sourcing
Check every number against its source
Nothing on this site is a score, a grade, an editorial rating or a weighted index. Every figure is a published survey estimate, reproduced with the question it came from and the year it was collected. Here is how to pull it yourself.
the file
Start with the dataset
the questions
Read the question number, not the headline
Three questions carry every figure on this site. They are not interchangeable, they were not asked of the same people, and two of them come from different years.
- Q36, class C07 Caregiving, survey year 2021. Percentage of older adults who provided care for a friend or family member within the past month. 48 states and territories reported it. Missouri 21.8 percent, rank 18. Illinois 16.4 percent. Used on the caregiver share page.
- Q39, class C07 Caregiving, survey year 2021. Average of 20 or more hours of care per week provided to a friend or family member. 48 states and territories reported it. Missouri 33.6 percent, rank 24. Illinois 39.9 percent. Used on the caregiving intensity page.
- Q46, class C01 Overall Health, survey year 2022. Percentage of older adults who report having a disability, a category the survey defines as including limitations related to sensory or mobility impairment. 54 states and territories reported it. Missouri 48.6 percent, rank 11. Illinois 40.4 percent. Used on the physical limits page.
the age cut
Hold the stratification steady
Every figure on this site uses the same slice: stratificationid1 = 65PLUS, which the
file labels "65 years or older", combined with stratificationid2 = OVERALL, meaning not
broken out by sex and not broken out by race. The dataset also carries a "50-64 years" band and an
"Overall" band that mixes the two. Neither is used here. If you pull the file and get a different
number, check that field first.
The data value type is always PRCTG, a percentage. Confidence limits shown on the
ranked pages are the low_confidence_limit and high_confidence_limit
columns exactly as published, at 95 percent.
de-duplication
Prefer the single year row, and say so when there is not one
The 2021 caregiving rows arrive twice for most locations. One row is stamped yearend 2021, a single survey year, and one is stamped yearend 2022, a 2021 to 2022 pooled estimate. They are two different series and they are not always the same number: of the 40 jurisdictions that appear in both, 7 report a different value on each question, and the gap reaches 6.0 points (Ohio publishes 25.9 percent in the single year series and 19.9 percent in the pooled series). The rule applied here: prefer the single year row, and fall back to the pooled row only where a jurisdiction publishes no single year row at all. Forty of the 48 jurisdictions on the two caregiving pages come from the single year series. Eight come from the pooled series because that is the only estimate published for them: Arizona, Colorado, Georgia, New Hampshire, Pennsylvania, Utah, Virginia and Washington. Those eight are marked on both ranked tables. Read the ranking as a snapshot of who reported what, not as a strict like for like comparison, because those eight rows rest on a wider survey window than the other forty. Without a de-duplication rule of some kind every state would appear twice in a ranking, which would be a fabricated list even though every number in it was real.
Rollup rows are also removed before ranking. The file contains a United States row and four census region rows (Midwest, Northeast, South, West) alongside the states and territories. Those are not jurisdictions and never take a rank. The United States row is shown at the bottom of each table, unranked and marked, purely for comparison.
missing years
Expect gaps, and expect them to be real
BRFSS modules are optional. Each state decides, year by year, whether to field a given module, so a state can report the caregiving questions in one year and not the next, and a question can cover 48 jurisdictions in one year and a different count in another. That is why no figure on this site is presented as "current" without a year attached, why the caregiving pages say 2021 and the physical limits page says 2022, and why we do not blend years to make a table look complete.
It also means a rank is a rank among the places that answered, not among all fifty states. Every ranked page states the count in its own methodology box.
what is not measured
Know what these numbers cannot tell you
These are self reported survey estimates of how many people in a place say a thing about themselves. They are not clinical assessments, not diagnoses, and not counts of anybody's need for help at home. A higher percentage in a state does not mean care is worse there, better there, cheaper there or easier to find there. This site makes no claim of that kind and no page here ranks a care provider on anything.
The checklist pages carry no survey data at all. Everything on the room by room page is cited to a named public guidance page from the National Institute on Aging or the CDC, with a live link, and everything on the rides, meals and errands page comes from the agencies' own published service areas.