	−0.22, interval −0.24 to −0.20, n = 75,717	−0.04, interval −0.17 to +0.06, n = 75,717	+0.06, interval +0.01 to +0.09, n = 75,717	−0.33, interval −0.34 to −0.32, n = 75,717	−0.09, interval −0.13 to −0.06, n = 75,717
Mortgage cost	+0.74, interval +0.70 to +0.78, n = 75,717	+0.86, interval +0.85 to +0.87, n = 75,717	+0.73, interval +0.71 to +0.76, n = 75,717	+1.00	+0.61, interval +0.57 to +0.65, n = 75,717	−0.02, interval −0.09 to +0.05, n = 75,717	−0.38, interval −0.41 to −0.34, n = 75,717	−0.46, interval −0.50 to −0.41, n = 75,717	−0.31, interval −0.34 to −0.27, n = 75,717	−0.10, interval −0.14 to −0.06, n = 75,717	−0.29, interval −0.32 to −0.27, n = 75,717	+0.10, interval −0.05 to +0.21, n = 75,717	+0.09, interval +0.01 to +0.14, n = 75,717	−0.30, interval −0.32 to −0.29, n = 75,717	−0.08, interval −0.15 to −0.02, n = 75,717
White-collar jobs	+0.79, interval +0.77 to +0.80, n = 75,717	+0.55, interval +0.52 to +0.59, n = 75,717	+0.52, interval +0.49 to +0.55, n = 75,717	+0.61, interval +0.57 to +0.65, n = 75,717	+1.00	+0.18, interval +0.14 to +0.22, n = 75,717	−0.50, interval −0.52 to −0.48, n = 75,717	−0.60, interval −0.62 to −0.59, n = 75,717	−0.34, interval −0.36 to −0.32, n = 75,717	−0.27, interval −0.29 to −0.25, n = 75,717	−0.46, interval −0.48 to −0.44, n = 75,717	−0.07, interval −0.18 to +0.03, n = 75,717	−0.35, interval −0.38 to −0.30, n = 75,717	−0.27, interval −0.28 to −0.25, n = 75,717	−0.49, interval −0.52 to −0.46, n = 75,717
Homeownership	+0.33, interval +0.28 to +0.37, n = 75,717	−0.05, interval −0.12 to +0.04, n = 75,717	+0.06, interval +0.02 to +0.11, n = 75,717	−0.02, interval −0.09 to +0.05, n = 75,717	+0.18, interval +0.14 to +0.22, n = 75,717	+1.00	−0.51, interval −0.52 to −0.49, n = 75,717	−0.52, interval −0.54 to −0.50, n = 75,717	−0.36, interval −0.38 to −0.34, n = 75,717	−0.26, interval −0.28 to −0.24, n = 75,717	−0.35, interval −0.38 to −0.32, n = 75,717	−0.54, interval −0.57 to −0.52, n = 75,717	−0.39, interval −0.43 to −0.35, n = 75,717	+0.02, interval −0.01 to +0.04, n = 75,717	−0.31, interval −0.35 to −0.27, n = 75,717
Poverty	−0.61, interval −0.62 to −0.59, n = 75,717	−0.31, interval −0.36 to −0.27, n = 75,717	−0.39, interval −0.41 to −0.37, n = 75,717	−0.38, interval −0.41 to −0.34, n = 75,717	−0.50, interval −0.52 to −0.48, n = 75,717	−0.51, interval −0.52 to −0.49, n = 75,717	+1.00	+0.91, interval +0.91 to +0.91, n = 75,717	+0.58, interval +0.57 to +0.59, n = 75,717	+0.41, interval +0.39 to +0.43, n = 75,717	+0.49, interval +0.47 to +0.50, n = 75,717	+0.38, interval +0.32 to +0.47, n = 75,717	+0.30, interval +0.27 to +0.33, n = 75,717	+0.30, interval +0.28 to +0.32, n = 75,717	+0.37, interval +0.34 to +0.40, n = 75,717
Below 150% of poverty	−0.70, interval −0.71 to −0.70, n = 75,717	−0.38, interval −0.43 to −0.34, n = 75,717	−0.45, interval −0.47 to −0.43, n = 75,717	−0.46, interval −0.50 to −0.41, n = 75,717	−0.60, interval −0.62 to −0.59, n = 75,717	−0.52, interval −0.54 to −0.50, n = 75,717	+0.91, interval +0.91 to +0.91, n = 75,717	+1.00	+0.57, interval +0.56 to +0.58, n = 75,717	+0.41, interval +0.40 to +0.43, n = 75,717	+0.53, interval +0.51 to +0.54, n = 75,717	+0.36, interval +0.29 to +0.46, n = 75,717	+0.35, interval +0.32 to +0.38, n = 75,717	+0.32, interval +0.30 to +0.34, n = 75,717	+0.44, interval +0.42 to +0.47, n = 75,717
Income disparity	−0.44, interval −0.46 to −0.42, n = 75,717	−0.25, interval −0.29 to −0.22, n = 75,717	−0.36, interval −0.38 to −0.34, n = 75,717	−0.31, interval −0.34 to −0.27, n = 75,717	−0.34, interval −0.36 to −0.32, n = 75,717	−0.36, interval −0.38 to −0.34, n = 75,717	+0.58, interval +0.57 to +0.59, n = 75,717	+0.57, interval +0.56 to +0.58, n = 75,717	+1.00	+0.26, interval +0.24 to +0.28, n = 75,717	+0.25, interval +0.24 to +0.27, n = 75,717	+0.29, interval +0.26 to +0.36, n = 75,717	+0.11, interval +0.09 to +0.13, n = 75,717	+0.23, interval +0.22 to +0.25, n = 75,717	+0.17, interval +0.16 to +0.19, n = 75,717
Unemployment	−0.29, interval −0.30 to −0.27, n = 75,717	−0.08, interval −0.13 to −0.04, n = 75,717	−0.09, interval −0.12 to −0.06, n = 75,717	−0.10, interval −0.14 to −0.06, n = 75,717	−0.27, interval −0.29 to −0.25, n = 75,717	−0.26, interval −0.28 to −0.24, n = 75,717	+0.41, interval +0.39 to +0.43, n = 75,717	+0.41, interval +0.40 to +0.43, n = 75,717	+0.26, interval +0.24 to +0.28, n = 75,717	+1.00	+0.29, interval +0.27 to +0.31, n = 75,717	+0.25, interval +0.22 to +0.30, n = 75,717	+0.21, interval +0.18 to +0.23, n = 75,717	+0.15, interval +0.12 to +0.18, n = 75,717	+0.22, interval +0.20 to +0.25, n = 75,717
Single parents	−0.49, interval −0.50 to −0.48, n = 75,717	−0.27, interval −0.30 to −0.25, n = 75,717	−0.22, interval −0.24 to −0.20, n = 75,717	−0.29, interval −0.32 to −0.27, n = 75,717	−0.46, interval −0.48 to −0.44, n = 75,717	−0.35, interval −0.38 to −0.32, n = 75,717	+0.49, interval +0.47 to +0.50, n = 75,717	+0.53, interval +0.51 to +0.54, n = 75,717	+0.25, interval +0.24 to +0.27, n = 75,717	+0.29, interval +0.27 to +0.31, n = 75,717	+1.00	+0.16, interval +0.09 to +0.24, n = 75,717	+0.30, interval +0.26 to +0.33, n = 75,717	+0.11, interval +0.09 to +0.13, n = 75,717	+0.34, interval +0.32 to +0.37, n = 75,717
No vehicle	−0.14, interval −0.26 to −0.03, n = 75,717	+0.11, interval −0.07 to +0.26, n = 75,717	−0.04, interval −0.17 to +0.06, n = 75,717	+0.10, interval −0.05 to +0.21, n = 75,717	−0.07, interval −0.18 to +0.03, n = 75,717	−0.54, interval −0.57 to −0.52, n = 75,717	+0.38, interval +0.32 to +0.47, n = 75,717	+0.36, interval +0.29 to +0.46, n = 75,717	+0.29, interval +0.26 to +0.36, n = 75,717	+0.25, interval +0.22 to +0.30, n = 75,717	+0.16, interval +0.09 to +0.24, n = 75,717	+1.00	+0.23, interval +0.15 to +0.31, n = 75,717	+0.10, interval +0.06 to +0.16, n = 75,717	+0.21, interval +0.16 to +0.27, n = 75,717
Crowding	−0.25, interval −0.28 to −0.22, n = 75,717	+0.10, interval +0.02 to +0.15, n = 75,717	+0.06, interval +0.01 to +0.09, n = 75,717	+0.09, interval +0.01 to +0.14, n = 75,717	−0.35, interval −0.38 to −0.30, n = 75,717	−0.39, interval −0.43 to −0.35, n = 75,717	+0.30, interval +0.27 to +0.33, n = 75,717	+0.35, interval +0.32 to +0.38, n = 75,717	+0.11, interval +0.09 to +0.13, n = 75,717	+0.21, interval +0.18 to +0.23, n = 75,717	+0.30, interval +0.26 to +0.33, n = 75,717	+0.23, interval +0.15 to +0.31, n = 75,717	+1.00	+0.01, interval −0.02 to +0.04, n = 75,717	+0.64, interval +0.57 to +0.70, n = 75,717
No plumbing	−0.28, interval −0.30 to −0.27, n = 75,717	−0.26, interval −0.28 to −0.25, n = 75,717	−0.33, interval −0.34 to −0.32, n = 75,717	−0.30, interval −0.32 to −0.29, n = 75,717	−0.27, interval −0.28 to −0.25, n = 75,717	+0.02, interval −0.01 to +0.04, n = 75,717	+0.30, interval +0.28 to +0.32, n = 75,717	+0.32, interval +0.30 to +0.34, n = 75,717	+0.23, interval +0.22 to +0.25, n = 75,717	+0.15, interval +0.12 to +0.18, n = 75,717	+0.11, interval +0.09 to +0.13, n = 75,717	+0.10, interval +0.06 to +0.16, n = 75,717	+0.01, interval −0.02 to +0.04, n = 75,717	+1.00	+0.07, interval +0.05 to +0.10, n = 75,717
Under 9 years of school	−0.36, interval −0.38 to −0.34, n = 75,717	−0.07, interval −0.14 to −0.01, n = 75,717	−0.09, interval −0.13 to −0.06, n = 75,717	−0.08, interval −0.15 to −0.02, n = 75,717	−0.49, interval −0.52 to −0.46, n = 75,717	−0.31, interval −0.35 to −0.27, n = 75,717	+0.37, interval +0.34 to +0.40, n = 75,717	+0.44, interval +0.42 to +0.47, n = 75,717	+0.17, interval +0.16 to +0.19, n = 75,717	+0.22, interval +0.20 to +0.25, n = 75,717	+0.34, interval +0.32 to +0.37, n = 75,717	+0.21, interval +0.16 to +0.27, n = 75,717	+0.64, interval +0.57 to +0.70, n = 75,717	+0.07, interval +0.05 to +0.10, n = 75,717	+1.00

Show data

Data for Home value correlates −0.31 with the poverty rate across tracts, a weaker link than family income has at −0.61
conf_high	conf_low	measure	n	other	r
n/a	n/a	Family income	n/a	Family income	1
0.729	0.637	Family income	75,717	Home value	0.682
0.666	0.609	Family income	75,717	Gross rent	0.638
0.781	0.696	Family income	75,717	Mortgage cost	0.739
0.796	0.774	Family income	75,717	White-collar jobs	0.786
0.369	0.284	Family income	75,717	Homeownership	0.326
−0.595	−0.618	Family income	75,717	Poverty	−0.606
−0.696	−0.715	Family income	75,717	Below 150% of poverty	−0.705
−0.425	−0.460	Family income	75,717	Income disparity	−0.442
−0.268	−0.304	Family income	75,717	Unemployment	−0.286
−0.477	−0.504	Family income	75,717	Single parents	−0.491
−0.034	−0.262	Family income	75,717	No vehicle	−0.145
−0.217	−0.283	Family income	75,717	Crowding	−0.252
−0.269	−0.297	Family income	75,717	No plumbing	−0.282
−0.345	−0.384	Family income	75,717	Under 9 years of school	−0.365
0.729	0.637	Home value	75,717	Family income	0.682
n/a	n/a	Home value	n/a	Home value	1
0.740	0.684	Home value	75,717	Gross rent	0.713
0.870	0.847	Home value	75,717	Mortgage cost	0.858
0.593	0.518	Home value	75,717	White-collar jobs	0.553
0.036	−0.120	Home value	75,717	Homeownership	−0.050
−0.271	−0.358	Home value	75,717	Poverty	−0.313
−0.339	−0.429	Home value	75,717	Below 150% of poverty	−0.381
−0.217	−0.295	Home value	75,717	Income disparity	−0.252
−0.036	−0.134	Home value	75,717	Unemployment	−0.084
−0.254	−0.297	Home value	75,717	Single parents	−0.274
0.262	−0.075	Home value	75,717	No vehicle	0.110
0.153	0.018	Home value	75,717	Crowding	0.100
−0.247	−0.281	Home value	75,717	No plumbing	−0.262
−0.011	−0.145	Home value	75,717	Under 9 years of school	−0.071
0.666	0.609	Gross rent	75,717	Family income	0.638
0.740	0.684	Gross rent	75,717	Home value	0.713
n/a	n/a	Gross rent	n/a	Gross rent	1
0.758	0.706	Gross rent	75,717	Mortgage cost	0.734
0.551	0.490	Gross rent	75,717	White-collar jobs	0.522
0.113	0.022	Gross rent	75,717	Homeownership	0.064
−0.373	−0.408	Gross rent	75,717	Poverty	−0.391
−0.434	−0.472	Gross rent	75,717	Below 150% of poverty	−0.453
−0.341	−0.383	Gross rent	75,717	Income disparity	−0.361
−0.061	−0.117	Gross rent	75,717	Unemployment	−0.089
−0.196	−0.236	Gross rent	75,717	Single parents	−0.217
0.056	−0.165	Gross rent	75,717	No vehicle	−0.040
0.090	0.012	Gross rent	75,717	Crowding	0.057
−0.317	−0.341	Gross rent	75,717	No plumbing	−0.328
−0.059	−0.131	Gross rent	75,717	Under 9 years of school	−0.092
0.781	0.696	Mortgage cost	75,717	Family income	0.739
0.870	0.847	Mortgage cost	75,717	Home value	0.858
0.758	0.706	Mortgage cost	75,717	Gross rent	0.734
n/a	n/a	Mortgage cost	n/a	Mortgage cost	1
0.648	0.571	Mortgage cost	75,717	White-collar jobs	0.609
0.052	−0.091	Mortgage cost	75,717	Homeownership	−0.024
−0.337	−0.415	Mortgage cost	75,717	Poverty	−0.376
−0.413	−0.496	Mortgage cost	75,717	Below 150% of poverty	−0.455
−0.274	−0.341	Mortgage cost	75,717	Income disparity	−0.307
−0.056	−0.143	Mortgage cost	75,717	Unemployment	−0.100
−0.270	−0.317	Mortgage cost	75,717	Single parents	−0.294
0.215	−0.054	Mortgage cost	75,717	No vehicle	0.099
0.143	0.009	Mortgage cost	75,717	Crowding	0.087
−0.292	−0.320	Mortgage cost	75,717	No plumbing	−0.305
−0.023	−0.149	Mortgage cost	75,717	Under 9 years of school	−0.082
0.796	0.774	White-collar jobs	75,717	Family income	0.786
0.593	0.518	White-collar jobs	75,717	Home value	0.553
0.551	0.490	White-collar jobs	75,717	Gross rent	0.522
0.648	0.571	White-collar jobs	75,717	Mortgage cost	0.609
n/a	n/a	White-collar jobs	n/a	White-collar jobs	1
0.224	0.143	White-collar jobs	75,717	Homeownership	0.184
−0.485	−0.517	White-collar jobs	75,717	Poverty	−0.501
−0.590	−0.618	White-collar jobs	75,717	Below 150% of poverty	−0.605
−0.317	−0.357	White-collar jobs	75,717	Income disparity	−0.337
−0.250	−0.286	White-collar jobs	75,717	Unemployment	−0.268
−0.440	−0.481	White-collar jobs	75,717	Single parents	−0.461
0.033	−0.184	White-collar jobs	75,717	No vehicle	−0.069
−0.304	−0.384	White-collar jobs	75,717	Crowding	−0.346
−0.249	−0.282	White-collar jobs	75,717	No plumbing	−0.265
−0.463	−0.516	White-collar jobs	75,717	Under 9 years of school	−0.490
0.369	0.284	Homeownership	75,717	Family income	0.326
0.036	−0.120	Homeownership	75,717	Home value	−0.050
0.113	0.022	Homeownership	75,717	Gross rent	0.064
0.052	−0.091	Homeownership	75,717	Mortgage cost	−0.024
0.224	0.143	Homeownership	75,717	White-collar jobs	0.184
n/a	n/a	Homeownership	n/a	Homeownership	1
−0.488	−0.523	Homeownership	75,717	Poverty	−0.505
−0.496	−0.537	Homeownership	75,717	Below 150% of poverty	−0.516
−0.345	−0.375	Homeownership	75,717	Income disparity	−0.359
−0.243	−0.285	Homeownership	75,717	Unemployment	−0.264
−0.323	−0.376	Homeownership	75,717	Single parents	−0.349
−0.516	−0.575	Homeownership	75,717	No vehicle	−0.542
−0.349	−0.433	Homeownership	75,717	Crowding	−0.394
0.037	−0.005	Homeownership	75,717	No plumbing	0.017
−0.265	−0.349	Homeownership	75,717	Under 9 years of school	−0.306
−0.595	−0.618	Poverty	75,717	Family income	−0.606
−0.271	−0.358	Poverty	75,717	Home value	−0.313
−0.373	−0.408	Poverty	75,717	Gross rent	−0.391
−0.337	−0.415	Poverty	75,717	Mortgage cost	−0.376
−0.485	−0.517	Poverty	75,717	White-collar jobs	−0.501
−0.488	−0.523	Poverty	75,717	Homeownership	−0.505
n/a	n/a	Poverty	n/a	Poverty	1
0.914	0.907	Poverty	75,717	Below 150% of poverty	0.911
0.593	0.571	Poverty	75,717	Income disparity	0.582
0.431	0.392	Poverty	75,717	Unemployment	0.412
0.500	0.469	Poverty	75,717	Single parents	0.486
0.472	0.322	Poverty	75,717	No vehicle	0.381
0.331	0.266	Poverty	75,717	Crowding	0.297
0.323	0.279	Poverty	75,717	No plumbing	0.301
0.397	0.345	Poverty	75,717	Under 9 years of school	0.371
−0.696	−0.715	Below 150% of poverty	75,717	Family income	−0.705
−0.339	−0.429	Below 150% of poverty	75,717	Home value	−0.381
−0.434	−0.472	Below 150% of poverty	75,717	Gross rent	−0.453
−0.413	−0.496	Below 150% of poverty	75,717	Mortgage cost	−0.455
−0.590	−0.618	Below 150% of poverty	75,717	White-collar jobs	−0.605
−0.496	−0.537	Below 150% of poverty	75,717	Homeownership	−0.516
0.914	0.907	Below 150% of poverty	75,717	Poverty	0.911
n/a	n/a	Below 150% of poverty	n/a	Below 150% of poverty	1
0.583	0.557	Below 150% of poverty	75,717	Income disparity	0.570
0.434	0.396	Below 150% of poverty	75,717	Unemployment	0.415
0.541	0.511	Below 150% of poverty	75,717	Single parents	0.527
0.461	0.295	Below 150% of poverty	75,717	No vehicle	0.363
0.381	0.315	Below 150% of poverty	75,717	Crowding	0.348
0.343	0.301	Below 150% of poverty	75,717	No plumbing	0.322
0.467	0.417	Below 150% of poverty	75,717	Under 9 years of school	0.443
−0.425	−0.460	Income disparity	75,717	Family income	−0.442
−0.217	−0.295	Income disparity	75,717	Home value	−0.252
−0.341	−0.383	Income disparity	75,717	Gross rent	−0.361
−0.274	−0.341	Income disparity	75,717	Mortgage cost	−0.307
−0.317	−0.357	Income disparity	75,717	White-collar jobs	−0.337
−0.345	−0.375	Income disparity	75,717	Homeownership	−0.359
0.593	0.571	Income disparity	75,717	Poverty	0.582
0.583	0.557	Income disparity	75,717	Below 150% of poverty	0.570
n/a	n/a	Income disparity	n/a	Income disparity	1
0.280	0.245	Income disparity	75,717	Unemployment	0.263
0.270	0.238	Income disparity	75,717	Single parents	0.254
0.361	0.259	Income disparity	75,717	No vehicle	0.295
0.127	0.087	Income disparity	75,717	Crowding	0.107
0.246	0.216	Income disparity	75,717	No plumbing	0.231
0.194	0.157	Income disparity	75,717	Under 9 years of school	0.175
−0.268	−0.304	Unemployment	75,717	Family income	−0.286
−0.036	−0.134	Unemployment	75,717	Home value	−0.084
−0.061	−0.117	Unemployment	75,717	Gross rent	−0.089
−0.056	−0.143	Unemployment	75,717	Mortgage cost	−0.100
−0.250	−0.286	Unemployment	75,717	White-collar jobs	−0.268
−0.243	−0.285	Unemployment	75,717	Homeownership	−0.264
0.431	0.392	Unemployment	75,717	Poverty	0.412
0.434	0.396	Unemployment	75,717	Below 150% of poverty	0.415
0.280	0.245	Unemployment	75,717	Income disparity	0.263
n/a	n/a	Unemployment	n/a	Unemployment	1
0.308	0.271	Unemployment	75,717	Single parents	0.290
0.296	0.219	Unemployment	75,717	No vehicle	0.251
0.234	0.181	Unemployment	75,717	Crowding	0.209
0.185	0.122	Unemployment	75,717	No plumbing	0.152
0.246	0.197	Unemployment	75,717	Under 9 years of school	0.222
−0.477	−0.504	Single parents	75,717	Family income	−0.491
−0.254	−0.297	Single parents	75,717	Home value	−0.274
−0.196	−0.236	Single parents	75,717	Gross rent	−0.217
−0.270	−0.317	Single parents	75,717	Mortgage cost	−0.294
−0.440	−0.481	Single parents	75,717	White-collar jobs	−0.461
−0.323	−0.376	Single parents	75,717	Homeownership	−0.349
0.500	0.469	Single parents	75,717	Poverty	0.486
0.541	0.511	Single parents	75,717	Below 150% of poverty	0.527
0.270	0.238	Single parents	75,717	Income disparity	0.254
0.308	0.271	Single parents	75,717	Unemployment	0.290
n/a	n/a	Single parents	n/a	Single parents	1
0.242	0.094	Single parents	75,717	No vehicle	0.157
0.330	0.260	Single parents	75,717	Crowding	0.296
0.128	0.087	Single parents	75,717	No plumbing	0.108
0.366	0.315	Single parents	75,717	Under 9 years of school	0.341
−0.034	−0.262	No vehicle	75,717	Family income	−0.145
0.262	−0.075	No vehicle	75,717	Home value	0.110
0.056	−0.165	No vehicle	75,717	Gross rent	−0.040
0.215	−0.054	No vehicle	75,717	Mortgage cost	0.099
0.033	−0.184	No vehicle	75,717	White-collar jobs	−0.069
−0.516	−0.575	No vehicle	75,717	Homeownership	−0.542
0.472	0.322	No vehicle	75,717	Poverty	0.381
0.461	0.295	No vehicle	75,717	Below 150% of poverty	0.363
0.361	0.259	No vehicle	75,717	Income disparity	0.295
0.296	0.219	No vehicle	75,717	Unemployment	0.251
0.242	0.094	No vehicle	75,717	Single parents	0.157
n/a	n/a	No vehicle	n/a	No vehicle	1
0.311	0.154	No vehicle	75,717	Crowding	0.227
0.164	0.057	No vehicle	75,717	No plumbing	0.101
0.266	0.161	No vehicle	75,717	Under 9 years of school	0.211
−0.217	−0.283	Crowding	75,717	Family income	−0.252
0.153	0.018	Crowding	75,717	Home value	0.100
0.090	0.012	Crowding	75,717	Gross rent	0.057
0.143	0.009	Crowding	75,717	Mortgage cost	0.087
−0.304	−0.384	Crowding	75,717	White-collar jobs	−0.346
−0.349	−0.433	Crowding	75,717	Homeownership	−0.394
0.331	0.266	Crowding	75,717	Poverty	0.297
0.381	0.315	Crowding	75,717	Below 150% of poverty	0.348
0.127	0.087	Crowding	75,717	Income disparity	0.107
0.234	0.181	Crowding	75,717	Unemployment	0.209
0.330	0.260	Crowding	75,717	Single parents	0.296
0.311	0.154	Crowding	75,717	No vehicle	0.227
n/a	n/a	Crowding	n/a	Crowding	1
0.041	−0.021	Crowding	75,717	No plumbing	0.007
0.698	0.571	Crowding	75,717	Under 9 years of school	0.636
−0.269	−0.297	No plumbing	75,717	Family income	−0.282
−0.247	−0.281	No plumbing	75,717	Home value	−0.262
−0.317	−0.341	No plumbing	75,717	Gross rent	−0.328
−0.292	−0.320	No plumbing	75,717	Mortgage cost	−0.305
−0.249	−0.282	No plumbing	75,717	White-collar jobs	−0.265
0.037	−0.005	No plumbing	75,717	Homeownership	0.017
0.323	0.279	No plumbing	75,717	Poverty	0.301
0.343	0.301	No plumbing	75,717	Below 150% of poverty	0.322
0.246	0.216	No plumbing	75,717	Income disparity	0.231
0.185	0.122	No plumbing	75,717	Unemployment	0.152
0.128	0.087	No plumbing	75,717	Single parents	0.108
0.164	0.057	No plumbing	75,717	No vehicle	0.101
0.041	−0.021	No plumbing	75,717	Crowding	0.007
n/a	n/a	No plumbing	n/a	No plumbing	1
0.100	0.046	No plumbing	75,717	Under 9 years of school	0.071
−0.345	−0.384	Under 9 years of school	75,717	Family income	−0.365
−0.011	−0.145	Under 9 years of school	75,717	Home value	−0.071
−0.059	−0.131	Under 9 years of school	75,717	Gross rent	−0.092
−0.023	−0.149	Under 9 years of school	75,717	Mortgage cost	−0.082
−0.463	−0.516	Under 9 years of school	75,717	White-collar jobs	−0.490
−0.265	−0.349	Under 9 years of school	75,717	Homeownership	−0.306
0.397	0.345	Under 9 years of school	75,717	Poverty	0.371
0.467	0.417	Under 9 years of school	75,717	Below 150% of poverty	0.443
0.194	0.157	Under 9 years of school	75,717	Income disparity	0.175
0.246	0.197	Under 9 years of school	75,717	Unemployment	0.222
0.366	0.315	Under 9 years of school	75,717	Single parents	0.341
0.266	0.161	Under 9 years of school	75,717	No vehicle	0.211
0.698	0.571	Under 9 years of school	75,717	Crowding	0.636
0.100	0.046	Under 9 years of school	75,717	No plumbing	0.071
n/a	n/a	Under 9 years of school	n/a	Under 9 years of school	1

The four dollar measures and white-collar work correlate strongly with one another, the hardship measures mostly correlate positively with one another, and the two groups correlate negatively; the home value row stays modest against the hardship columns. Each cell is the correlation between two measures across tracts, weighted by tract population, with a 95% interval from a bootstrap over counties. Rows and columns share one order, the measures of advantage first.Note: Cells toward the teal end of the key are positive correlations and cells toward the brown end negative ones. The diagonal is one by construction and has no interval.Note: The correlations use the measures as the ACS reports them: dollars for the four dollar measures, proportions for the rest and the log ratio for income disparity. The staged ACS file has no margins of error.Source: ACS 2019-2023 five-year tract estimates (staged ACS file and Census API pull of housing measures); population-weighted (ACS tract population)

In raw units, home value and family income set the score

Write the score of tract $i$ as a sum over the measures $k$ of a coefficient $c_k$ times the measure $x_{ik}$ . The Atlas uses the measures as they come. A standardized version subtracts each measure's mean $\bar x_k$ and divides by its standard deviation $s_k$ :

S_i=\sum_{k} c_k\,x_{ik} \qquad\text{and}\qquad S_i^{z}=\sum_{k} c_k\,\frac{x_{ik}-\bar x_k}{s_k}.

The first is the replica of the published score. The second applies the same coefficients, with the population-weighted mean and standard deviation of each measure across tracts. A sum is dominated by the terms that vary most, and variance is measured in squared units, so a measure with values in the hundreds of thousands contributes more variance than a proportion does. Figure 3 splits the variance of each score among its measures. A measure's allocated share is the covariance of its term with the score divided by the variance of the score, and each covariance between two terms is shared equally between them, so the shares add to the whole of the variance. In the raw-units score, median home value is allocated 85.4% (95% CI 83.1% to 87.0%) and median family income 14.2% (95% CI 12.6% to 16.5%). The two terms are correlated, so the allocation matters: the home value term alone has 74.3% (95% CI 70.5% to 77.0%) of the variance, and the home value and family income terms together have 99.2% (95% CI 99.2% to 99.3%). Rent and the mortgage cost are allocated a further 0.38% (95% CI 0.36% to 0.41%), and the measures that are not dollar amounts 0.0003% (95% CI 0.0002% to 0.0003%) together. After standardization, no measure is allocated more than 12.4% (95% CI 12.2% to 12.6%), which is the near-poverty share, and home value is allocated 5.3% (95% CI 5.0% to 5.7%).

Figure 3

In raw units home value accounts for 74.3% of the variance of the score on its own and 85.4% with its covariance with income shared out, against 5.3% after standardization

Share of the population-weighted variance of the replica score allocated to each measure, same coefficients, measures in raw units and in standard deviation units, 75,717 tracts

In raw units home value and family income carry 99.2% of the variance of the score between them

Share of the variance of the score in raw units allocated to each measure, percent

Measure

Home value's bar is the longest and family income's the second; every other bar is nearly zero.

Show data

Data for In raw units home value accounts for 74.3% of the variance of the score on its own and 85.4% with its covariance with income shared out, against 5.3% after standardization, Raw units
conf_high	conf_low	measure	n	share
86.99	83.08	Home value	75,717	85.40
16.52	12.64	Family income	75,717	14.23
0.245	0.213	Mortgage cost	75,717	0.226
0.164	0.143	Gross rent	75,717	0.152
0.000	0.000	Income disparity	75,717	0.000
0.000	0.000	White-collar jobs	75,717	0.000
0.000	0.000	Below 150% of poverty	75,717	0.000
0.000	0.000	Poverty	75,717	0.000
0.000	0.000	Single parents	75,717	0.000
0.000	0.000	Under 9 years of school	75,717	0.000
0.000	0.000	No plumbing	75,717	0.000
0.000	0.000	Unemployment	75,717	0.000
0.000	0.000	Homeownership	75,717	0.000
0.000	0.000	Crowding	75,717	0.000
0.000	0.000	No vehicle	75,717	0.000

In raw units two bars, home value and family income, hold 99.2% of the variance between them, and after standardization the bars are spread over the measures. A bar is the covariance of a measure's term in the score with the score, divided by the variance of the score, so each covariance between two terms is shared equally between them and the bars add to the whole of the variance; whiskers are 95% intervals from a bootstrap over counties. The tabs share one order and one axis.Note: Raw units: dollars for the four dollar measures, proportions between zero and one for the rest, and the log ratio for income disparity, as the Atlas applies Singh's coefficients. Standardized: each measure minus its population-weighted mean over its population-weighted standard deviation.Note: The two measures that are not rebuilt (high school diploma and internet access) are proportions, so in raw units they would add little to the variance of the score.Source: ACS 2019-2023 five-year tract estimates (staged ACS file and Census API pull of housing measures); Singh (2003) coefficients; population-weighted (ACS tract population)

The same coefficients therefore give weights that depend on the scale of the measures. Petterson showed the pattern across the block groups of the Atlas, and Hannan and colleagues found that in New York State the Atlas ADI mainly reflects home value.[1, 2] The shares here describe a population-weighted frame of tracts, a different population from either study.

The replica reproduces the published rank

Ranking the replica's score among the frame's tracts gives a population-weighted rank correlation of +0.991 (95% CI 0.990 to 0.992) with the published tract rank. The match is close and not exact, because the replica uses tract estimates where the Atlas uses block groups, omits two measures and does not impute missing values. Standardizing the measures before applying the same coefficients lowers the correlation with the published rank to +0.72 (95% CI 0.66 to 0.77). The published ADI correlates −0.988 (95% CI −0.989 to −0.986) with the rank of median home value. The raw-units replica correlates −0.987 (95% CI −0.989 to −0.986) with it and the standardized replica −0.66 (95% CI −0.71 to −0.60).

The scale of the measures moves the ranking, and the choice of coefficients barely does. The first principal component of the standardized measures lets the data choose the weights, and it ranks the tracts almost as Singh's coefficients applied to the standardized measures do, with a rank correlation of +0.997 (95% CI 0.996 to 0.998).

The staged SVI file carries the two missing measures for ACS 2018 to 2022. Adding them to the replica, with the telephone measure's coefficient on households without internet because the Atlas does not publish one, moves the rank correlation of the standardized replica with the published rank to +0.71 (95% CI 0.65 to 0.77) and the share of residents of high-poverty tracts it places at or below the 50th percentile to 0.61% (95% CI 0.32% to 0.93%). The raw-units replica stays at +0.991 (95% CI 0.990 to 0.992).

The published ADI sits apart from rankings that standardize

Figure 4 compares six rankings of the same tracts: the published ADI, its raw-units replica, the standardized replica, the first principal component of the standardized measures, the socioeconomic theme of the Social Vulnerability Index (SVI) of the CDC and ATSDR, and the poverty rate.[6, 7] The SVI converts each measure to a percentile among tracts before summing within a theme, so it is standardized by construction. The published ADI and its replica form one pair, and the standardized replica and the principal component form another. The principal component, the SVI theme and the poverty rate each correlate more strongly with the standardized replica than with the published ADI.

Figure 4

The published ADI correlates +0.991 with its raw-units replica and +0.72 with the standardized version of the same measures

Population-weighted rank correlations among six rankings of 2020 tracts, measures from ACS 2019 to 2023

0.401.00

A six by six grid of rank correlations. The published ADI and its raw-units replica are almost perfectly correlated, the standardized replica and the principal component are almost perfectly correlated, and the SVI theme and poverty rate correlate strongly with the standardized rankings and only moderately with the published ADI.
Ranking	Published ADI	Replica, raw units	Replica, standardized	First principal component	SVI socioeconomic theme	Poverty rate
Published ADI	1.00	0.99, interval 0.99 to 0.99, n = 75,717	0.72, interval 0.66 to 0.77, n = 75,717	0.76, interval 0.70 to 0.80, n = 75,717	0.48, interval 0.41 to 0.56, n = 75,717	0.54, interval 0.49 to 0.59, n = 75,717
Replica, raw units	0.99, interval 0.99 to 0.99, n = 75,717	1.00	0.74, interval 0.68 to 0.79, n = 75,717	0.77, interval 0.72 to 0.82, n = 75,717	0.51, interval 0.43 to 0.57, n = 75,717	0.55, interval 0.50 to 0.59, n = 75,717
Replica, standardized	0.72, interval 0.66 to 0.77, n = 75,717	0.74, interval 0.68 to 0.79, n = 75,717	1.00	1.00, interval 1.00 to 1.00, n = 75,717	0.86, interval 0.85 to 0.87, n = 75,717	0.85, interval 0.84 to 0.86, n = 75,717
First principal component	0.76, interval 0.70 to 0.80, n = 75,717	0.77, interval 0.72 to 0.82, n = 75,717	1.00, interval 1.00 to 1.00, n = 75,717	1.00	0.84, interval 0.83 to 0.85, n = 75,717	0.84, interval 0.84 to 0.85, n = 75,717
SVI socioeconomic theme	0.48, interval 0.41 to 0.56, n = 75,717	0.51, interval 0.43 to 0.57, n = 75,717	0.86, interval 0.85 to 0.87, n = 75,717	0.84, interval 0.83 to 0.85, n = 75,717	1.00	0.76, interval 0.75 to 0.77, n = 75,717
Poverty rate	0.54, interval 0.49 to 0.59, n = 75,717	0.55, interval 0.50 to 0.59, n = 75,717	0.85, interval 0.84 to 0.86, n = 75,717	0.84, interval 0.84 to 0.85, n = 75,717	0.76, interval 0.75 to 0.77, n = 75,717	1.00

Show data

Data for The published ADI correlates +0.991 with its raw-units replica and +0.72 with the standardized version of the same measures
conf_high	conf_low	n	other	r	ranking
n/a	n/a	n/a	Published ADI	1	Published ADI
0.992	0.990	75,717	Replica, raw units	0.991	Published ADI
0.773	0.659	75,717	Replica, standardized	0.720	Published ADI
0.800	0.703	75,717	First principal component	0.755	Published ADI
0.556	0.406	75,717	SVI socioeconomic theme	0.485	Published ADI
0.587	0.486	75,717	Poverty rate	0.539	Published ADI
0.992	0.990	75,717	Published ADI	0.991	Replica, raw units
n/a	n/a	n/a	Replica, raw units	1	Replica, raw units
0.789	0.680	75,717	Replica, standardized	0.738	Replica, raw units
0.817	0.723	75,717	First principal component	0.773	Replica, raw units
0.575	0.429	75,717	SVI socioeconomic theme	0.506	Replica, raw units
0.595	0.496	75,717	Poverty rate	0.548	Replica, raw units
0.773	0.659	75,717	Published ADI	0.720	Replica, standardized
0.789	0.680	75,717	Replica, raw units	0.738	Replica, standardized
n/a	n/a	n/a	Replica, standardized	1	Replica, standardized
0.998	0.996	75,717	First principal component	0.997	Replica, standardized
0.866	0.852	75,717	SVI socioeconomic theme	0.859	Replica, standardized
0.859	0.843	75,717	Poverty rate	0.851	Replica, standardized
0.800	0.703	75,717	Published ADI	0.755	First principal component
0.817	0.723	75,717	Replica, raw units	0.773	First principal component
0.998	0.996	75,717	Replica, standardized	0.997	First principal component
n/a	n/a	n/a	First principal component	1	First principal component
0.850	0.832	75,717	SVI socioeconomic theme	0.841	First principal component
0.853	0.836	75,717	Poverty rate	0.844	First principal component
0.556	0.406	75,717	Published ADI	0.485	SVI socioeconomic theme
0.575	0.429	75,717	Replica, raw units	0.506	SVI socioeconomic theme
0.866	0.852	75,717	Replica, standardized	0.859	SVI socioeconomic theme
0.850	0.832	75,717	First principal component	0.841	SVI socioeconomic theme
n/a	n/a	n/a	SVI socioeconomic theme	1	SVI socioeconomic theme
0.769	0.753	75,717	Poverty rate	0.761	SVI socioeconomic theme
0.587	0.486	75,717	Published ADI	0.539	Poverty rate
0.595	0.496	75,717	Replica, raw units	0.548	Poverty rate
0.859	0.843	75,717	Replica, standardized	0.851	Poverty rate
0.853	0.836	75,717	First principal component	0.844	Poverty rate
0.769	0.753	75,717	SVI socioeconomic theme	0.761	Poverty rate
n/a	n/a	n/a	Poverty rate	1	Poverty rate

Two blocks stand out: the published ADI with its raw-units replica at top left, and the standardized score with its principal component, the SVI theme and the poverty rate. Each cell is the correlation between the percentile ranks of two rankings across tracts, weighted by tract population, with a 95% interval from a bootstrap over counties. The replicas sum Singh's coefficients times the measures, in raw units or in standard deviation units; the principal component is the first of the standardized measures.Note: Each cell uses the tracts that have both rankings; pairs with the SVI socioeconomic theme use the tracts that also have an SVI value, and each cell's tooltip gives its tract count.Note: The ranks are percentiles among those tracts, and each cell is the population-weighted Pearson correlation of the two percentile ranks (Spearman's rank correlation would count each tract once); the published ADI is the Atlas's block-group ranks averaged over each tract.Source: Neighborhood Atlas ADI 2023; ACS 2019-2023; CDC/ATSDR Social Vulnerability Index 2022; population-weighted (ACS tract population) rank correlations

The ADI reads expensive places as less deprived

The disagreement between the published and the standardized ranking follows the cost of housing. Figure 5 groups tracts by the median home value of their county, in quartiles with equal numbers of residents. County median home values are above $450,562 in the highest-cost quartile and below $216,301 in the lowest-cost quartile. The published percentile minus the standardized percentile averages +11.1 (95% CI 10.5 to 11.6) points in the lowest-cost counties and −16.5 (95% CI −21.3 to −11.5) points in the highest-cost counties. At the same standardized percentile, tracts in the highest-cost counties have a lower published ADI than tracts in the lowest-cost counties in every tenth of the standardized ranking, and the gap widens toward the most deprived tenth.

The gap matters most where people are poor. 5,100 tracts in the frame, with 17.2 million residents, have poverty rates of 30% or more. The published ADI places 16.1% (95% CI 11.3% to 21.4%) of those residents in tracts ranked at or below the 50th national percentile, and the raw-units replica places 15.8% (95% CI 11.0% to 20.7%). The standardized replica places 0.34% (95% CI 0.15% to 0.55%). Across all 6,718 tracts of the 50 states and DC that have a published rank and a poverty rate, including tracts that lack some rebuilt measures, the published share is 20.6% (95% CI 14.1% to 27.9%).

The second tab of the figure lists the 15 tracts with the largest gaps among the 5,049 tracts that have a poverty rate of 30% or more and at least 1,000 residents. 12 are in New York City. In the first row, a tract in San Francisco where 35.9% of residents are below the poverty line, the published ADI is 1, among the least deprived in the nation, and the standardized percentile is 99.7. The median home value of that tract is at the ACS top code, which applies to 495 tracts in the frame and caps a dollar value that would otherwise be higher.

Figure 5

The published ADI percentile minus the standardized percentile averages −16.5 in the highest-cost counties and +11.1 in the lowest-cost counties

Tract percentile on the published ADI against the percentile of the standardized replica score, by the cost of the tract's county, and the tracts with the largest gaps

At the same standardized percentile, tracts in the highest-cost counties have a lower published ADI than tracts in the lowest-cost counties

Published ADI against the standardized percentile, one dot per sampled tract, colored by the median home value of the tract's county

Loading data

The line of binned means for the lowest-cost counties lies above the line for the highest-cost counties in every tenth of the standardized ranking, and the gap widens toward the most deprived tenth.

At the same standardized percentile the binned means of the published ADI are higher in the lowest-cost counties than in the highest-cost counties in every tenth, and the 95% intervals of the two groups do not overlap. The two rankings correlate at +0.72, so the diagonal of equal percentiles is not a null for a binned mean, and the evidence is the gap between the groups. The first tab plots a seeded sample of 3,000 tracts with population-weighted binned means; the second lists the 15 tracts with a poverty rate of at least 30% and at least 1,000 residents where the published percentile falls furthest below the standardized one, of which 12 are in New York City.Note: Counties are grouped by their median home value into quartiles with equal numbers of residents; the middle two quartiles are drawn together in gray. Markers are population-weighted means of the published ADI within tenths of the standardized percentile for the lowest-cost and highest-cost groups, with 95% intervals from a bootstrap over counties; cells with few tracts are left out.Note: The dots are a random sample of the frame's tracts drawn with a fixed seed (seed and share in the methods note); the means, intervals and the table use every tract.Note: A tract's ACS medians rest on the homes and renters in the tract, which are few in some tracts, and the staged file has no margins of error; a college-town tract can show a high poverty rate because of student households. Each row illustrates what the formula does and is not an estimate of its tract's deprivation.Source: Neighborhood Atlas ADI 2023; ACS 2019-2023; population-weighted (ACS tract population) binned means, modeled estimates

One Medicare payment model, ACO REACH, uses the ADI. The model adjusts payments for each beneficiary by the decile of a score that combines the ADI with the beneficiary's status as dually eligible for Medicare and Medicaid or as receiving the low-income subsidy. For performance year 2024 the Centers for Medicare and Medicaid Services added the state ADI to the score, weighted the national ADI, the state ADI and that status equally, and said the change would better identify underserved beneficiaries in areas with a high cost of living.[8] The state ADI compares tracts within a state, and this chapter does not test how much of the dominance of home value it removes.

The measures describe more than one kind of deprivation

The correlation matrix behind Figure 2 has 3 eigenvalues above the average, the usual criterion for the number of factors worth keeping, and the first 3 principal components carry 66.0% (95% CI 65.3% to 66.9%) of the variance of the measures, the first alone 39.6% (95% CI 38.4% to 41.1%). A factor model describes the correlations among many measures with a few unobserved factors, and a measure's loading is its correlation with a factor. A maximum-likelihood factor model of the same population-weighted correlation matrix, with 3 factors and a varimax rotation that turns the factors so that each measure loads mainly on one, sorts the measures into 3 groups (Figure 6). The first, housing money, gathers home value, mortgage cost, rent, family income and white-collar work, and holds 24.7% of the variance of the measures. The second, economic hardship, gathers the poverty rate, the near-poverty share, income disparity, unemployment and single parenthood, with homeownership on the other side, and holds 21.5%. The third gathers crowding and the share of adults who left school before ninth grade, which correlate +0.64 (95% CI 0.57 to 0.70) with each other, and holds 11.9%.

Figure 6

The measures sort into housing money, economic hardship, and crowding with little schooling, which together hold 58.1% of their variance

Varimax-rotated loadings of the 15 standardized measures on 3 maximum-likelihood factors, 75,717 tracts, ACS 2019 to 2023

Home value, mortgage cost, family income, rent and white-collar work load on the housing money factor

Loadings on the housing money factor, which holds 24.7% of the variance of the measures

Measure

The five longest bars are the dollar measures and white-collar work.

Show data

Data for The measures sort into housing money, economic hardship, and crowding with little schooling, which together hold 58.1% of their variance, Housing money
loading	measure
0.930	Mortgage cost
0.900	Home value
0.750	Gross rent
0.750	Family income
0.640	White-collar jobs
−0.370	Below 150% of poverty
−0.290	No plumbing
−0.270	Single parents
−0.270	Poverty
−0.250	Income disparity
0.160	No vehicle
0.100	Crowding
−0.090	Homeownership
−0.080	Under 9 years of school
−0.060	Unemployment

Each tab is one factor, and a long bar means the measure moves with that factor; bars pointing left load negatively. Housing money gathers the dollar measures and white-collar work, economic hardship gathers poverty and its relatives, and the third gathers crowding and little schooling. Loadings describe all the tracts in the frame and carry no interval.Note: Maximum-likelihood factor analysis of the population-weighted correlation matrix of the measures (the matrix in the first correlation figure), 3 factors chosen by the Kaiser criterion, varimax rotation. The fit starts from seeded random loadings and the solution does not depend on the seed. Each factor is signed so its largest loading is positive.Note: The factors describe these measures, not deprivation in general: other measures would give other factors.Source: ACS 2019-2023 five-year tract estimates (staged ACS file and Census API pull of housing measures); population-weighted (ACS tract population) correlation matrix

Kolak and colleagues, using a different set of tract measures, also needed several principal components and named the first socioeconomic advantage.[9] A single ranking folds these factors into one number. In raw units the number follows the housing-money factor, as the correlation with home value shows.

Standardized rankings track upward mobility more closely than the published ADI does, and life expectancy about as closely

Two outcomes that no ranking used as an input test the recipes. One is life expectancy at birth from USALEEP, for deaths in 2010 to 2015 in 71,874 tracts. The other is the mean adult household income rank of children raised in the tract by parents at the 25th income percentile, from the Opportunity Atlas, for children born in 1978 to 1983 in 75,667 tracts. Figure 7 shows the rank correlation of eight rankings with each outcome. Every ranking runs from least to most deprived, so every correlation is negative, and the figure plots absolute values. Comparisons below that rest on point estimates say so, and the others use intervals from paired draws of the same counties.

For life expectancy the published ADI correlates 0.615 (95% CI 0.594 to 0.633) and the standardized replica 0.619 (95% CI 0.596 to 0.642). The difference, 0.005 (95% CI −0.015 to 0.027), has an interval that includes zero, and so does the difference between family income alone, at 0.609 (95% CI 0.587 to 0.630), and the ADI, −0.006 (95% CI −0.022 to 0.012). The standardized replica exceeds family income alone by 0.010 (95% CI 0.002 to 0.020), and the first principal component, which has the highest point estimate at 0.633 (95% CI 0.613 to 0.653), exceeds it by 0.025 (95% CI 0.019 to 0.031). Both intervals exclude zero, by small margins. The raw-units replica is at 0.620 (95% CI 0.601 to 0.638), and the poverty rate alone has the lowest point estimate, 0.508 (95% CI 0.484 to 0.529).

For upward mobility the order changes. The standardized replica correlates 0.582 (95% CI 0.560 to 0.603) and the published ADI 0.483 (95% CI 0.455 to 0.511), a difference of 0.100 (95% CI 0.076 to 0.123). The raw-units replica correlates 0.492 (95% CI 0.466 to 0.520) and home value alone 0.443 (95% CI 0.415 to 0.473). Family income alone, at 0.564 (95% CI 0.543 to 0.586), correlates more strongly than the published ADI does, by 0.082 (95% CI 0.063 to 0.101), and less strongly than the standardized replica, which exceeds it by 0.018 (95% CI 0.010 to 0.026), and than the first principal component, which exceeds it by 0.023 (95% CI 0.017 to 0.029). The rankings that standardize therefore edge out family income alone for both outcomes, and the published ADI does not for either. The poverty rate alone, at 0.504 (95% CI 0.486 to 0.522), does not differ detectably from the ADI, by 0.021 (95% CI −0.002 to 0.044).

Figure 7

The published ADI tracks life expectancy about as closely as the standardized score does but upward mobility less closely, by 0.100 in correlation

Absolute population-weighted rank correlation of eight rankings with tract life expectancy at birth and with the adult income rank of children from low-income families, 71,874 and 75,667 tracts

Life expectancy at birth
Upward mobility

Show data

Data for The published ADI tracks life expectancy about as closely as the standardized score does but upward mobility less closely, by 0.100 in correlation
conf_high	conf_low	estimate	n	outcome	ranking
0.653	0.613	0.633	71,874	Life expectancy at birth	First principal component
0.607	0.566	0.587	75,667	Upward mobility	First principal component
0.638	0.601	0.620	71,874	Life expectancy at birth	Replica, raw units
0.520	0.466	0.492	75,667	Upward mobility	Replica, raw units
0.642	0.596	0.619	71,874	Life expectancy at birth	Replica, standardized
0.603	0.560	0.582	75,667	Upward mobility	Replica, standardized
0.633	0.594	0.615	71,874	Life expectancy at birth	Published ADI
0.511	0.455	0.483	75,667	Upward mobility	Published ADI
0.630	0.587	0.609	71,874	Life expectancy at birth	Family income
0.586	0.543	0.564	75,667	Upward mobility	Family income
0.605	0.563	0.584	71,874	Life expectancy at birth	Home value
0.473	0.415	0.443	75,667	Upward mobility	Home value
0.553	0.485	0.518	71,874	Life expectancy at birth	SVI socioeconomic theme
0.571	0.519	0.545	75,667	Upward mobility	SVI socioeconomic theme
0.529	0.484	0.508	71,874	Life expectancy at birth	Poverty rate
0.522	0.486	0.504	75,667	Upward mobility	Poverty rate

The dots for life expectancy bunch together across the composite rankings and family income. For upward mobility the published ADI, its raw replica and home value alone have the lowest point estimates, all below family income alone, and the poverty rate alone is not distinguishable from the published ADI. Each dot is the absolute value of the correlation between a ranking and an outcome across tracts, weighted by tract population, with a 95% interval from a bootstrap over counties; every ranking runs from least to most deprived, so both correlations are negative.Note: Both outcomes were modeled and neither was an input to any ranking. Each outcome uses the tracts that have it and every ranking, and institutional tracts are dropped from the life expectancy comparison. USALEEP covers deaths in 2010 to 2015 and its 2020-tract file is interpolated from 2010 tracts; the Opportunity Atlas covers children born in 1978 to 1983 and its 2020-tract file is interpolated from 2010 tracts.Note: Family income and home value enter as negative values so that a higher rank means more deprived. The SVI theme and the poverty rate are not Atlas products.Source: Neighborhood Atlas ADI 2023; ACS 2019-2023; CDC/ATSDR Social Vulnerability Index 2022; NCHS USALEEP 2010-2015; Opportunity Atlas 1978-1983 birth cohorts; population-weighted (ACS tract population) rank correlations, modeled estimates

What these data cannot show

Every result describes tracts and rankings of tracts. A tract's rank says little about how deprived any one resident is, because the data are averages for places (ecological inference), and a poor household in a tract with high home values takes the tract's rank.

The replica is a tract-level rebuild of a block-group index. It omits two measures, substitutes for some, does not impute missing values and uses ACS tract estimates that carry sampling error the staged file does not report, which is largest in small tracts and for medians that rest on few homes. A tract with many students living off campus can show a high poverty rate that reflects student households. The close match to the published rank, +0.991 (95% CI 0.990 to 0.992), shows that this formula reproduces the Atlas's ranks, and it does not show that the Atlas's code is the same. The frame leaves out tracts that lack a median, and the ADI ranks some of them. It also leaves out Puerto Rico, which the Atlas ranks with the states.

The data cannot say which recipe is right. A ranking dominated by home value measures the cost of housing, a property of a market, and some uses of a deprivation index may want that. The correlations with life expectancy and mobility are one test, and the two outcomes disagree about the recipes. Both outcomes are older than the ACS measures, deaths in 2010 to 2015 and children born in 1978 to 1983, and both are modeled estimates placed on 2020 tracts by interpolation from 2010 tracts. The SVI theme and the poverty rate are not Atlas products.

A factor model describes the measures it is given. Other measures, such as violence, isolation or air quality, would give other factors, and the count of factors rests on a common cutoff. The chapter does not test the state ADI that Medicare now uses, and it says nothing about how payments change under any formula.

Data sources

Area Deprivation IndexNeighborhood Atlas ADI, 2023 release (version 4.0.1), national percentile ranks of block groups averaged to 2020 tractsYears: 2023 release (version 4.0.1; built on ACS 2019-2023) · Weights: none (index); tract values are unweighted means of block-group ranks · License: Free for non-commercial use; cite the Neighborhood Atlas and Kind and Buckingham
American Community SurveyACS 2019-2023 five-year tract estimates: the staged tract file and a Census API pull of housing, occupation, income-distribution and schooling tablesYears: 2019-2023 (5-year) · Weights: ACS 2019-2023 tract population · License: Public domain (U.S. Census Bureau)
CDC/ATSDR Social Vulnerability IndexCDC/ATSDR Social Vulnerability Index 2022, tracts: the socioeconomic theme percentile and, for a sensitivity check, the shares of adults without a high school diploma and of households without internet (ACS 2018 to 2022)Years: 2022 release (ACS 2018-2022) · Weights: none (index); percentile ranks of tracts · License: Public domain (CDC/ATSDR)
USALEEP life expectancyNCHS small-area life expectancy at birth, deaths 2010 to 2015, 2020 tracts (interpolated), institutional tracts droppedYears: 2010-2015 · Weights: ACS 2019-2023 tract population · License: Public domain (NCHS, CDC)
Opportunity AtlasMean adult household income rank of children from low-income families, 1978 to 1983 birth cohorts, 2020 tracts (interpolated)Years: 1978-1983 birth cohorts (2018 release) · Weights: ACS 2019-2023 tract population · License: Public use (Opportunity Insights and the U.S. Census Bureau)

All data sources

Methods

This chapter rebuilds the Neighborhood Atlas Area Deprivation Index from ACS 2019-2023 tract estimates and compares the replica with the published rank, with standardized rankings and with a principal component, across 75,717 census tracts of the 50 states and DC. Fifteen of the Atlas's 17 measures are rebuilt, with substitutions for some, so the replica approximates the Atlas and is not its own file; the Puerto Rico tracts are left out. Life expectancy (deaths in 2010 to 2015) and upward mobility (children born 1978 to 1983) describe earlier periods than the ACS measures, and the ACS file carries no margins of error, so the only intervals come from resampling whole counties.

Which tracts are in the frame

The frame is every 2020 census tract of the 50 states and DC with a published ADI rank (the Atlas's block-group ranks averaged over the tract's block groups without population weights), a population above zero in the ACS 2019-2023 tract file and all 15 rebuilt measures: 75,717 of the 84,415 tracts of the 50 states and DC in the staged ADI file (adi_atlas_tract2020_2023, the Atlas's 2023 release, version 4.0.1, whose staged documentation records ACS 2019 to 2023 as its source data).

The 981 Puerto Rico tracts are left out: the SVI and USALEEP have none, and the Atlas ranks them together with the states, so percentiles of the replicas are among the tracts of the 50 states and DC.

How the measures are rebuilt

Fifteen of the Atlas's 17 measures are rebuilt from the staged ACS tract file (poverty, homeownership, unemployment and single-parent shares) and from a Census API pull of ACS tables (B25077, B25064, B25088, B19113, B19001, C17002, C24010, B25044, B25047, B25014 and B15003) archived under data/raw/acs/deprivation_components_2019_2023 with a manifest; b2.load has no loader for that pull, so the chapter registers the files with b2.load and reads them through load_staged.

Substitutions: persons below the poverty line for families, management, business, science and arts occupations for white-collar work, owner costs of homes with a mortgage for the monthly mortgage, adults who stopped at 8th grade for adults with under 9 years of education, and an income disparity of ln(100 x (households under $10,000 + 0.5) / (households of $50,000 or more + 0.5)), with the 0.5 added so that tracts with no household under $10,000 keep a value.

The two measures not rebuilt are adults with at least a high school diploma and households without internet. Both are proportions, and a sensitivity check adds them from the SVI 2022 file (ACS 2018 to 2022; the share without a diploma enters as one minus the share, with coefficient -0.0970, and the share without internet takes the telephone measure's coefficient, 0.0877, because the Atlas does not publish one for internet).

Proportions enter as numbers between zero and one, the convention of Petterson's Table 1 (0.41 for a block group with 41 percent below 150 percent of the poverty line); entered as percentages, home value would carry 85.4 percent and family income 14.2 percent of the variance of the raw-units score.

Figures explained here: Figure 1 and Figure 2

Outcomes and comparison indexes

Life expectancy is USALEEP for deaths in 2010 to 2015 on 2020 tracts (interpolated from 2010 tracts), institutional tracts dropped; mobility is the Opportunity Atlas rank of children of parents at the 25th income percentile (1978 to 1983 birth cohorts), interpolated to 2020 tracts.

The SVI socioeconomic theme is the 2022 CDC/ATSDR percentile rank (theme 1).

Figure explained here: Figure 7

How scores, rankings and groups are formed

Recipes: the replica score is the sum of Singh's factor score coefficients (Singh 2003, as printed in Petterson 2023, Table 1) times the measures in their own units; the standardized score uses the same coefficients on measures minus their population-weighted means over their population-weighted standard deviations; the principal component is the first eigenvector of the population-weighted correlation matrix.

Rankings are percentiles among the tracts compared, and correlations of rankings are population-weighted Pearson correlations of those percentiles.

Shares of the score's variance are cov(c_k x_k, S) / var(S) for the measures' terms c_k x_k and the score S, computed from the weighted covariance matrix, so each covariance between two terms is shared equally between them; the own-variance shares are var(sum of terms) / var(S) with no covariance with other terms allocated.

Counties are grouped by the ACS 2019-2023 median home value of the county (the Census API county pull) into quartiles with equal numbers of residents of the frame. High-poverty tracts have a poverty rate of at least 30 percent, and the table of the largest gaps lists the 15 tracts with the lowest published-minus-standardized percentile among those with at least 1,000 residents.

Figures explained here: Figure 3, Figure 4 and Figure 5

Weights and uncertainty

Weights are ACS 2019-2023 tract population.

Intervals are 95% percentile intervals from 2,000 draws of whole counties with replacement (seed 12345; Connecticut's planning regions stand in for counties, 3,126 clusters in the frame), one set of draws for all statistics; differences of two correlations on the same tracts use paired draws.

The ACS file carries no margins of error.

Factor model and plotted sample

The factor model is a maximum-likelihood fit to the population-weighted correlation matrix (statsmodels Factor; its EM start draws random loadings, seeded with 12345, and a second seed gives the same loadings to four decimals) with a varimax rotation; it and the eigenvalue shares describe the whole frame, and the eigenvalue shares also have bootstrap intervals.

The scatter plot shows a random sample of 3,000 tracts drawn with the same seed; every statistic uses all tracts.

Figure explained here: Figure 6

References

Hannan, Edward L.; Wu, Yifeng; Cozzens, Kimberly; Anderson, Brett (2023). The Neighborhood Atlas Area Deprivation Index For Measuring Socioeconomic Status: An Overemphasis On Home Value. Health Affairs 42(5): 702-709. doi:10.1377/hlthaff.2022.01406
Petterson, Stephen (2023). Deciphering the Neighborhood Atlas Area Deprivation Index: the consequences of not standardizing. Health Affairs Scholar 1(5): qxad063. doi:10.1093/haschl/qxad063
Singh, Gopal K. (2003). Area Deprivation and Widening Inequalities in US Mortality, 1969-1998. American Journal of Public Health 93(7): 1137-1143. doi:10.2105/AJPH.93.7.1137
Kind, Amy J. H.; Buckingham, William R. (2018). Making Neighborhood-Disadvantage Metrics Accessible: The Neighborhood Atlas. New England Journal of Medicine 378(26): 2456-2458. doi:10.1056/NEJMp1802313
University of Wisconsin School of Medicine and Public Health (2026). Neighborhood Atlas: Changes between versions of the ADI. Neighborhood Atlas web site, change log with entries dated 2019 to 2026. www.neighborhoodatlas.medicine.wisc.edu/changelog
Flanagan, Barry E.; Gregory, Edward W.; Hallisey, Elaine J.; Heitgerd, Janet L.; Lewis, Brian (2011). A Social Vulnerability Index for Disaster Management. Journal of Homeland Security and Emergency Management 8(1): Article 3. doi:10.2202/1547-7355.1792
Agency for Toxic Substances and Disease Registry (2024). CDC/ATSDR Social Vulnerability Index 2022 Documentation. CDC/ATSDR web document, October 2024. www.atsdr.cdc.gov/place-health/media/pdfs/2024/10/SVI2022Documentation.pdf
Centers for Medicare & Medicaid Services (2024). ACO REACH Model Performance Year 2024 Model Update: Health Equity Benchmark Adjustment. CMS Innovation Center web page. www.cms.gov/priorities/innovation/innovation-models/reach-py24-model-perf
Kolak, Marynia; Bhatt, Jay; Park, Yoon Hong; Padrón, Norma A.; Molefe, Ayrin (2020). Quantification of Neighborhood-Level Social Determinants of Health in the Continental United States. JAMA Network Open 3(1): e1919928. doi:10.1001/jamanetworkopen.2019.19928

How to cite

@misc{singh2026anatomyofdeprivation,
  author = {Singh, Vishal},
  title = {The anatomy of a deprivation index},
  year = {2026},
  url = {https://vishalsingh.org/zipcode/anatomy-of-deprivation},
  note = {In ZIP Code Destiny. Published 9 October 2026.}
}

Same question at another grain

These chapters take up the same axis with a different unit of analysis.

Deprivation index
- Chapter 10.3, A deprivation index at two zooms Neighborhoods

Reading paths

Methods first