United States · indicators

274 county measures

Each measure is listed with what it actually counts, who publishes it, and the publisher’s own code for the series. That way any figure on this site can be checked against its source instead of taken on trust.

Two columns here show how much any link involving a measure is worth. A link is measured by r: how closely two measures rise and fall together, from −1 (opposite) to +1 (in step). Density loading is how much the measure just reflects how urban a place is: its r with population density (in percentage terms). A measure with a loading of 0.85 is mostly reporting how urban a county is, whatever its name says. Moran’s I shows how much nearby places look alike on the measure. High values mean neighbouring counties resemble each other. Then the number of counties overstates how much separate information the measure carries.

Some of this data goes missing a whole state at a time. 111 of the 274measures have no data for at least one entire state. That is because each state, not each county, decides whether to take part in the survey behind them. Kentucky and Pennsylvania report none of the 2023 health measures at all, and the newest seven are missing Texas, Florida and six other states completely. A measure reported for “2,957 counties” that really covers 48 states is a different thing from one covering 2,957 counties spread across 51. So the state count is shown next to the county count. A link involving one of these quietly describes only the states that took part. That is a different problem from the scattered single counties the American Community Survey drops for a wide margin of error (too much uncertainty in the estimate). And it is different again from the 105 measures whose only gap is Puerto Rico, which several agencies leave out altogether.

The split is clean, and it is no accident. All 54 counts have a density loading between +0.28 and +0.89. All 212 rates and medians fall between +0.01 and +0.74. A count of anything is mostly a count of people.

274 of 274


Population (people) How many people live here, as the Bureau of Economic Analysis counts them to work out per-person figures.	BEA Regional Accounts `BEA CAINC1 line 2`	count	add up	3,091	51 no PR	0.89	0.33
Population aged 18 and over (people) The number of residents aged 18 and over. The PLACES health percentages are shares of this group. The adult population that the CDC PLACES rates are shares of.	CDC PLACES `PLACES totalpop18plus`	count	add up	3,144	51 no PR	0.87	0.32
Personal income (thousands of US$) All income that residents receive, from wages, investments and government benefits. It is counted where people live, not where they work.	BEA Regional Accounts `BEA CAINC1 line 1`	count	add up	3,091	51 no PR	0.87	0.32
Population in the same house one year ago (people) The number of people who lived in the same home a year earlier. It measures how settled a place is, the opposite of how often people move in and out.	Census ACS 5-year `ACS table B07001_017`	count	add up	3,141	51 no PR	0.87	0.33
Population (people) How many people live here, as County Health Rankings counts them. The population used to give bigger places more weight when County Health Rankings rates are combined. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v051`	count	add up	3,144	51 no PR	0.87	0.32
Population, decennial census (people) The census count for that decade (2010 or 2020), which the Religion Census divides by. It is kept separate from the American Community Survey estimate so the religion rates have their own exact base. The population the study divides by: that decade's full census count, not an estimate from the Census survey (ACS). It is kept so the membership rates have their own exact population when areas are combined. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census 2020 files; 2010 from the ARDA, RCMSCY10 and RCMSST10`	count	add up	3,144	51 no PR	0.87	0.32
Population for whom poverty status is determined (people) The people whose poverty status is worked out, used as the base for the poverty rate. It is calculated back from the poverty count and the poverty rate. Worked out as the number of people in poverty divided by the poverty rate.	Census SAIPE `Census SAIPE, derived from the count and rate`	count	add up	3,143	51 no PR	0.87	0.32
Presidential votes cast, 2024 (votes) Every presidential vote counted in the state in 2024, including third-party and write-in votes. It is a count, so it mostly follows population size. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2024 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MEDSL 2024 official state returns`	count	add up	3,112	50 no AK, PR	0.87	0.34
Presidential votes cast, 2016 (votes) Every presidential vote counted in 2016. It is a count, so it mostly follows population size. It is also used to give bigger places more weight when that year's vote shares are combined. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2016 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,113	50 no AK, PR	0.87	0.33
Tax returns filed (returns) The number of tax returns filed by people living in the area. Counted from tax returns filed. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N1`	count	add up	3,143	51 no PR	0.87	0.33
Presidential votes cast, 2020 (votes) Every presidential vote counted in 2020, the presidential election with the highest turnout in a century. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2020 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,113	50 no AK, PR	0.87	0.33
People covered by a tax return (people) The number of people covered by those tax returns, roughly the number of people in tax-filing households. Counted from tax returns filed. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N2`	count	add up	3,143	51 no PR	0.87	0.34
Presidential votes cast, 2012 (votes) The total number of presidential votes cast in 2012. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2012 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,112	50 no AK, PR	0.86	0.33
People who did not move (people) Tax filers who filed from the same address in both years. People on tax returns filed from the same address in both years. Every rate here is worked out per 100 of these people. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	count	add up	3,143	51 no PR	0.86	0.33
Presidential votes cast, 2008 (votes) The total number of presidential votes cast in 2008. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2008 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,112	50 no AK, PR	0.86	0.33
Tests taken (tests) The number of tests the county's figures are based on. The number of tests the county figures are based on. Time period: pooled over 2009-2019, the years the archive covers; these are pre-pandemic results and describe a decade of testing rather than a single year.	Stanford Education Data Archive `SEDA tot_asmts`	count	add up	3,136	50 no DC, PR	0.86	0.31
Presidential votes cast, 2004 (votes) The total number of presidential votes cast in 2004. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2004 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,112	50 no AK, PR	0.86	0.33
Aggregate travel time to work (minutes) The total minutes all commuters together spend getting to work in a day. A total of everyone's minutes, so it can be added up across areas. The average commute is this total divided by the number of workers, worked out again for each combined area.	Census ACS 5-year `ACS table B08013_001`	count	add up	3,208	52	0.85	0.34
Women aged 15 to 50 (people) The number of women aged 15 to 50, the group the survey asks about recent births.	Census ACS 5-year `ACS table B13002_001`	count	add up	3,215	52	0.85	0.31
Population aged 15 and over (people) The number of residents aged 15 and over. The Census asks this group about marriage, and every marriage percentage here is a share of it.	Census ACS 5-year `ACS table B12001_001`	count	add up	3,220	52	0.85	0.32
Population aged 25 and over (people) The number of residents aged 25 and over. Education levels are measured for this age group, since most people have finished school by then.	Census ACS 5-year `ACS table B15003_001`	count	add up	3,220	52	0.85	0.32
Total population (people) How many people live here, as counted by the American Community Survey. This comes from the survey, not from the Census Bureau's separate population estimates.	Census ACS 5-year `ACS table B01003_001`	count	add up	3,220	52	0.85	0.32
Population for whom poverty status is determined (people) The people whose poverty status can be worked out, used as the base for the poverty rate. It leaves out people in institutions, military barracks and college dorms, so it is a little smaller than the population.	Census ACS 5-year `ACS table B17001_001`	count	add up	3,220	52	0.85	0.32
Presidential votes cast, 2000 (votes) The total number of presidential votes cast in 2000. Every vote counted for president. When counties are combined, this number sets how much weight each gets in that year's vote share. Time period: from the 2000 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	count	add up	3,111	50 no AK, PR	0.85	0.32
Unemployed (people) People without a job who looked for one in the past four weeks and could start work.	BLS LAUS `BLS LAUS county annual averages`	count	add up	3,221	52	0.84	0.32
Occupied housing units (units) Homes and apartments that someone lives in.	Census ACS 5-year `ACS table B25003_001`	count	add up	3,219	52	0.84	0.31
Owner-occupied housing units (units) Lived-in homes where the people living there own the home.	Census ACS 5-year `ACS table B25003_002`	count	add up	3,219	52	0.84	0.33
Workers aged 16 and over who commute (people) The number of workers aged 16 and over who travel to a workplace.	Census ACS 5-year `ACS table B08303_001`	count	add up	3,218	52	0.84	0.32
Labour force (people) Everyone who is working or actively looking for work. People who have stopped looking are not included. So a falling unemployment rate can mean either more jobs or fewer people looking. People with a job plus people looking for one. Can be added up across areas.	BLS LAUS `BLS LAUS county annual averages`	count	add up	3,221	52	0.84	0.31
Population aged 65 and over (people) The number of residents aged 65 and over. This is a count of people, not a percentage. Added up from the twelve age-and-sex groups that make up people aged 65 and over.	Census ACS 5-year `ACS table B01001_020…B01001_049 (12 cells)`	count	add up	3,218	52	0.84	0.35
Employed (people) People who have a job, including part-time work and self-employment.	BLS LAUS `BLS LAUS county annual averages`	count	add up	3,221	52	0.84	0.31
Population below the poverty line (people) The number of people below the poverty line. It is an estimate from a statistical model, not a direct count.	Census SAIPE `Census SAIPE SAEPOVALL_PT`	count	add up	3,142	51 no PR	0.84	0.27
Population with a bachelor's degree or higher (people) The number of adults aged 25 and over who have at least a bachelor's degree. Bachelor's, master's, professional and doctoral degrees, added together.	Census ACS 5-year `ACS table B15003_022…B15003_025 (4 cells)`	count	add up	3,206	52	0.83	0.32
Housing units (units) All homes and apartments, whether or not anyone lives in them.	Census ACS 5-year `ACS table B25002_001`	count	add up	3,221	52	0.83	0.31
Business establishments (establishments) Business locations with paid employees in March, from the Census Bureau's list of businesses. Each location counts once, so a chain's stores count separately. Business sites with paid employees, counted in March of the year.	Census Business Dynamics Statistics `Census Business Dynamics Statistics API, ESTAB`	count	add up	3,144	51 no PR	0.83	0.30
Population (firearm measures) (people) How many people live here, from the 2017-2021 American Community Survey. The gun measures are adjusted using this population. Residents counted by the Census survey (ACS) in 2017-2021. The firearm measures are worked out over this population.	CDC WONDER mortality; RAND firearm ownership database `ACS 2017-2021 five-year, B01001_001`	count	add up	2,220	51 no PR	0.82	0.30
Population below the poverty line (people) The number of people living below the federal poverty line.	Census ACS 5-year `ACS table B17001_002`	count	add up	3,144	52	0.82	0.26
Children with below-median parental income (people) The number of children whose parents earned less than the middle income. The connectedness figures are measured over this group. The population the connectedness measures are worked out over. It is not a Facebook figure. The atlas takes it from the Opportunity Atlas, the same source as the mobility measures here. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights num_below_p50`	count	add up	3,089	51 no PR	0.82	0.24
Business establishments opened (establishments) Business locations that had employees in March and none a year earlier: new businesses, or new locations of old ones. A business moving in from another state counts here too. The data cannot tell the two apart. Business sites with employees in March that had none a year earlier.	Census Business Dynamics Statistics `Census Business Dynamics Statistics API, ESTABS_ENTRY`	count	add up	3,144	51 no PR	0.81	0.29
Women who gave birth in the past year (people) The number of women who said they gave birth in the past twelve months.	Census ACS 5-year `ACS table B13002_002`	count	add up	2,189	52	0.81	0.29
Children in the mobility cohort (people) The number of children the upward-mobility figures are based on. The number of children the mobility measures are worked out over. When areas are combined, each gets weight by this number. Time period: measured around 2015 for children born 1978-1983 who grew up in this county; most no longer live there.	Opportunity Atlas `Opportunity Insights pooled_pooled_count`	count	add up	3,136	51 no PR	0.81	0.24
Business establishments closed (establishments) Business locations that had employees a year earlier and none in March: closings, including a business moving a location out of the place. Business sites that had employees a year earlier but none in March.	Census Business Dynamics Statistics `Census Business Dynamics Statistics API, ESTABS_EXIT`	count	add up	3,144	51 no PR	0.80	0.29
Gross domestic product (thousands of US$) Gross domestic product: the value of everything produced in the area, in dollars of that year (not adjusted for inflation).	BEA Regional Accounts `BEA CAGDP2 line 1`	count	add up	3,091	51 no PR	0.80	0.23
Private jobs (jobs) Private-sector jobs, averaged over the year, as employers report them to the state's unemployment insurance program. Jobs are counted where the job is, not where the worker lives. Private-sector jobs reported to the unemployment insurance system, averaged over the year.	BLS Quarterly Census of Employment and Wages `BLS QCEW annual averages, private, all industries, annual_avg_emplvl`	count	add up	3,214	52	0.80	0.25
Population (WONDER) (people) How many people live here, as CDC WONDER counts them. Its yearly homicide rate is per this number of residents. Residents: the population that CDC WONDER's rates are worked out over.	CDC WONDER mortality `CDC WONDER population, by county and year`	count	add up	401	45 no CT, ID, MT, ND, PR, VT, WY	0.74	0.23
Share of some other race, not Hispanic (%) The percentage of residents who report some other race alone and are not Hispanic. It is small almost everywhere. It is included so the eight shares cover everyone, with nobody left over. Included so the eight group shares add up to the whole population. Without it they added up to 99.7% on average, and as little as 91% in the counties where this group is largest.	Census ACS 5-year `ACS table B03002_008, divided by total population`	rate	average weighted by total population, so bigger places count more	1,020	52	0.74	0.55
Local income tax collected (thousands of US$) Income tax collected from people by every local government in the place (county, cities, townships, school districts), in thousands of dollars. For counties, it covers census years only (1967, 1972 ... 2022), when every government is counted. Income tax on individuals collected by every local government in the area: counties, cities, townships and school districts. Counties have figures only for Census years; see the source note.	Census local government finances `Census local government finances, item T40, every local government`	count	add up	554	13 no AK, AR, AZ, CA, CO, CT, DE, FL, GA, HI, ID, IL, LA, MA, ME, MN, MS, MT, NC, ND, NE, NH, NJ, NM, NV, OK, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	0.72	0.29
Congregations per 10,000 people (per 10,000) How many separate congregations a place has per ten thousand residents. That is a different question from how many people belong. Rural counties have many small churches, and cities have fewer large ones. How many separate congregations a place supports. That is a different question from how many people belong to them. Rural counties have many small churches, and cities have fewer, larger ones. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census 2020 files; 2010 from the ARDA, RCMSCY10 and RCMSST10`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.71	0.43
Foreign-born population (people) The number of residents born outside the United States, whether or not they are citizens.	Census ACS 5-year `ACS table B05002_013`	count	add up	2,452	52	0.70	0.29
Vacant housing units (units) Homes and apartments that nobody lives in, including vacation homes and places for sale or rent. So a high number can mean a resort town or a town that is emptying out.	Census ACS 5-year `ACS table B25002_003`	count	add up	3,202	52	0.69	0.30
Share of taxes from severance (% of all taxes) The percentage from severance taxes, charged on oil, gas, coal and minerals as they are taken from the ground. This reflects geology before policy. North Dakota raises 41% of all its tax money this way, Alaska 36%, New Mexico 28% and Wyoming 26%, while 19 states raise nothing at all. It is the most uneven measure in this source. Any state at the top taxes its residents lightly, since it can tax oil and minerals instead. Charged on oil, gas, coal and minerals as they are taken out of the ground. So it says more about what lies under a state than about its tax choices. A named part of share of taxes from other taxes, so it is already counted inside it.	Census government finances `Census government finances LF0037`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.68	—
Cholesterol screening (%) The estimated percentage of adults who have been screened for cholesterol. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CHOLSCREEN, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.67	0.70
Share of taxes from property transfer (% of all taxes) The percentage from taxes charged when a property deed or a stock changes hands, not on owning either. It rises with how often property is sold as much as with the tax rate. So Florida and Washington are near the top, alongside Delaware and DC. Taxes charged when a property deed or a share of stock changes hands, not for owning either. A named part of share of taxes from other taxes, so it is already counted inside it.	Census government finances `Census government finances LF0036`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.67	—
Checkup in the past year (% of adults) Adults who saw a doctor for a routine checkup in the past year. Neither end is better: more checkups can mean better access to care or more illness. Adults who saw a doctor for a routine checkup within the year. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS CHECKUP1, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.66	—
State tax collected, all taxes (thousands of US$) Every tax the state government collected in its fiscal year, in thousands of dollars. Local governments' taxes are not included. All taxes, from the Census yearly survey of state government tax collections.	Census state tax collections `Census Annual Survey of State Government Tax Collections, historical database (T40, C105)`	count	add up	50 states	50 no DC, PR	0.64	—
Flu vaccination among Medicare enrollees (%) The percentage of people on Medicare who got a flu shot. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v155`	rate	average weighted by population, so bigger places count more	3,117	50 no CT, PR	0.63	0.46
Motor vehicle crash deaths (deaths per 100,000) Deaths in car and other motor vehicle crashes per 100,000 people. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v039`	rate	average weighted by population, so bigger places count more	2,491	51 no PR	0.62	0.44
Firearm suicide rate (per 100,000 a year) Suicides by gun per 100,000 residents a year, 2018-2023. Firearm suicides per 100,000 residents a year, 2018-2023. Not adjusted for age.	CDC WONDER mortality; RAND firearm ownership database `CDC WONDER underlying cause of death, suicide by intent, firearm by mechanism, 2018-2023`	rate	average weighted by population (firearm measures), so bigger places count more	2,220	51 no PR	0.62	0.45
Suicide rate (deaths per 100,000) Suicides per 100,000 people, adjusted so places with older or younger populations can be compared fairly. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v161`	rate	average weighted by population, so bigger places count more	2,049	50 no CT, PR	0.61	0.41
Share Asian, not Hispanic (%) The percentage of residents who are Asian alone and not Hispanic.	Census ACS 5-year `ACS table B03002_006, divided by total population`	rate	average weighted by total population, so bigger places count more	2,069	52	0.59	0.47
State and local tax collections (thousands of US$) Every tax dollar collected by a state and all the governments inside it: counties, cities, townships, school districts and special districts together. The governments report these themselves, so this is not a survey sample and not an estimate from a model. Every tax collected by the state and by every local government inside it.	Census government finances `Census government finances LF0008`	count	add up	51 states	51 no PR	0.58	—
People in households earning $200,000 or more who did not move (people) People in households reporting $200,000 or more who filed from the same address in both years. The three high-income moving rates are measured against this group. The three rates below are worked out per 100 of these people. From the IRS Gross Migration file. It splits moves by the size of adjusted gross income on the second year's tax return. It is published for states only, since the IRS does not split counties by income. This file leaves out returns with an income below zero. So its totals differ slightly from the state-to-state files the other migration measures come from.	IRS county migration flows `IRS SOI gross migration, AGI class 7`	count	add up	51 states	51 no PR	0.57	—
Vacancy rate (%) The percentage of all homes and apartments that nobody lives in. The middle county is 15%, which surprises people. It counts vacation and seasonal homes, so the top of the list is resort and outdoor-recreation areas, such as Daggett County, Utah at 77%, not abandoned towns. A high number can tell two very different stories, and this measure cannot tell them apart.	Census ACS 5-year `ACS table B25002_003, divided by housing units`	rate	average weighted by housing units, so bigger places count more	3,203	52	0.56	0.38
Friending clustering (share) How often a person's friends are also friends with each other. It shows how tight-knit the local network is, which is separate from how far it reaches across income groups. On average, how often two of a person's friends are also friends with each other. It shows how tight-knit local friend groups are. That is separate from how far they reach across class. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights clustering_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,091	51 no PR	0.56	0.53
Share of taxes from death and gift (% of all taxes) The percentage from taxes on what people leave when they die, and on gifts. Estate taxes are charged on what is left, and inheritance taxes on what each heir receives. Only sixteen states and DC collect much, and even for them it is rarely above 2% of tax money. So this is close to a yes-or-no measure of state policy, not a scale. A named part of share of taxes from other taxes, so it is already counted inside it.	Census government finances `Census government finances LF0035`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.55	—
Hearing disability (%) The estimated percentage of adults who report serious difficulty hearing. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES HEARING, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.53	0.64
Orthodox Christian adherents (% of population) Members of Orthodox Christian churches, such as Greek, Russian and Antiochian Orthodox, as a share of the population. Greek, Russian, Antiochian and other Orthodox bodies. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census ORTHRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.52	0.44
State individual income tax collected (thousands of US$) Income tax collected from people by the state government in its fiscal year, in thousands of dollars. Local income taxes are not included. Individual income tax, from the Census yearly survey of state government tax collections (line T40).	Census state tax collections `Census Annual Survey of State Government Tax Collections, historical database (T40, C105)`	count	add up	50 states	50 no DC, PR	0.52	—
Mutual-friend share (share) The share of local friendships where the two friends also share a third friend nearby. The share of friendships within the county where the two friends also share a third friend in the same county. The atlas calls this the 'support ratio'. That sounds like a measure of support, but it measures how tight-knit friend groups are. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights support_ratio_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,091	51 no PR	0.52	0.42
Colorectal cancer screening (%) The estimated percentage of adults who aged 45 to 75 are up to date with colon and rectal cancer screening. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES COLON_SCREEN, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	3,143	51 no PR	0.51	0.62
Returns claiming charitable contributions (%) The percentage of tax returns claiming a deduction for gifts to charity. Only people who itemize their deductions can claim it, and the 2017 tax law cut their number roughly in half. So this tracks who finds itemizing worth it as much as who gives. Shows up only on returns that list deductions one by one (itemise). The 2017 tax law roughly halved the number of people who itemise. So this now measures who finds itemising worth it as much as who gives. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N19700 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,143	51 no PR	0.50	0.65
Routine checkup in the past year (%) The estimated percentage of adults who saw a doctor for a routine checkup in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CHECKUP, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.50	0.88
Republican share of the two-party vote, 2024 (% of the two-party vote) The Republican share of the votes for the two main parties in the 2024 presidential election, by state. Third-party votes are left out, so this shows which way the two-party vote leaned and nothing about turnout. It describes how a state voted and does not judge it. No measure here treats either party as the better end. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2024 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MEDSL 2024 official state returns`	rate	average weighted by presidential votes cast, 2024, so bigger places count more	3,112	50 no AK, PR	0.49	0.60
Republican share of the two-party vote, 2020 (% of the two-party vote) The Republican share of the votes for the two main parties in the 2020 presidential election. It is shown with 2016 and 2024 so you can see which way a county is moving, not only where it stands. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2020 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2020, so bigger places count more	3,113	50 no AK, PR	0.49	0.60
Firearm fatalities (deaths per 100,000) Deaths involving a gun per 100,000 people. Most are suicides and homicides, counted together. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v148`	rate	average weighted by population, so bigger places count more	2,013	51 no PR	0.48	0.49
Median gross rent (US$/month) The rent paid by the middle renting household, plus any utilities the renter pays. That is why it is called 'gross' rent, not just the rent in the lease.	Census ACS 5-year `ACS table B25064_001`	rate	cannot combine	3,185	52	0.48	0.70
Local income tax as a share of income (% of personal income) Local income tax as a percentage of what residents earned. It is zero in most of the country. About 4,800 local governments in a dozen states charge one, and New York City's is the largest. When a city spans several counties, its tax is split between them by where its residents' income is. Local income tax collected, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis).	Census local government finances `Census local government finances, item T40, every local government`	rate	cannot combine	554	13 no AK, AR, AZ, CA, CO, CT, DE, FL, GA, HI, ID, IL, LA, MA, ME, MN, MS, MT, NC, ND, NE, NH, NJ, NM, NV, OK, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	0.47	0.59
Millionaire tax returns (returns) Tax returns filed from the state with adjusted gross income of $1 million or more. Residents, not movers: the count rises when millionaires move in, and also when residents have a good year in the stock market. There are no figures for 2002-2009, when the IRS state tables stopped at $200,000. Individual tax returns reporting adjusted gross income of $1 million or more.	IRS Statistics of Income, state data `IRS SOI Historic Table 2, AGI $1,000,000 or more`	count	add up	52 states	52	0.47	—
Republican share of the two-party vote, 2016 (% of the two-party vote) The Republican candidate's share of the votes for the two main parties in the 2016 presidential election. Third-party votes are left out, so they do not push it up or down. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2016 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2016, so bigger places count more	3,113	50 no AK, PR	0.45	0.60
Adult income rank of children raised poor here (rank 0-1) Where children who grew up here with poor parents ended up as adults, ranked against everyone in the country by income, on a scale from 0 to 1. A value of 0.5 would mean that on average they reached the middle of the national income ladder. The average household income rank at age 35 (0 = poorest, 1 = richest) of children whose parents were at the 25th percentile, that is, poorer than about three in four families. The Opportunity Atlas's main measure. Time period: measured around 2015 for children born 1978-1983 who grew up in this county; most no longer live there.	Opportunity Atlas `Opportunity Insights kfr_pooled_pooled_p25`	rate	average weighted by children in the mobility cohort, so bigger places count more	3,208	52	0.45	0.75
Safety score (standard deviations) Deaths from violence and injury (guns, crashes and overdoses), flipped so higher is safer. Homicide, suicide, drunk-driving and total injury deaths are left out, since they would count the same deaths twice. Gun deaths already include most homicides and suicides, and injury deaths include all four. The average of 4 measures from 2 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 4 measures (BRFSS, CHR)`	rate	cannot combine	2,086	51 no PR	0.44	0.48
Child mortality (deaths per 100,000 under 18) Deaths of children under 18 per 100,000 children. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v128`	rate	average weighted by population, so bigger places count more	1,735	51 no PR	0.44	0.38
Share of taxes from motor vehicle licences (% of all taxes) The percentage from vehicle licence taxes: registration and license plates. Most states charge a flat fee per vehicle. So it makes up a larger share of tax money where incomes are low and people have to drive. A named part of share of taxes from licences, so it is already counted inside it.	Census government finances `Census government finances LF0029`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.44	—
Current smoking (% of adults) Adults who have smoked at least a hundred cigarettes in their life and now smoke every day or some days. Adults who smoke every day or some days. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFSMOK3, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.43	—
Share Native Hawaiian or Pacific Islander, not Hispanic (%) The percentage of residents who are Native Hawaiian or other Pacific Islander alone and not Hispanic. It is under 0.1% in most counties. So outside Hawaii and a few Pacific-coast counties, it barely varies, and there is little to compare.	Census ACS 5-year `ACS table B03002_007, divided by total population`	rate	average weighted by total population, so bigger places count more	1,891	52	0.43	0.36
Effective individual income tax rate (% of personal income) Income tax collected from people, as a percentage of their total income. It is the share paid across everyone, not the rate on any one person's top dollar. It runs from a true zero in the nine states without an income tax to about 4.7% in Oregon and New York. Individual income tax collected, divided by personal income. Zero in the states that have none. Census figures for state and local tax collected, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). It shows the total tax paid as a share of income. It is not the rate written in law, and not the rate on the next dollar earned. Taxes are counted over the government's budget year and income over the calendar year, and state and local taxes are combined.	`Census tax collections times the share of taxes from individual income, divided by BEA personal income`	rate	average weighted by personal income, so bigger places count more	51 states	51 no PR	0.42	—
Black Protestant adherents (% of population) Members of the historically Black Protestant churches, as a share of the population. Their central records are thinner than most, so this is the least complete of the six groups. Which of these churches reported also differs between the 2010 and 2020 studies, so part of any change between them is who reported, not real membership. The historically Black denominations. Their central records are thinner than most, so this is the least complete of the six. Which of them reported differs between the 2010 and 2020 studies. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census BPRTRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.41	0.54
Always wears a seatbelt (% of adults) Adults who say they always wear a seatbelt in a car. Adults who say they always wear a seatbelt. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFSEAT2, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.41	—
Mammography in the past two years (%) The estimated percentage of adults who women aged 50 to 74 had a mammogram in the past two years. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES MAMMOUSE, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	3,143	51 no PR	0.41	0.59
High cholesterol (% of adults) Adults who have ever been told that their blood cholesterol is high. Adults ever told their cholesterol is high. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFCHOL3, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.41	—
Share of suicides by firearm (%) The percentage of suicides that were by gun, 2018-2023. Where more households own guns, more suicides are by gun. So researchers use this as the standard stand-in for gun ownership, which nobody counts directly. Firearm suicides divided by all suicides, with 2018-2023 taken together.	CDC WONDER mortality; RAND firearm ownership database `CDC WONDER underlying cause of death, suicide by intent, firearm by mechanism, 2018-2023`	rate	average weighted by population (firearm measures), so bigger places count more	2,220	51 no PR	0.40	0.51
Median home value (US$) What the owner of the middle owner-lived-in home says it would sell for. It is the owner's own guess, not an actual sale price.	Census ACS 5-year `ACS table B25077_001`	rate	cannot combine	3,207	52	0.40	0.71
Share never married (%) The percentage of people aged 15 and over who have never married. The Census counts from age 15, not 18, so every county's figure includes some teenagers.	Census ACS 5-year `ACS table B12001_003…B12001_012 (2 cells), divided by population aged 15 and over`	rate	average weighted by population aged 15 and over, so bigger places count more	3,213	52	0.40	0.33
Share aged 65 and over (%) The percentage of all residents, children included, who are 65 or older. It is a share of everyone, not just of adults. The typical county is 19.7%, and four in five are between 14% and 26%. The extremes come from one big group living there, not from places that aged differently: Sumter County, Florida reaches 57.8% since The Villages retirement community is there. Chattahoochee County, Georgia is 3.3% since it is mostly Fort Benning, an army base, and Madison County, Idaho is 5.3% since it is a university town.	Census ACS 5-year `ACS table B01001_020…B01001_049 (12 cells), divided by total population`	rate	average weighted by total population, so bigger places count more	3,218	52	0.39	0.35
Friending bias (share) How much less often low-income people make high-income friends than you would expect from how often they meet them. A high value means the chance to mix is there but is not taken. One minus (connectedness divided by exposure). It shows how much less likely lower-status people are to befriend higher-status people than you would expect from how many of them they meet. It is built from the same numbers as economic connectedness, so the two are bound to match to some degree. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights bias_grp_mem_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,014	51 no PR	0.39	0.22
Share with a bachelor's degree or higher (%) The percentage of adults aged 25 and over who have at least a bachelor's degree. It counts degrees people already have, not people who are in college now. A share of people aged 25 and over. When areas are combined, each gets weight by its number of people aged 25 and over. Using total population instead would tilt the result toward places with many children. This is one of the four factors the site adjusts for by default when it compares US places.	Census ACS 5-year `ACS table B15003_022…B15003_025 (4 cells), divided by population aged 25 and over`	rate	average weighted by population aged 25 and over, so bigger places count more	3,207	52	0.39	0.43
Public-good pages per 1,000 users (per 1,000 Facebook users) Facebook pages for community groups and public-good efforts, per thousand Facebook users. It is per user, not per resident. Facebook Pages that a computer model judges to serve the public good, per thousand Facebook users in the county. It is a rate per user, not per resident, so it cannot be compared with a per-person figure. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights civic_organizations_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,091	51 no PR	0.39	0.28
Fine particulate matter (micrograms per cubic metre) The average amount of fine particle pollution in the air, in micrograms per cubic metre. These are bits of soot small enough to get into the blood. It is an estimate from a statistical model, since most counties have no air monitor. Average daily level of fine dust and soot particles in the air (PM2.5). It comes from an air-quality computer model, not from a monitor in every county. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v125`	rate	average weighted by population, so bigger places count more	3,059	50 no CT, PR	0.38	0.73
Homicide rate (deaths per 100,000) Deaths recorded as homicide per 100,000 people. It comes from death certificates, not police reports. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v015`	rate	average weighted by population, so bigger places count more	1,144	51 no PR	0.37	0.44
Share of taxes from corporate income (% of all taxes) The percentage from taxes on company profits. A zero does not mean businesses go untaxed, and this measure alone will mislead about any state. Businesses are taxed three ways, filed in three places. Tax on profit is this line. Tax on total sales, which Nevada, Texas, Washington and Wyoming use instead, is filed under general sales. Tax for the right to be incorporated is the corporation licence tax. Delaware raises 30% of its tax money that third way and 6% this way. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0024`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.37	—
Firearm share of suicides, standardised (index, 100 = population mix) The gun share of suicides compared with what you would expect if each sex, age and race group here used guns at its national rate. 100 means it matches what the local population mix would predict. Above 100 means more gun suicide than that. It is the best county measure of gun access available, and it closely tracks RAND's state estimates of gun ownership. The share of suicides by firearm, compared with the share expected from the place's mix of sex, age and race if national rates applied. Times 100, so 100 means exactly what that mix would predict.	CDC WONDER mortality; RAND firearm ownership database `CDC WONDER underlying cause of death, suicide by intent, firearm by mechanism, 2018-2023`	rate	average weighted by population (firearm measures), so bigger places count more	2,220	51 no PR	0.37	0.53
Mean income of departing households (US$) The average income on the tax returns of people who moved out. Adjusted gross income (the IRS's main measure of income) per tax return of departing households. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,111	51 no PR	0.36	0.48
Share of taxes from individual income (% of all taxes) The percentage from income tax on people. It is zero in Alaska, Florida, Nevada, South Dakota, Texas and Wyoming, a real zero and not missing data. New Hampshire's 1.8% is its tax on interest and dividends, and Washington's 1.4% is its tax on investment profits (capital gains). Capital gains have no line of their own, since almost every state taxes them as regular income, so they are already in this figure. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0023`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.36	—
Difficulty concentrating or deciding (% of adults) Adults with serious trouble concentrating, remembering or making decisions. Adults with serious difficulty concentrating, remembering or making decisions. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS DECIDE, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.36	—
Median household income (model-based) (US$) The income of the middle household. It is an estimate from a statistical model that combines survey answers with tax and government-program records. The middle value: half of households earn more and half less. SAIPE publishes it separately for each level.	Census SAIPE `Census SAIPE SAEMHI_PT`	rate	cannot combine	3,143	51 no PR	0.35	0.58
Share widowed (%) The percentage of people aged 15 and over whose spouse has died. It mostly follows how old the population is, so it is close to a second measure of a county's age.	Census ACS 5-year `ACS table B12001_009…B12001_018 (2 cells), divided by population aged 15 and over`	rate	average weighted by population aged 15 and over, so bigger places count more	3,173	52	0.35	0.33
Republican share of the two-party vote, 2012 (% of the two-party vote) The Republican candidate's share of the votes for the two main parties in the 2012 presidential election, Romney against Obama. Third-party votes are left out, just as in the other years. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2012 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2012, so bigger places count more	3,112	50 no AK, PR	0.35	0.62
Depression (% of adults) Adults who have ever been told they have depression (a depressive disorder). People report this themselves, so it counts being told by a doctor as well as being ill. Adults ever told they have a depressive disorder. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS ADDEPEV3, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.35	—
Births per 1,000 women aged 15 to 50 (per 1,000) Births in the past year for every 1,000 women aged 15 to 50. It is a survey estimate, not a count of birth certificates. In small counties it rests on very few answers and can be far off, so the highest and lowest values are mostly random noise from the survey.	Census ACS 5-year `ACS table B13002_002, divided by women aged 15 to 50`	rate	average weighted by women aged 15 to 50, so bigger places count more	2,195	52	0.34	0.17
Infant mortality (deaths per 1,000 live births) Deaths of babies before their first birthday per 1,000 babies born alive. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v129`	rate	average weighted by population, so bigger places count more	1,172	50 no CT, PR	0.34	0.39
Republican share of the two-party vote, 2000 (% of the two-party vote) The Republican candidate's share of the votes for the two main parties in the 2000 presidential election. Third-party votes are left out, just as in the other years. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2000 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2000, so bigger places count more	3,111	50 no AK, PR	0.34	0.59
Median property tax bill (US$/year) The yearly property tax bill, in dollars, of the middle home-owning household that pays one. Households with no bill are left out, so this is what payers pay, not the average across all owners. The middle value: the tax paid by the middle household among owners who get a tax bill. Households paying none are left out. The Census stops counting at $10,001. Sixteen counties hit that cap in 2023. For those, the true figure is at least this much, not exactly this much.	Census ACS 5-year `ACS table B25103_001`	rate	cannot combine	3,182	52	0.33	0.79
Homicide rate, annual (deaths per 100,000) Deaths recorded as homicide in the year, per 100,000 residents, from death certificates (CDC WONDER). It is shown only for places and years with 10 or more homicides, so small counties appear rarely or never. Deaths recorded as homicide in the year, per 100,000 residents. Shown only for county-years with 10 or more such deaths.	CDC WONDER mortality `CDC WONDER underlying cause of death, homicide by intent, by county and year`	rate	average weighted by population (wonder), so bigger places count more	401	45 no CT, ID, MT, ND, PR, VT, WY	0.33	0.40
Share American Indian or Alaska Native, not Hispanic (%) The percentage of residents who are American Indian or Alaska Native alone and not Hispanic. They live mostly on and near reservations: Oglala Lakota County, South Dakota is 91%, while most counties are under 1%.	Census ACS 5-year `ACS table B03002_005, divided by total population`	rate	average weighted by total population, so bigger places count more	1,399	52	0.33	0.41
Mean adjusted gross income per return (US$) Average adjusted gross income (income after a few deductions, as reported to the IRS) per tax return. Business and farm losses are subtracted. It counts only people who file, and people who earn too little to have to file are missing, which makes the poorest places look better off. Total adjusted gross income (the IRS's main measure of a filer's income) divided by the number of returns filed. The IRS reports it in thousands of dollars, and it is converted here. People who do not file a return do not show up at all. So this overstates income most where the fewest people file. The population is filers, not residents.	IRS Statistics of Income `IRS SOI A00100 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,124	51 no PR	0.33	0.38
Arthritis (% of adults) Adults who have ever been told they have arthritis, rheumatoid arthritis, gout, lupus or fibromyalgia. Adults ever told they have some form of arthritis. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _DRDXAR2, age-standardised`	rate	cannot combine	46 states	46 no KY, NM, NV, PA, PR, SD	0.33	—
Republican share of the two-party vote, 2008 (% of the two-party vote) The Republican candidate's share of the votes for the two main parties in the 2008 presidential election, McCain against Obama. Third-party votes are left out, just as in the other years. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2008 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2008, so bigger places count more	3,112	50 no AK, PR	0.32	0.60
Health score (standard deviations) One number for a county's health. It averages illness, health habits, preventive care and death rates, each put on the same scale and turned so higher is better. Twenty-six of its thirty measures are CDC PLACES estimates from a statistical model. So measures are averaged within each publisher first, and PLACES counts once, not twenty-six times. The average of 41 measures from 3 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 41 measures (BRFSS, CDC, CHR)`	rate	cannot combine	2,956	49 no KY, PA, PR	0.32	0.58
Republican share of the two-party vote, 2004 (% of the two-party vote) The Republican candidate's share of the votes for the two main parties in the 2004 presidential election, Bush against Kerry. Third-party votes are left out, just as in the other years. Republican votes divided by Republican plus Democratic votes. Votes for third parties are left out. So a county where a third party did well does not look less Republican for that reason alone. Time period: from the 2004 presidential election; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2004, so bigger places count more	3,112	50 no AK, PR	0.31	0.58
Obesity (% of adults) Adults with a body mass index (a measure of weight for height) of 30 or more, worked out from the height and weight they reported themselves. Adults with a body mass index of 30 or more, from reported height and weight. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _BMI5CAT, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.30	—
Share of taxes from sports betting and pari-mutuels (% of all taxes) The percentage from taxes on sports betting and horse-race betting. The Census only split this out starting in 2023, so there is no 2022 figure. That year is marked unknown, not zero, since states taxed betting long before it had its own line. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0019`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.30	—
Net domestic migration rate (per 1,000 residents) Net moves from other states per 1,000 residents, every year from 2001. July 2009 to July 2010 is missing, since it falls between two editions of the estimates. It is an estimate built mostly from addresses on tax returns, so it is a longer, smoother relative of the IRS moving data, not a separate count. Net domestic migration divided by the population a year earlier, per 1,000 people. It comes from the Census population estimates, which are built mostly from tax returns and Medicare enrolment.	Census population estimates `Census population estimates, DOMESTICMIG over POPESTIMATE of the prior year`	rate	cannot combine	51 states	51 no PR	0.30	—
State income tax as a share of income (% of personal income) State income tax collected as a percentage of what residents earned, every year from 1970. This is the share actually paid, not the rate written in law. A tax increase shows as a step up that stays, and a one-time windfall from investment profits as a spike that falls back. Local income taxes, such as New York City's, Maryland's counties' and Ohio's cities', are not included. Individual income tax collected by the state government, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). This is the share of income actually paid, from 1970 onward. State government only, local income taxes excluded; fiscal-year collections over calendar-year personal income.	Census state tax collections `Census Annual Survey of State Government Tax Collections, historical database (T40, C105)`	rate	cannot combine	50 states	50 no DC, PR	0.30	—
Mean travel time to work (minutes) The average one-way trip to work, in minutes, for people who commute. People who work from home are left out, not counted as zero. Total minutes spent getting to work, divided by the number of commuters. It is minutes, not a percentage.	Census ACS 5-year `ACS table B08013_001, divided by workers aged 16 and over who commute`	rate	average weighted by workers aged 16 and over who commute, so bigger places count more	3,213	52	0.29	0.49
Mean income of arriving households (US$) The average income on the tax returns of people who moved in. Adjusted gross income (the IRS's main measure of income) per tax return of arriving households. It is per return, not per person, since income is reported by the household. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,096	51 no PR	0.29	0.45
Homeownership rate (%) The percentage of lived-in homes that are owned, not rented. Empty homes are left out of the count entirely. The share of lived-in homes whose owners live in them. Empty homes are left out. When areas are combined, each gets weight by its lived-in homes, not by all homes.	Census ACS 5-year `ACS table B25003_002, divided by occupied housing units`	rate	average weighted by occupied housing units, so bigger places count more	3,219	52	0.29	0.19
Injury deaths (deaths per 100,000) Deaths from injury per 100,000 people: overdoses, crashes, falls, suicides and homicides together. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v135`	rate	average weighted by population, so bigger places count more	2,978	50 no CT, PR	0.29	0.40
People per tax return (people) People covered per tax return: household size as the tax system sees it. Exemptions (people claimed on a return) divided by returns. It is a measure of household size from tax filings, not from a survey. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N2 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,143	51 no PR	0.28	0.56
Other religion adherents (% of population) Members of every group not counted in the other five, including Jewish, Muslim, Hindu, Buddhist, Latter-day Saints and more. It is a leftover group, so its members have little in common except not fitting the other five. Everything not Christian and not sorted elsewhere: Jewish, Muslim, Hindu, Buddhist, Latter-day Saints and more. So it is a leftover category, not a group with anything in common. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census OTHRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.28	0.78
Net domestic migration (people) People moving in from other states minus people moving out to them, from 1 July to 1 July, from the Census population estimates. People moving in from other states minus people moving out, from 1 July one year to 1 July the next.	Census population estimates `Census population estimates, DOMESTICMIG`	count	add up	51 states	51 no PR	0.28	—
Median age (years) The middle age of the county: half the residents are older and half are younger. The typical county is 41.5, and four in five are between 36 and 49. The extremes come from one big group of residents, such as retirees or students: 68.9 in Loving County, Texas and 68.4 in Sumter County, Florida, against 21.4 in Madison County, Idaho, where a university sets the middle age. Since it is a middle value, county figures cannot be averaged into a state figure. The middle value: half of residents are older and half younger. The Census publishes it for every kind of area, so it is never worked out by averaging smaller areas.	Census ACS 5-year `ACS table B01002_001`	rate	cannot combine	3,221	52	0.28	0.26
Republican swing, 2000 to 2024 (percentage points) The change in the Republican share of the two-party presidential vote from 2000 to 2024, in percentage points. Positive means the area moved toward Republicans. That is a different question from which party it votes for. How far the county moved over twenty-four years. That is a different question from where it stands, and it goes with different things. Positive means it moved towards the Republican candidate. Time period: from the 2000 and 2024 presidential elections; how a county votes changes slowly, so an election result is a fair stand-in for how it leans now, but it is not a direct measure of that.	MIT Election Data and Science Lab `MIT county presidential returns`	rate	average weighted by presidential votes cast, 2024, so bigger places count more	3,111	50 no AK, PR	0.27	0.60
High cholesterol (%) The estimated percentage of adults who have been told they have high cholesterol. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES HIGHCHOL, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.27	0.83
Share of taxes from motor fuel (% of all taxes) The percentage from taxes on gasoline and diesel. Almost everywhere it is charged per gallon, not as a percentage of the price. So, after inflation, the money it raises falls whenever fuel prices rise or cars use less fuel, without anyone voting to cut it. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0015`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.27	—
Economy score (standard deviations) Four income measures from four agencies: income per person, middle household income (an estimate from a statistical model), personal income per person, and average adjusted gross income on tax returns. Read it as four ways of measuring income, not as a broad economic summary. It moves very closely with middle household income (a link of 0.91 on a scale where 1 is perfect), and the site already adjusts for that. So a link with this score and a link with income are close to the same question. The average of 4 measures from 4 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 4 measures (ACS, BEA, IRS, SAIPE)`	rate	cannot combine	3,143	51 no PR	0.27	0.55
Share foreign-born (%) The percentage of residents born outside the United States, whether or not they are citizens. The middle county is 2.7%, and the numbers are very uneven: Miami-Dade is 54% and Queens 48%, while half of all counties are under 3%. The count alone mostly ranks counties by size, which is why this percentage is shown next to it.	Census ACS 5-year `ACS table B05002_013, divided by total population`	rate	average weighted by total population, so bigger places count more	2,453	52	0.27	0.61
Current cigarette smoking (%) The estimated percentage of adults who smoke cigarettes every day or some days. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CSMOKING, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.27	0.60
Millionaire returns as a share of all returns (% of returns) Millionaire tax returns as a percentage of all returns filed from the state. Tax returns with adjusted gross income of $1 million or more, as a share of all returns.	IRS Statistics of Income, state data `IRS SOI Historic Table 2, AGI $1,000,000 or more`	rate	cannot combine	52 states	52	0.26	—
Share now married (%) The percentage of people aged 15 and over who are married now, including those whose spouse lives elsewhere. The middle county is 52.4%. Never married, married, widowed and divorced cover everyone 15 and over and add up to exactly 100%.	Census ACS 5-year `ACS table B12001_004…B12001_013 (2 cells), divided by population aged 15 and over`	rate	average weighted by population aged 15 and over, so bigger places count more	3,221	52	0.26	0.35
Went without care over cost (% of adults) Adults who needed to see a doctor in the past year but did not, because the cost stopped them. People report this themselves, so it records what they said held them back. Adults who needed to see a doctor in the past year and did not, citing cost. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS MEDCOST1, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.26	—
Has health insurance (% of adults) Adults with any kind of health insurance. Neither end is marked as better here, just as for the county version of this measure. Adults with any form of health coverage. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _HLTHPL1, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.26	—
Share of taxes from general sales and gross receipts (% of all taxes) The percentage of total tax money from general sales and 'gross receipts' taxes. Note the second part: the Census puts taxes on a business's total sales here. So Oregon seems to raise 4.2% from 'sales' with no sales tax at all. That figure is its Corporate Activity Tax. Washington's B&O tax and Texas's margin tax are here too, which is why those states show no corporate income tax. Includes taxes on a business's total sales, so a state with no sales tax can still show a figure here. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0011`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.26	—
Coronary heart disease (%) The estimated percentage of adults who have been told they have coronary heart disease (narrowed heart arteries). It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CHD, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.25	0.56
Median household income (US$) The income of the middle household: half of households earn more, and half earn less. It counts all money income before taxes. Since it is the middle value, not the average, a few very high incomes do not change it. The middle value: half of households earn more and half earn less. A middle value cannot be built up from smaller areas. Averaging county middle values does not give the state's middle value. Incomes are lopsided, with a long tail of high earners, so the average would land in the wrong place.	Census ACS 5-year `ACS table B19013_001`	rate	cannot combine	3,208	52	0.25	0.63
Two weeks of poor physical health (% of adults) Adults who say their physical health was not good on fourteen or more of the last thirty days. Adults reporting fourteen or more days in the last thirty when their physical health was not good. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _PHYS14D, age-standardised`	rate	cannot combine	47 states	47 no KY, NV, PA, PR, SD	0.25	—
State and local income tax as a share of income (% of personal income) State and local income tax together, as a percentage of residents' income. This is the fair comparison for New York, Maryland, Ohio and Pennsylvania, where the state rate alone leaves out a large local layer of tax. Income tax collected by the state government (from Census state tax records) plus every local government's, divided by BEA personal income.	Census local government finances `Census state tax collections T40 plus local government finances T40, divided by BEA personal income`	rate	cannot combine	12 states	12 no AK, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, ID, IL, LA, MA, ME, MN, MS, MT, NC, ND, NE, NH, NJ, NM, NV, OK, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	0.25	—
Effective general sales tax rate (% of personal income) General sales tax collected, as a percentage of residents' income. It is the flip side of the income tax rate: states without an income tax raise more here, except New Hampshire, which has neither. General sales tax and taxes on business sales, divided by personal income. States without an income tax usually rely more on this one. Census figures for state and local tax collected, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). It shows the total tax paid as a share of income. It is not the rate written in law, and not the rate on the next dollar earned. Taxes are counted over the government's budget year and income over the calendar year, and state and local taxes are combined.	`Census tax collections times the share of taxes from general sales and gross receipts, divided by BEA personal income`	rate	average weighted by personal income, so bigger places count more	51 states	51 no PR	0.25	—
Economic connectedness in high school (ratio) How many of a low-income teenager's high school friends are high-income, compared with an even mix. It is counted in the county of the school, not where the person lives as an adult, so it describes where people grew up. The same measure, worked out over high-school friendships and assigned to the county of the school. So it describes where people grew up, not where they live now. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights child_ec_county`	rate	average weighted by children with below-median parental income, so bigger places count more	2,731	51 no PR	0.25	0.58
Diabetes (% of adults) Adults who have ever been told they have diabetes, not counting diabetes that only appeared during pregnancy. Adults ever told they have diabetes, excluding gestational. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS DIABETE4, age-standardised`	rate	cannot combine	45 states	45 no DC, KY, NH, NV, PA, PR, SD	0.25	—
Share of taxes from licences (% of all taxes) The percentage from all licence taxes together: vehicle registration, corporation and business licences, driver's licences, hunting and fishing, and the rest. The Census publishes no total for this group, so it is added up here from its nine parts. Without it, the other six main shares would add to 96%, with nothing to explain the gap. Added up from nine kinds of licence, since the Census publishes no total for licences. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0025, LF0026, LF0027, LF0028, LF0029, LF0030, LF0031, LF0032, LF0033`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.24	—
Top state income tax rate (%) The highest rate in the state's own income tax on wages and similar income, for a single filer, for the tax year. It includes statewide extra taxes on top earners and leaves out local income taxes. It is compiled from each state's laws and tax forms, and is 0 where the state has no broad income tax. The highest rate in the state's own income tax: the rate on each extra dollar in the top bracket. For a single person filing alone. Extra statewide taxes on top earners are included, local taxes are not.	State tax law, compiled from statutes and tax forms `Compiled from state statutes, session laws and tax forms; sources per change in top_rate_changes.csv`	rate	cannot combine	50 states	50 no DC, PR	0.24	—
Disability-free score (standard deviations) How common disability is, across seven kinds, flipped so higher means less disability. It is kept out of the health score on purpose. Disability goes with age and with how people report it as much as with anything a county does. Adding it would make an older county look sicker just for being older. How common disability is goes with age, and with how people report it, as much as with anything a county does. So it is kept out of the health score. Otherwise an older county would look sicker just for being older. The average of 13 measures from 2 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 13 measures (BRFSS, CDC)`	rate	cannot combine	2,956	49 no KY, PA, PR	0.24	0.60
Net migration per 100 staying, $200,000 or more (per 100) Households reporting $200,000 or more moving in minus those moving out, for every hundred such households that stayed. Positive means the state gained high earners overall. Arrivals minus departures, for the top income group only. From the IRS Gross Migration file. It splits moves by the size of adjusted gross income on the second year's tax return. It is published for states only, since the IRS does not split counties by income. This file leaves out returns with an income below zero. So its totals differ slightly from the state-to-state files the other migration measures come from.	IRS county migration flows `IRS SOI gross migration, AGI class 7`	rate	average weighted by people in households earning $200,000 or more who did not move, so bigger places count more	51 states	51 no PR	0.23	—
Millionaires' share of all income (% of adjusted gross income) The share of all income reported on the state's tax returns that is on millionaires' returns. It shows how much of the state's income is held at the very top. It rises and falls with the stock market, since so much of that income is profit from selling investments. Adjusted gross income on returns of $1 million or more, as a share of all adjusted gross income.	IRS Statistics of Income, state data `IRS SOI Historic Table 2, AGI $1,000,000 or more`	rate	cannot combine	52 states	52	0.23	—
Per capita income (US$) All the money income in the area divided by everyone who lives there, children included. It is an average, so a few high earners pull it up, unlike the middle household income. An average per person, so bigger places get more weight when areas are combined.	Census ACS 5-year `ACS table B19301_001`	rate	average weighted by total population, so bigger places count more	3,220	52	0.22	0.59
Share Black, not Hispanic (%) The percentage of residents who are Black or African American alone and not Hispanic. The middle county is under 2%, and this group is one of the most concentrated in the data. A handful of counties in the Mississippi Delta and the Black Belt are above 80%.	Census ACS 5-year `ACS table B03002_004, divided by total population`	rate	average weighted by total population, so bigger places count more	2,390	52	0.22	0.76
Years of potential life lost (per 100,000 under 75) Years of life lost before age 75, per 100,000 people. A death at 25 counts as fifty years lost and a death at 70 as five, so earlier deaths count for more. Years of life lost to deaths before age 75, per 100,000 people. Adjusted so places with older or younger populations can be compared fairly. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v001`	rate	average weighted by population, so bigger places count more	2,897	50 no CT, PR	0.22	0.49
Share in the same house a year ago (%) The percentage of residents who lived in the same home a year earlier. The middle county is 88%, so the low end is the interesting one: college towns, military bases and resort areas turn over fastest. Read it as the opposite of moving around, not as good or bad. A place nobody wants to leave and a place nobody can afford to leave look the same here.	Census ACS 5-year `ACS table B07001_017, divided by total population`	rate	average weighted by total population, so bigger places count more	3,144	51 no PR	0.22	0.12
Dental visit in the past year (%) The estimated percentage of adults who visited a dentist in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES DENTAL, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	3,143	51 no PR	0.22	0.79
Total tax burden (% of personal income) Every state and local tax collected, as a percentage of what the state's residents earned. It is the broadest measure of what living there costs in tax, and it is not the same as any rate written in law. All state and local tax collections as a share of personal income. The headline number in most comparisons between states. Census figures for state and local tax collected, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). It shows the total tax paid as a share of income. It is not the rate written in law, and not the rate on the next dollar earned. Taxes are counted over the government's budget year and income over the calendar year, and state and local taxes are combined.	`Census tax collections, divided by BEA personal income`	rate	average weighted by personal income, so bigger places count more	51 states	51 no PR	0.21	—
Cognitive disability (%) The estimated percentage of adults who report serious difficulty concentrating or remembering. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES COGNITION, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.21	0.66
Any disability (%) The estimated percentage of adults who report any disability. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES DISABILITY, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.21	0.67
Arrivals per 100 people staying (per 100) Tax filers who moved in, for every hundred who stayed put. People who moved in from elsewhere, per 100 people who stayed. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,096	51 no PR	0.21	0.45
Taking blood pressure medication (%) The estimated percentage of adults who with high blood pressure are taking medication for it. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES BPMED, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.20	0.89
Child poverty rate, ages 0-17 (%) The percentage of children under 18 in poverty. It is an estimate from a statistical model, combining survey answers with tax and food-assistance records. When areas are combined, each should ideally get weight by its number of children. SAIPE does not publish that number for each county, so the all-ages population is used instead, and combined figures are approximate. Where precision matters, read the county values directly.	Census SAIPE `Census SAIPE SAEPOVRT0_17_PT`	rate	average weighted by population for whom poverty status is determined, so bigger places count more	3,142	51 no PR	0.20	0.58
Binge drinking (%) The estimated percentage of adults who report binge drinking (several drinks at one time) in the past month. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES BINGE, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.20	0.79
Effective property tax rate on income (% of personal income) Property tax collected, as a percentage of residents' income. It is measured against income, not property value. 'Effective property tax rate' can also mean tax against home value, which is a separate measure here. Property tax collected, divided by personal income. It is not divided by property value. That other meaning of the phrase is shown separately, as the effective property tax rate on home value. Census figures for state and local tax collected, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). It shows the total tax paid as a share of income. It is not the rate written in law, and not the rate on the next dollar earned. Taxes are counted over the government's budget year and income over the calendar year, and state and local taxes are combined.	`Census tax collections times the share of taxes from property, divided by BEA personal income`	rate	average weighted by personal income, so bigger places count more	51 states	51 no PR	0.20	—
Average test score (grade levels) How far above or below the national average children here score on tests, in grade levels. Minus one means about a year behind. It reflects everything in children's lives, not only their schools. How far above or below the national average a county's children score, measured in grade levels. It reflects everything that happened to them before and outside school as much as anything inside it. Time period: pooled over 2009-2019, the years the archive covers; these are pre-pandemic results and describe a decade of testing rather than a single year.	Stanford Education Data Archive `SEDA cs_mn_avg_ol`	rate	average weighted by tests taken, so bigger places count more	2,772	50 no DC, PR	0.20	0.47
Alcohol-involved share of driving deaths (%) The percentage of driving deaths that involved alcohol. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v134`	rate	average weighted by population, so bigger places count more	2,401	50 no CT, PR	0.20	0.13
Social connection score (standard deviations) Social connection: friendships across income lines, friendships across income lines in high school, volunteering, community Facebook pages, membership groups, and loneliness and lack of support (flipped). Friending bias is left out, since it is built from the same numbers as friendships across income lines. The average of 7 measures from 3 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 7 measures (CDC, CHR, SC)`	rate	cannot combine	3,065	51 no PR	0.20	0.61
Frequent physical distress (%) The estimated percentage of adults who report poor physical health on 14 or more days a month. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES PHLTH, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.19	0.57
Two weeks of poor mental health (% of adults) Adults who say their mental health was not good on fourteen or more of the last thirty days. Adults reporting fourteen or more days in the last thirty when their mental health was not good. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _MENT14D, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.19	—
Adults under 65 without health insurance (%) The estimated percentage of adults who aged 18 to 64 have no health insurance. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES ACCESS2, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.19	0.75
Loneliness (%) The estimated percentage of adults who always or usually feel lonely. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES LONELINESS, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.19	0.74
Meets aerobic guidelines (% of adults) Adults who get the recommended amount of aerobic exercise, such as brisk walking, each week. Adults meeting the recommended amount of aerobic activity. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _PAINDX3, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.19	—
Share White, not Hispanic (%) The percentage of residents who are White alone and not Hispanic or Latino. It comes from a table that sorts people by both race and Hispanic origin, so nobody is counted twice. The eight race and ethnicity shares here add up to exactly 100% in every county.	Census ACS 5-year `ACS table B03002_003, divided by total population`	rate	average weighted by total population, so bigger places count more	3,162	52	0.18	0.71
Share of taxes from corporation licences (% of all taxes) The percentage from corporation licence and franchise taxes, charged for being incorporated in the state at all, not on anything earned there. This is how Delaware makes its money: 30.1% of all tax money collected by Delaware and every government inside it comes from companies registered there that operate elsewhere. No other state reaches 5%. Franchise taxes: what a state charges a company for being legally set up there, apart from any tax on its profit. A named part of share of taxes from licences, so it is already counted inside it.	Census government finances `Census government finances LF0027`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.18	—
Independent-living disability (%) The estimated percentage of adults who report difficulty doing errands alone. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES INDEPLIVE, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.18	0.59
Poverty rate (model-based) (%) The percentage of people in poverty, as an estimate from a statistical model, not a direct count. The Census Bureau combines survey answers with tax and food-assistance records, to get figures that small counties' survey answers cannot give on their own. When areas are combined, each gets weight by the number of people whose poverty status the Census can check.	Census SAIPE `Census SAIPE SAEPOVRTALL_PT`	rate	average weighted by population for whom poverty status is determined, so bigger places count more	3,143	51 no PR	0.18	0.56
All natural teeth lost, age 65 and over (%) The estimated percentage of adults who aged 65 and over have lost all their natural teeth. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES TEETHLOST, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	3,143	51 no PR	0.17	0.49
Chronic obstructive pulmonary disease (%) The estimated percentage of adults who have been told they have COPD (chronic obstructive pulmonary disease, a long-term lung disease). It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES COPD, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.17	0.63
Self-care disability (%) The estimated percentage of adults who report difficulty dressing or bathing. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES SELFCARE, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.17	0.58
Membership of volunteering groups (%) The percentage of Facebook users who belong to a volunteering or activism group. The percentage of Facebook users who belong to a group that a computer model judges to be about volunteering or activism. It records joining a group, not actual volunteering. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights volunteering_rate_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,091	51 no PR	0.16	0.41
No leisure-time physical activity (%) The estimated percentage of adults who did no physical activity outside work in the past month. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES LPA, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.16	0.66
Stroke (%) The estimated percentage of adults who have ever been told they had a stroke. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES STROKE, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.16	0.59
Share Hispanic or Latino, any race (%) The percentage of residents who are Hispanic or Latino, of any race. The Census asks about Hispanic origin separately from race. So this figure includes all races, and the race shares next to it count only people who are 'not Hispanic'.	Census ACS 5-year `ACS table B03002_012, divided by total population`	rate	average weighted by total population, so bigger places count more	2,958	52	0.16	0.89
Vision disability (%) The estimated percentage of adults who report serious difficulty seeing. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES VISION, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.15	0.57
Fair or poor self-rated health (%) The estimated percentage of adults who rate their own health as fair or poor. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES GHLTH, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.15	0.62
Membership organisations (per 10,000 people) Membership groups per 10,000 people, such as churches, clubs, unions and sports leagues. They are counted from the government's list of businesses and organizations. Civic, religious, sports and political organisations, counted from the official register of businesses. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v140`	rate	average weighted by population, so bigger places count more	3,144	51 no PR	0.15	0.39
Share of taxes from insurance premiums (% of all taxes) The percentage from taxes on insurance policies sold in the state. Almost every state charges it, and almost none argues about it. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0014`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.15	—
Mobility disability (%) The estimated percentage of adults who report serious difficulty walking or climbing stairs. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES MOBILITY, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.14	0.64
Premature mortality (deaths per 100,000) Deaths before age 75 per 100,000 people. It is adjusted so places with older or younger populations can be compared fairly. Deaths before age 75, adjusted so places with older or younger populations can be compared fairly. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v127`	rate	average weighted by population, so bigger places count more	3,080	50 no CT, PR	0.14	0.53
Obesity (%) The estimated percentage of adults who have a body mass index (a measure of weight for height) of 30 or above. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES OBESITY, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.14	0.61
Arrivals per 100 staying, $200,000 or more (per 100) People moving in for every hundred who stayed, counting only households reporting $200,000 or more. Income is measured in the year they arrived, so someone who moved and then earned more counts as high-income. People in the top income group who moved into the state, per 100 of that group who stayed. From the IRS Gross Migration file. It splits moves by the size of adjusted gross income on the second year's tax return. It is published for states only, since the IRS does not split counties by income. This file leaves out returns with an income below zero. So its totals differ slightly from the state-to-state files the other migration measures come from.	IRS county migration flows `IRS SOI gross migration, AGI class 7`	rate	average weighted by people in households earning $200,000 or more who did not move, so bigger places count more	51 states	51 no PR	0.14	—
Returns reporting unemployment compensation (%) The percentage of tax returns reporting unemployment benefits during the year. Filers who received unemployment insurance payments at some point in the year. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N02300 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,143	51 no PR	0.14	0.71
Net migration per 100 people staying (per 100) People moving in minus people moving out, for every hundred who stayed. Positive means the area gained filers. It says nothing about how many moved each way. Arrivals minus departures. It is exactly the first rate minus the second, since both are worked out per 100 of the same people. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,087	51 no PR	0.13	0.34
Departures per 100 people staying (per 100) Tax filers who moved out, for every hundred who stayed put. People who moved away, per 100 people who stayed. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,111	51 no PR	0.13	0.32
Incarceration rate (per 100,000 aged 15-64) People from this county in jail or prison on an average day, per 100,000 residents aged 15 to 64. That is the age group most often locked up, not the whole population. The middle county is 1,014, about one working-age resident in a hundred. Jail and prison together: everyone from this county who is locked up on an average day, wherever they are held. Rates are per 100,000 residents aged 15 to 64, not per 100,000 people. Time period: from the jail series the Bureau of Justice Statistics collected until 2019, which is also the last year Vera publishes county prison data.	`Vera Incarceration Trends total_incarceration_rate`	rate	average weighted by total population, so bigger places count more	2,479	38 no AK, AR, AZ, CT, DC, DE, HI, ID, LA, MT, PR, RI, VA, VT	0.13	0.17
Gini index of income inequality How unequally household income is shared, on a scale from 0 to 1. Zero would mean every household has the same income. One would mean a single household has all of it. US counties run from about 0.32 to 0.60. It is measured on money income before taxes and government payments, so US figures sit above countries that measure it after taxes and payments. A score of income inequality, from 0 (everyone has the same income) to 1 (one household has it all). The inequality of a larger area cannot be worked out from its parts. Two counties where everyone earns the same can still make an unequal state, if one county is much richer than the other.	Census ACS 5-year `ACS table B19083_001`	rate	cannot combine	3,222	52	0.13	0.25
Net business formation rate (% of establishments) The opening rate minus the closing rate, as a percentage of the place's business locations. It is positive when more businesses open than close. The opening rate minus the closing rate.	Census Business Dynamics Statistics `Census Business Dynamics Statistics, ESTABS_ENTRY_RATE minus ESTABS_EXIT_RATE`	rate	average weighted by business establishments, so bigger places count more	3,144	51 no PR	0.13	0.06
Binge drinking (% of adults) Adults who, at least once in the past month, had five or more drinks at one time (for a man) or four or more (for a woman). Adults reporting binge drinking in the past month. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFBING6, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.13	—
Sleeping less than 7 hours (%) The estimated percentage of adults who usually sleep less than seven hours a night. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES SLEEP, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	3,143	51 no PR	0.12	0.66
Preventable hospital stays (per 100,000 Medicare enrollees) Hospital stays per 100,000 people on Medicare for illnesses that good care from a regular doctor should have prevented. It measures access to care, as seen through hospital stays. From Medicare claims, so it is per 100,000 people enrolled in Medicare, not per resident. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v005`	rate	average weighted by population, so bigger places count more	3,061	50 no CT, PR	0.12	0.44
Lack of reliable transportation (%) The estimated percentage of adults who lacked reliable transport for daily life in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES LACKTRPT, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.12	0.61
Mainline Protestant adherents (% of population) Members of mainline Protestant churches, such as United Methodist, Presbyterian Church (USA), Episcopal and Evangelical Lutheran, as a share of the population. United Methodist, Presbyterian Church (USA), Episcopal, Evangelical Lutheran and their kin. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census MPRTRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.12	0.63
Business opening rate (% of establishments) Openings as a percentage of the place's business locations (the average of this March and last). This is the Census Bureau's own measure. Openings divided by the average number of business sites this year and last.	Census Business Dynamics Statistics `Census Business Dynamics Statistics API, ESTABS_ENTRY_RATE`	rate	average weighted by business establishments, so bigger places count more	3,144	51 no PR	0.12	0.21
Life expectancy (years) How long a baby born here would live if today's death rates stayed the same all its life. It sums up death rates now. It is not a prediction for any real child. Worked out from death rates at each age. It cannot be built up from smaller areas. For two counties together, it has to be worked out from their combined deaths and years lived, not by averaging their two figures. County Health Rankings publishes it for each level, so the site uses the published figure. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v147`	rate	cannot combine	3,060	50 no CT, PR	0.11	0.53
Material security score (standard deviations) Poverty, child poverty, income inequality, and four basic-needs measures from CDC PLACES: food, housing, utilities and transport. It is flipped so higher means less hardship, which is why it is named for security, not for what it counts. Turned around so that higher is better. So it measures the absence of hardship, not hardship itself. The average of 9 measures from 4 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 9 measures (ACS, BRFSS, CDC, SAIPE)`	rate	cannot combine	3,108	51 no PR	0.11	0.59
Mobility score (standard deviations) What happened to children raised poor in this county: their income rank as adults, and their chance of being locked up (flipped). Both describe children born around 1980. So this tells you about the county of forty years ago as much as the county today. The average of 2 measures from 1 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 2 measures (OA)`	rate	cannot combine	1,987	52	0.11	0.45
Incarceration of children raised poor here (share) The share of those same children, who grew up here with poor parents, who were in prison on census day in 2010. Share locked up in jail or prison on 1 April 2010, among children whose parents were at the 25th percentile of income (poorer than about three in four families). Time period: measured around 2015 for children born 1978-1983 who grew up in this county; most no longer live there.	Opportunity Atlas `Opportunity Insights jail_pooled_pooled_p25`	rate	average weighted by children in the mobility cohort, so bigger places count more	1,987	52	0.11	0.33
Education score (standard deviations) Test scores, how much children improve each school year, the trend in scores, and high school graduation. The share of adults with a college degree is left out. Every comparison on this site already adjusts for it, so a score that contained it would be adjusting for part of itself. This is the score here least tied to income. The average of 4 measures from 2 publishers. Each measure is first put on a common scale: how many typical gaps between places (standard deviations) it sits above or below the average. Each is also flipped where needed, so higher is always better. Measures from the same publisher are averaged first, then the publishers are averaged together. Each level is scaled on its own. A county's score shows where it stands among counties, and a state's shows where it stands among states. The two are not on one scale, so do not compare them with each other. This is a summary, not a measurement.	`derived from 4 measures (CHR, SEDA)`	rate	cannot combine	2,859	50 no DC, PR	0.11	0.21
Share separated (%) The percentage of people aged 15 and over who are separated. This is not a fifth group next to the other four. The Census counts a separated person as married with the spouse away, so this is part of the married share and must never be added to it. This is part of 'now married', not a separate group. The Census counts a separated person as married, with the spouse living elsewhere.	Census ACS 5-year `ACS table B12001_007…B12001_016 (2 cells), divided by population aged 15 and over`	rate	average weighted by population aged 15 and over, so bigger places count more	2,256	52	0.10	0.46
Economic connectedness (ratio) How many of a low-income person's friends are high-income, compared with an even mix. One means their friends match the local mix of rich and poor. Below one means their friendships are more divided by income than that. Socioeconomic status means standing in income, schooling and similar. For each lower-status person, take the share of their friends who are higher-status and double it, then average over everyone lower-status. The social capital atlas's main measure. Time period: measured over Facebook users aged 25-44 in 2022, with noise added to protect privacy.	Social Capital Atlas `Opportunity Insights ec_county`	rate	average weighted by children with below-median parental income, so bigger places count more	3,020	51 no PR	0.10	0.64
Black-white residential segregation (index, 0 to 100) How unevenly Black and white residents are spread across the county's neighborhoods, from 0 (evenly mixed) to 100 (completely separate). It is the share of either group who would have to move for the two groups to be spread the same way. It is only calculated where both groups live in large enough numbers. Measured across the county's census tracts (small neighbourhoods). It is the share of Black or white residents who would have to move for the two groups to be spread evenly. Worked out only where both groups live in large enough numbers. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v141`	rate	cannot combine	2,064	51 no PR	0.10	0.43
Religious adherents (% of population) Members, their children, and others who attend regularly, as a share of the population. It is above 100% in thirty very small counties, where a congregation draws people from beyond the county it sits in. Members, their children and others who attend regularly, as a share of the population. It is over 100% in thirty very small counties, where a congregation draws people from a wider area than its own county. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census 2020 files; 2010 from the ARDA, RCMSCY10 and RCMSST10`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.10	0.36
Threat of utility shut-off (%) The estimated percentage of adults who received a notice threatening to shut off a utility in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES SHUTUTILITY, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.10	0.64
Trend in average test score (grade levels per cohort) Whether each new group of students scores higher or lower than the group before, in grade levels per year. Change in the average test score from one year's group of students to the next. Time period: pooled over 2009-2019, the years the archive covers; these are pre-pandemic results and describe a decade of testing rather than a single year.	Stanford Education Data Archive `SEDA cs_mn_tav_ol`	rate	average weighted by tests taken, so bigger places count more	2,227	50 no DC, PR	0.09	0.31
How much faster the top bracket arrives than everyone else (percentage points) How much faster high earners arrive than everyone else. It is the net moving rate for households with $200,000 or more, minus the net rate for all households. Positive means a state gains high earners faster than it gains people in general, which is different from gaining people at all. Florida does both. Vermont only does the first. The top income group's net migration rate minus the rate for all income groups, both from the same file. Positive means a state is gaining high earners faster than it is gaining people in general. That is a different question from whether it is gaining people at all. It is usually the question people mean when they argue about taxes and moving. From the IRS Gross Migration file. It splits moves by the size of adjusted gross income on the second year's tax return. It is published for states only, since the IRS does not split counties by income. This file leaves out returns with an income below zero. So its totals differ slightly from the state-to-state files the other migration measures come from.	IRS county migration flows `IRS SOI gross migration, AGI class 7`	rate	average weighted by people in households earning $200,000 or more who did not move, so bigger places count more	51 states	51 no PR	0.09	—
High blood pressure (% of adults) Adults who have ever been told by a doctor that they have high blood pressure. Adults ever told they have high blood pressure. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFHYPE6, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.09	—
Catholic adherents (% of population) Catholic members as a share of the population. They are counted from church records kept by each diocese, which makes this one of the study's most complete figures. Counted from the records of each diocese (a local church district). That makes this one of the most complete series in the study. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census CATHRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.09	0.56
Food insecurity (%) The estimated percentage of adults who were sometimes or often worried about running out of food. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES FOODINSECU, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.08	0.63
Share of taxes from public utilities (% of all taxes) The percentage from taxes on utility sales: electricity, gas, water and phone and internet service. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0020`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.08	—
Income gap between arrivals and departures (US$) How much richer the people moving in are than the people moving out. Positive means the area is trading up in income, whatever happens to its population. A shrinking place can still be trading up. Average income of arrivals minus average income of departures. Positive means the households moving in earn more than those moving out, whatever happens to the total number of people. A place can be shrinking and still be gaining richer households at the same time. Published for counties and states. Metro area figures are worked out here from county-to-county moves. Moves between counties inside the same metro area are left out, since they do not count as moving to or from the metro area.	IRS county migration flows `IRS SOI migration, county aggregates`	rate	average weighted by people who did not move, so bigger places count more	3,087	51 no PR	0.08	0.19
Frequent mental distress (%) The estimated percentage of adults who report poor mental health on 14 or more days a month. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES MHLTH, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.08	0.68
Depression (%) The estimated percentage of adults who have been told they have a depressive disorder. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES DEPRESSION, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.08	0.80
Fair or poor health (% of adults) Adults who describe their own health as fair or poor, not good, very good or excellent. Adults rating their own general health fair or poor. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _RFHLTH, age-standardised`	rate	cannot combine	49 states	49 no KY, PA, PR	0.07	—
Share of taxes from property (% of all taxes) The percentage of a state's total tax money that comes from property tax. It runs from 62% in New Hampshire, which has no general sales tax and no broad income tax, to under 15% in states that rely on taxes on oil, gas and minerals or on sales. This shows how a state raises its money, not what any owner pays. For that, see the effective property tax rate, the one tax measure here available for each county. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0009`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.07	—
Per capita personal income (US$) Personal income divided by the number of residents, children included. An average per person, so bigger places get more weight when areas are combined.	BEA Regional Accounts `BEA CAINC1 line 3`	rate	average weighted by population, so bigger places count more	3,091	51 no PR	0.07	0.32
Private job growth (% a year) The percentage change in private-sector jobs from the year before. Change in the yearly average number of private jobs from the year before.	BLS Quarterly Census of Employment and Wages `BLS QCEW annual averages, private, all industries, change in annual_avg_emplvl`	rate	average weighted by private jobs, so bigger places count more	3,212	52	0.07	0.01
Unemployment rate (%) Unemployed people as a percentage of the labor force (people working or looking for work), not of the whole population. A place where few people look for work at all can have low unemployment and still have few people working. When areas are combined, each gets weight by its labour force (people working or looking for work), not by population. Using population would let counties with many retirees or children throw off the average.	BLS LAUS `BLS LAUS county annual averages`	rate	average weighted by labour force, so bigger places count more	3,221	52	0.07	0.62
Share of taxes from other taxes (% of all taxes) The percentage from taxes that fit none of the other headings: estate and gift, property transfer, oil, gas and mineral taxes, and a small leftover. It is worth reading only through its named parts, which are shown separately, since the heading itself groups unlike things. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0034`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.07	—
Poverty rate (%) The percentage of people whose family income is below the federal poverty line. The line depends on how many people are in the family and their ages. When areas are combined, each gets weight by the number of people whose poverty status the Census can check, not by total population.	Census ACS 5-year `ACS table B17001_002, divided by population for whom poverty status is determined`	rate	average weighted by population for whom poverty status is determined, so bigger places count more	3,149	52	0.07	0.68
Pretrial share of the jail population (%) Of everyone in the county's jails, the percentage held before trial, not after being found guilty. The middle county is 64.5%, so the typical county jail mostly holds people not yet convicted of the charge they are held on. It is calculated here as people held before trial divided by all people in jail. The publisher's rate column is not a rate per resident, so it is not used. People held before trial, as a percentage of everyone in the county's jails. It is about two thirds nationally. It shows whether a jail mostly holds people found guilty or people only accused. It is worked out here as people held before trial divided by the jail population. Vera's own pretrial rate column is not used, since it is not a rate per resident. Time period: from the jail series the Bureau of Justice Statistics collected until 2019, which is also the last year Vera publishes county prison data.	`Vera Incarceration Trends total_pretrial_custody / total_jail_pop`	rate	average weighted by total population, so bigger places count more	2,744	45 no AK, CT, DE, HI, PR, RI, VT	0.07	0.15
Returns reporting capital gains (%) The percentage of tax returns reporting a capital gain: profit from selling something like stock or property. Filers who made a profit selling things like shares or property during the year. This goes with owning such assets, but also with choosing to sell them. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N01000 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,143	51 no PR	0.06	0.58
High blood pressure (%) The estimated percentage of adults who have been told they have high blood pressure. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES BPHIGH, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.06	0.81
Children in single-parent households (%) The percentage of children who live in a household headed by one parent. Share of children living with one parent. From the American Community Survey, with five years of answers pooled together. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v082`	rate	average weighted by population, so bigger places count more	2,826	51 no PR	0.06	0.47
Jail population rate (per 100,000 aged 15-64) People held in this county's own jails, per 100,000 residents aged 15 to 64. The middle county is 416. It counts everyone in the jail, including people held for other counties, for US Marshals and for immigration enforcement. So a small county with a large regional jail can be in the tens of thousands: Issaquena County, Mississippi is 26,667. Read it as what happens inside the county, not as what its own residents do. People held in this county's jails on an average day. Most of them are waiting for trial, not serving a sentence after being found guilty. Rates are per 100,000 residents aged 15 to 64, not per 100,000 people. Time period: from the jail series the Bureau of Justice Statistics collected until 2019, which is also the last year Vera publishes county prison data.	`Vera Incarceration Trends total_jail_pop_rate`	rate	average weighted by total population, so bigger places count more	2,902	45 no AK, CT, DE, HI, PR, RI, VT	0.06	0.25
Share divorced (%) The percentage of people aged 15 and over who are divorced and have not remarried. It counts who is divorced now, not how many divorces happen. A county where many marriages end but people quickly remarry can show a low figure.	Census ACS 5-year `ACS table B12001_010…B12001_019 (2 cells), divided by population aged 15 and over`	rate	average weighted by population aged 15 and over, so bigger places count more	3,185	52	0.06	0.33
Share of taxes from selective sales (% of all taxes) The percentage from taxes on particular things, not on sales in general: gasoline and diesel, tobacco, alcohol, insurance, utilities and entertainment. The money from these taxes shrinks as people do less of the thing being taxed. A share of the tax money the state and its local governments collect, not a rate anyone pays.	Census government finances `Census government finances LF0012`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.05	—
Share of taxes from occupation and business licences (% of all taxes) The percentage from job and business licences not listed anywhere else: the fees a state or city charges for permission to do business. It is a catch-all by design, so it shows how much a state licenses more than it measures any one tax. A named part of share of taxes from licences, so it is already counted inside it.	Census government finances `Census government finances LF0032`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.05	—
Prison population rate (per 100,000 aged 15-64) People in state prison who were sentenced in this county, wherever they are held now, per 100,000 residents aged 15 to 64. The middle county is 593. It is the reverse of the jail rate: it follows the county's own residents wherever they are sent, instead of counting whoever is inside its walls. People sent to state prison by this county's courts, held anywhere in the state. It describes the county's courts, not a building in the county. Rates are per 100,000 residents aged 15 to 64, not per 100,000 people. Time period: from the jail series the Bureau of Justice Statistics collected until 2019, which is also the last year Vera publishes county prison data.	`Vera Incarceration Trends total_prison_pop_rate`	rate	average weighted by total population, so bigger places count more	2,593	38 no AK, AR, AZ, CT, DC, DE, HI, ID, LA, MT, PR, RI, VA, VT	0.05	0.35
Difficulty walking or climbing stairs (% of adults) Adults with serious trouble walking or climbing stairs. Adults with serious difficulty walking or climbing stairs. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS DIFFWALK, age-standardised`	rate	cannot combine	47 states	47 no DC, KY, NV, PA, PR	0.05	—
High school graduation (%) The percentage of students who finish high school within four years. The share of students who graduate within four years of starting high school, allowing for students who change schools. From state education agencies. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v021`	rate	average weighted by population, so bigger places count more	2,489	47 no CT, ME, NM, OK, PR	0.05	0.28
Share of two or more races, not Hispanic (%) The percentage of residents who report two or more races and are not Hispanic. It rose sharply between the 2010 and 2020 censuses, partly since the Census changed how it sorts written-in answers. So a rise here does not simply mean a change in who lives there.	Census ACS 5-year `ACS table B03002_009, divided by total population`	rate	average weighted by total population, so bigger places count more	2,726	52	0.05	0.56
Share of taxes from tobacco (% of all taxes) The percentage from tobacco taxes. New Hampshire is highest at 2.6%. That reflects its neighbors' high tax rates more than its own smoking: a low-tax state surrounded by high-tax ones sells to people from both. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0017`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.04	—
Share of taxes from gambling (% of all taxes) The percentage from entertainment taxes, which is mostly casino and gambling money. Nevada raises 6.5% this way and Rhode Island 5.0%. For most states it rounds to nothing. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0013`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.04	—
Departures per 100 staying, $200,000 or more (per 100) People moving out for every hundred who stayed, counting only households reporting $200,000 or more. People in the top income group who left the state, per 100 of that group who stayed. From the IRS Gross Migration file. It splits moves by the size of adjusted gross income on the second year's tax return. It is published for states only, since the IRS does not split counties by income. This file leaves out returns with an income below zero. So its totals differ slightly from the state-to-state files the other migration measures come from.	IRS county migration flows `IRS SOI gross migration, AGI class 7`	rate	average weighted by people in households earning $200,000 or more who did not move, so bigger places count more	51 states	51 no PR	0.03	—
State taxes as a share of income (% of personal income) All state government taxes as a percentage of residents' income, from 1970. It covers the state government only, so it is lower than the combined state and local tax load wherever local governments tax heavily. Every state tax collected by the state government, divided by personal income as measured by the BEA (the federal Bureau of Economic Analysis). This is the share of income actually paid, from 1970 onward. State government only, local income taxes excluded; fiscal-year collections over calendar-year personal income.	Census state tax collections `Census Annual Survey of State Government Tax Collections, historical database (T40, C105)`	rate	cannot combine	50 states	50 no DC, PR	0.03	—
Effective property tax rate (% of home value) The middle property tax bill as a percentage of the middle home value. This is the usual 'effective rate': what a typical owner pays each year per dollar of home value. The middle county is 0.82%, ranging from under 0.2% in rural Alaska and Louisiana to over 3% in southern New Jersey and upstate New York. It is the one tax rate here that really varies from county to county, not only from state to state. Even so, much of the gap follows state lines: New Jersey and New York counties sit near 2.2%, and Hawaii and Alabama counties near 0.3%, almost whichever county you pick. The middle tax bill divided by the middle home value. This is the usual published way to measure it. But it is not the same as the middle household's own rate. It cannot be built up from smaller areas, since neither middle value can. The Census caps the tax bill at $10,001. Where bills hit that cap, the rate shows too low, so the most expensive counties look lower than they are.	Census ACS 5-year `ACS table B25103_001, divided by median home value`	rate	cannot combine	3,124	52	0.03	0.77
Received food stamps (%) The estimated percentage of adults who received food stamps in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES FOODSTAMP, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.03	0.58
Share of births to unmarried women (%) Of women who gave birth in the past year, the percentage who were not married: never married, widowed or divorced. A separated woman counts as married. The middle county is 31.8%. It is a share of women who gave birth, not of all women, so it says nothing about how many births there were. A share of women who gave birth in the past year, not of all women. 'Unmarried' here means never married, widowed or divorced. A separated woman counts as married.	Census ACS 5-year `ACS table B13002_007, divided by women who gave birth in the past year`	rate	average weighted by women who gave birth in the past year, so bigger places count more	1,621	52	0.03	0.30
Difficulty running errands alone (% of adults) Adults who have trouble doing errands alone, such as going to a doctor or shopping. Adults with difficulty doing errands alone, such as visiting a doctor or shopping. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS DIFFALON, age-standardised`	rate	cannot combine	46 states	46 no DC, KY, NV, PA, PR, SD	0.03	—
Housing insecurity (%) The estimated percentage of adults who were unable to pay rent or mortgage on time in the past year. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES HOUSINSECU, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.02	0.64
Cancer other than skin cancer (%) The estimated percentage of adults who have ever been told they had cancer other than skin cancer. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CANCER, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.02	0.71
Current asthma (% of adults) Adults who were told they have asthma and say they still have it. It counts current asthma, not everyone who ever had it. Adults who were told they have asthma and still have it. People report this themselves in the Behavioral Risk Factor Surveillance System, a national health survey. This is the survey result for each state, not an estimate from a statistical model. CDC PLACES builds its county estimates from this same survey. Here it is used at the state level, where the survey is designed to give a fair picture. Rates are adjusted to the age mix of the 2000 US population, so the figures do not just reflect which states have older people.	`BRFSS _CASTHM1, age-standardised`	rate	cannot combine	47 states	47 no DC, KY, NM, PA, PR	0.02	—
Drug overdose deaths (deaths per 100,000) Deaths from drug overdose per 100,000 people, from any drug. Source: County Health Rankings, which compiles each measure from its own source and filed here under the last year of the period it covers.	County Health Rankings `County Health Rankings v138`	rate	average weighted by population, so bigger places count more	1,774	51 no PR	0.02	0.52
Learning rate (grade levels per year) How much children improve each school year, in grade levels, where 0 is the national average pace. This is the part that happens in school, and it behaves quite differently from the score level. How much children gain per year of schooling: what changes while they are in school. It moves with income very differently from the average test score does, which is why both are here. Time period: pooled over 2009-2019, the years the archive covers; these are pre-pandemic results and describe a decade of testing rather than a single year.	Stanford Education Data Archive `SEDA cs_mn_lrn_ol`	rate	average weighted by tests taken, so bigger places count more	2,043	50 no DC, PR	0.02	0.33
Arthritis (%) The estimated percentage of adults who have been told they have arthritis. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES ARTHRITIS, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.02	0.78
Returns claiming the earned income credit (%) The percentage of tax returns claiming the earned income tax credit, which goes to working households with low incomes. It is a count from tax records of the working poor, not a survey estimate. A count from tax records of low-income working families. It is separate from the survey and model-based estimates of poverty used elsewhere on this site. The population is filers, not residents.	IRS Statistics of Income `IRS SOI N59660 / N1`	rate	average weighted by tax returns filed, so bigger places count more	3,143	51 no PR	0.02	0.66
Jail admission rate (per 100,000 aged 15-64) Bookings into this county's jails during the year, per 100,000 residents aged 15 to 64. The middle county is 2,033, five times its jail population rate, since most jail stays are short. A county that books many people for one night each has a high admission rate and a low population rate. The two describe different kinds of systems. Bookings into this county's jails over the year. Admissions and population measure different things. A county that books many people for one night each has a high admission rate but a low population rate. Rates are per 100,000 residents aged 15 to 64, not per 100,000 people. Time period: from the jail series the Bureau of Justice Statistics collected until 2019, which is also the last year Vera publishes county prison data.	`Vera Incarceration Trends total_jail_admits_rate`	rate	average weighted by total population, so bigger places count more	2,832	45 no AK, CT, DE, HI, PR, RI, VT	0.02	0.21
Lacking social and emotional support (%) The estimated percentage of adults who rarely or never get the social and emotional support they need. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES EMOTIONSPT, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,299	40 no CO, FL, KY, OR, PA, PR, SD, TN, TX, VT, WA, WY	0.01	0.73
Diagnosed diabetes (%) The estimated percentage of adults who have been told they have diabetes, excluding during pregnancy. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES DIABETES, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.01	0.66
Current asthma (%) The estimated percentage of adults who currently have asthma. It is adjusted so places with older or younger populations can be compared fairly. It is an estimate from a statistical model, not a direct count from a local survey. An estimate from a statistical model, not a direct measurement. The model starts from interviews in a national health survey (BRFSS). It fills in local detail using Census population counts and figures from the American Community Survey (ACS). So the estimate partly reflects ACS measures that also appear on this site. When counties are combined, bigger ones get more weight. That gives a close figure, not an exact one. Each county rate is age-adjusted: adjusted to the age mix of the 2000 US population, so places with older or younger people can be compared fairly. That makes the rates nearly combinable. The exact figure would need counts by age group, which the CDC does not release. Where precision matters, read the county values directly.	CDC PLACES `PLACES CASTHMA, age-adjusted`	rate	average weighted by population aged 18 and over, so bigger places count more	2,956	49 no KY, PA, PR	0.01	0.63
Evangelical Protestant adherents (% of population) Members of evangelical Protestant churches, such as Southern Baptist, Assemblies of God, Lutheran Church-Missouri Synod and about two hundred others, as a share of the population. Southern Baptist, Assemblies of God, Lutheran Church-Missouri Synod and some 200 other bodies. The six groups add up to the published total. Adherents are counted where the congregation is, not where the member lives, and a denomination that did not report is absent rather than zero.	`Religion Census EVANRATE (2010, 2020)`	rate	average weighted by population, decennial census, so bigger places count more	3,144	51 no PR	0.01	0.53
Business closing rate (% of establishments) Closings as a percentage of the place's business locations (the average of this March and last). Closings divided by the average number of business sites this year and last.	Census Business Dynamics Statistics `Census Business Dynamics Statistics API, ESTABS_EXIT_RATE`	rate	average weighted by business establishments, so bigger places count more	3,144	51 no PR	0.01	0.12
Share of taxes from alcohol (% of all taxes) The percentage from taxes on alcoholic drinks. It leaves out the profits of the seventeen states that sell liquor themselves, which count as business income, not tax. So one of those states can look as though it barely taxes alcohol. A named part of share of taxes from selective sales, so it is already counted inside it.	Census government finances `Census government finances LF0016`	rate	average weighted by state and local tax collections, so bigger places count more	51 states	51 no PR	0.01	—
Guns first sold here and recovered in crimes elsewhere (per 100,000 residents) Crime guns recovered in other states that were first sold in this one, per 100,000 of its residents. It shows how much the state supplies crime guns to other states. Crime guns found in other states that were first sold in a shop in this one, per 100,000 residents of this state.	ATF Firearms Trace Data `ATF Firearms Trace Data, Source and Recovery States`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Handgun share of crime guns traced (%) Pistols, revolvers and derringers as a percentage of traced crime guns whose type is known. Pistols, revolvers and derringers, as a share of traced guns whose type is known.	ATF Firearms Trace Data `ATF Firearms Trace Data, Types by State`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Crime guns first sold in another state (%) The percentage of crime guns recovered in the state, and traced to a buyer, that were first sold in another state. It shows how much of the state's crime-gun supply comes across its borders. Share of the crime guns found in the state that were first sold in another state.	ATF Firearms Trace Data `ATF Firearms Trace Data, Source and Recovery States`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Crime guns recovered within a year of first sale (%) The percentage of traced crime guns recovered less than a year after their first sale in a store. The ATF treats a short time from sale to crime as a standard sign of gun trafficking or 'straw' buying, where someone buys a gun for a person who cannot. Only traces that name a buyer and a sale date count. Share of traced crime guns found less than a year after they were first sold in a shop. The ATF treats a short gap as a warning sign of illegal gun dealing, or of someone buying a gun for a person who cannot.	ATF Firearms Trace Data `ATF Firearms Trace Data, Time-to-Crime by State`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Average time from first sale to crime (years) The average number of years between a traced crime gun's first sale in a store and its recovery by police. Average number of years between a traced crime gun's first sale in a shop and the day police found it.	ATF Firearms Trace Data `ATF Firearms Trace Data, Time-to-Crime by State`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Crime guns traced (firearms) Guns recovered by police in the state during the year and traced by the ATF (the federal firearms agency) back to their first sale in a store. Guns found by police in the state during the year and traced by the ATF, the federal agency that tracks where guns were first sold.	ATF Firearms Trace Data `ATF Firearms Trace Data, Types by State`	count	add up	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Crime guns traced per 100,000 residents (per 100,000) Crime guns traced per 100,000 residents: guns recovered by police and traced by the ATF. It rises with how many guns police recover and trace, as well as with how many guns are around. It includes guns recovered from homicides. Guns found by police and traced, per 100,000 residents.	ATF Firearms Trace Data `ATF Firearms Trace Data, Types by State`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—
Adults in a household with a gun (%) RAND's estimate of the percentage of adults living in a home with a gun, 1980-2016. It is an estimate from a statistical model, built from surveys, hunting licences, background checks and the gun share of suicides. Guns people will not tell a survey about, including most illegal guns, are largely missing. An estimate from a statistical model by RAND, a research group, not a direct count. Covers 1980-2016. It mostly leaves out guns held illegally.	CDC WONDER mortality; RAND firearm ownership database `RAND State-Level Firearm Ownership Database, HFR (ODbL 1.0)`	rate	cannot combine	0 states	0 no AK, AL, AR, AZ, CA, CO, CT, DC, DE, FL, GA, HI, IA, ID, IL, IN, KS, KY, LA, MA, MD, ME, MI, MN, MO, MS, MT, NC, ND, NE, NH, NJ, NM, NV, NY, OH, OK, OR, PA, PR, RI, SC, SD, TN, TX, UT, VA, VT, WA, WI, WV, WY	—	—

What “cannot combine” means

A metro area’s median (middle) household income is not the average of its counties’ medians. That number, an average of middle values, means nothing. It is also wrong in a direction that depends on how lopsided the incomes are. Measures marked cannot combineare shown only for the levels their publisher calculates them for. That is why they are missing from the metro row of every pair’s results by level.

Every Moran’s I here is unlikely to be chance, as far as the test can tell. A p-value is the chance of seeing a pattern this strong by luck alone. The test shuffled the data 999 times, so the smallest p-value it can report is 0.001. Nothing measured across US counties is scattered at random on the map. Back to the US overview