To determine the county with the worst health insurance coverage in every state, 24/7 Wall St. reviewed five-year estimates of the percentage of the noninstitutionalized civilian population under 65 without health insurance from the U.S. Census Bureau’s 2019 American Community Survey.
Of the 3,220 counties or county equivalents, 3,141 had boundaries that fell within one of the 50 states, while the rest were in the District of Columbia or Puerto Rico.
Counties were excluded if the noninstitutionalized civilian population under 65 was less than 1,000, or if the sampling error associated with a county’s data was deemed too high.
The sampling error was defined as too high if the coefficient of variation — a statistical assessment of how reliable an estimate is — for a county’s under 65 uninsured rate was above 15% and greater than two standard deviations above the mean CV for all counties’ under 65 uninsured rates. We similarly excluded counties that had a sampling error too high for their under 65 noninstitutionalized civilian population, using the same definition.
We selected the under 65 age group because Americans become eligible for Medicare at age 65, and the uninsured rate for the population above this age is less than 1% nationwide. However, because the Census doesn’t publish insurance coverage estimates specifically for the under 65 age group, we aggregated the data from more granular age breakouts.
To ensure each aggregate estimate’s sampling error could be assessed using the definition above, we derived a margin of error for each aggregate estimate using the successive differences replication variance estimation methodology recommended and used by the U.S. Census Bureau.
The remaining 2,945 places were ranked within their state based on their under 65 uninsured rates. To break ties, we used the number of insured people in the same population group.
The share of the population covered by each type of insurance — Medicare, Medicaid, VA, employer, direct-purchase, and Tricare/military — are for the same cohort and are also aggregated from five-year ACS estimates. The estimates reflect people who are covered by that type of insurance alone or in combination with other types on the list. So, when a person is covered by more than one type of insurance, they are included in each group.