How to turn 51 state profiles into a five-state shortlist
A useful shortlist does not begin with somebody else’s “best states” list. It begins by stating your priorities, scoring every state the same way, and then trying to disprove the result.
By statenumbers.com · · 8 min read
Fifty states and the District of Columbia are too many places to research with equal care. The usual solution is to search for a generic ranking, inherit its hidden assumptions, and investigate whichever names appear at the top. A better solution is to build a transparent first-pass model whose assumptions belong to you.
The goal is not to calculate the objectively best place to live. No dataset knows your family, career, climate tolerance, politics, health needs, or community. The goal is to reduce a large search space to a handful of candidates worth examining locally.
Start with deal-breakers, not weights
A weighted score can let a superb result on one measure compensate for an unacceptable result on another. Real decisions do not always work that way. If a state must contain a licensed employer, be within one flight of family, protect a particular right, or meet a climate constraint, apply that as a filter before scoring.
Write these requirements down in plain language. “Reasonable weather” cannot be tested consistently; “fewer than 20 days above 95°F” can. State Numbers does not carry every possible filter, but the structure matters: remove impossible destinations first, then rank the feasible ones.
An illustrative renter-first model
To show the method, we scored all 51 state-level profiles on four measures already in our dataset. The weights describe a renter who values purchasing power and a healthy labor market, while giving substantial attention to housing:
- 35% real median household income: Census median household income divided by the BEA all-items price level; higher is better.
- 25% unemployment: the latest seasonally adjusted BLS state rate in the snapshot; lower is better.
- 25% renter cost burden: median gross rent as a share of renter household income; lower is better.
- 15% housing-rents price level: the BEA housing component relative to the national level; lower is better.
For each measure, the lowest state receives a score near one end of a 0–100 percentile scale and the highest near the other. Ties share the middle of their positions. We multiply the four percentile scores by the stated weights and add them. This avoids mixing dollars, percentages, and index points directly.
What that exact model returns
| Model rank | State | Score | Real median income | Unemployment | Rent burden | Housing RPP |
|---|---|---|---|---|---|---|
| 1 | North Dakota | 86.1 | $86,131 | 2.2% | 25.0% | 71.4 |
| 2 | South Dakota | 83.8 | $84,742 | 2.0% | 25.9% | 67.6 |
| 3 | Iowa | 80.0 | $85,489 | 3.2% | 27.4% | 65.3 |
| 4 | Wyoming | 76.2 | $82,175 | 3.0% | 27.0% | 71.1 |
| 5 | Nebraska | 75.6 | $84,878 | 2.9% | 27.9% | 75.2 |
BEA Regional Price Parities · SARPP · 2024 · rel. February 19, 2026 · Official data
US Census ACS 5-Year Estimates · B19013 / B25071 · 2024 · rel. December 2025 · Official data
BLS Local Area Unemployment Statistics · LAUS · July 2026 · rel. Aug. 21, 2026 · Official data
North Dakota and South Dakota lead this model because they perform well across all four chosen measures. That is not a declaration that they are America’s two best places to live. It is evidence that these weights strongly reward the combination those states currently offer.
Change the question and the answer should change. An owner focused on long-term home value would replace rent burden. A nurse or electrician should use occupational wages instead of household income. A remote worker with a fixed salary might remove local unemployment. Someone who will not tolerate cold winters would filter much of this shortlist before the scoring begins.
A model is useful when it is sensitive to your priorities. If changing a meaningful weight never changes the shortlist, inspect the formula. If tiny weight changes completely reorder it, treat the result as fragile and investigate a wider group.
Do not add a fake tax score
It is easy to place a top income-tax rate beside a combined sales-tax rate and call the result “tax friendliness.” That would create false precision. A top marginal rate is not a household’s effective rate, and sales-tax exposure depends on taxable spending. Property tax depends on the home and locality.
Keep taxes in the model only when you can estimate a comparable dollar liability for the household. Otherwise, use the state tax table as a second-stage checklist. The same principle applies to insurance, child care, health care, and commuting: a personal quote is better than a broad proxy when one is available.
Stress-test before trusting the five
- Change one weight at a time. Double the factor you care about most, then cut it in half. Note which states remain.
- Try alternative measures. Replace overall income with occupational pay or renter burden with likely monthly rent.
- Inspect near misses. The states ranked sixth through tenth may be practically tied once sampling error and personal circumstances enter.
- Move down in geography. Compare metros, then neighborhoods, employers, homes, and commute routes.
- Add the unscored life. Visit, talk to residents, and evaluate community, climate, services, and support networks.
The shortlist is the beginning of the work
A transparent score is valuable because it leaves an audit trail. You can explain why a state appeared, identify which input made it rise, and revise the result when your priorities or the data change. A generic ranking asks you to trust its verdict. A good shortlist gives you five hypotheses to test.
That is the right level of ambition for statewide data: disciplined enough to reveal candidates you might have missed, humble enough to hand the final decision back to the household making it.