← Back to Cities

The Livewell Project

Methodology

50 metrics across 9 factors — all from public, citable data sources

The Thesis

The metrics used to rank "best places to live" measure how well a place works for people who can afford to leave. That's not the same thing as a good place to live.

Orinda, CA has excellent schools and a $252k median income — and the highest "Very High Fire Hazard" acreage of any Contra Costa city, with a largely broken insurance market. Traditional rankings miss this.

The Livewell Project asks harder questions. All methodology, scoring weights, and source data are open source on GitHub.

The 9 Factors

Click a factor to expand its metrics and data sources.

Scoring & Normalization

All raw values are normalized to a 0–100 scale where higher always means better. For metrics where lower is desirable (crime, cost, pollution, risk), scores are inverted before ranking. The displayed value is always the raw real-world number (e.g. $10,800/pupil, 720 violent crimes/100k) — the normalized score is only used for ranking and comparison.

Each city receives a factor rating (Excellent / Good / Fair / Poor) based on the average of its normalized sub-metric scores: Excellent ≥ 85, Good 70–84, Fair 50–69, Poor < 50.

Data resolution note: Renewable energy % and electricity cost are at the state/utility-territory level (EIA). Grid reliability (SAIDI) is by utility territory. Water stress is county/watershed level (WRI Aqueduct). These are the best available public datasets — limitations are noted here for transparency.

Exact Scoring Rubrics

These thresholds define how each raw metric maps to a 0–100 score. All logic is in calculate_all_scores.js on GitHub.

Schools & Education

Public school opt-in rate

Higher = better

≥85% → Excellent · 75–84% → Good · 65–74% → Fair · <65% → Poor
Direct family confidence signal

Per-pupil spending

Higher = better

≥$18k → Excellent · $15–18k → Good · $12–15k → Fair · <$12k → Poor
NCES national median ~$14.3k

Student-teacher ratio

Lower = better

≤15:1 → Excellent · 15–18:1 → Good · 18–22:1 → Fair · >22:1 → Poor
Displayed as rounded integer; used in letters only

Bachelor's attainment

Higher = better

≥75% → Excellent · 60–74% → Good · 40–59% → Fair · <40% → Poor
Adults 25+, Census ACS

Safety

Violent crime / 100k

Lower = better (inverted)

≤150 → Excellent · 150–300 → Good · 300–500 → Fair · >500 → Poor
National avg ~380. Score = 100 − scaled rate.

Property crime / 100k

Lower = better (inverted)

≤1,200 → Excellent · 1,200–2,000 → Good · 2,000–3,500 → Fair · >3,500 → Poor
National avg ~2,100.

Cost of Living

Total family burden

Lower = better (inverted)

≤40% → Excellent · 40–55% → Good · 55–70% → Fair · >70% → Poor
Housing + childcare + food + utilities as % of income

Housing cost burden

Lower = better (inverted)

≤20% → Excellent · 20–28% → Good · 28–35% → Fair · >35% → Poor
HUD standard: >30% = cost-burdened

Childcare burden

Lower = better (inverted)

≤10% → Excellent · 10–18% → Good · 18–25% → Fair · >25% → Poor
% of median HH income spent on one child

Median home price

Display only

Used in affordability ratio (home price ÷ income). Not directly scored.
Zillow ZHVI

Climate & Environment

Wildfire / Flood / Earthquake risk

Lower risk = better (inverted)

FEMA NRI scores inverted: raw score 0 → display 100 (no risk). High risk scores compress toward 0.
Pre-inverted in database. High score = safe.

Air quality (AQI)

Lower = better

Annual median AQI ≤35 → Excellent · 35–50 → Good · 50–75 → Fair · >75 → Poor
EPA Good = ≤50, Moderate = 51–100

Extreme heat days (>95°F)

Lower = better

≤5 days → Excellent · 5–20 → Good · 20–50 → Fair · >50 → Poor
Annual days exceeding 95°F, 30-yr avg

Electricity cost (¢/kWh)

Lower = better

Ranked 1–30. Lower cost = better rank. Displayed as raw ¢/kWh.
EIA state-level residential rate 2023

Drinking water violations

Lower = better

0 → Excellent · 1–2 → Good · 3–5 → Fair · >5 → Poor
Health-based violations 2019–2024

Healthcare

Uninsured rate

Lower = better (inverted)

≤5% → Excellent · 5–8% → Good · 8–12% → Fair · >12% → Poor
CDC PLACES / Census SAHIE

Cancer rate / 100k

Lower = better (inverted)

≤380 → Excellent · 380–420 → Good · 420–460 → Fair · >460 → Poor
Age-adjusted incidence, CDC USCS

Adult obesity rate

Lower = better (inverted)

≤20% → Excellent · 20–25% → Good · 25–32% → Fair · >32% → Poor
BMI ≥30, CDC PLACES

Jobs & Economy

Unemployment rate

Lower = better (inverted)

≤3% → Excellent · 3–4.5% → Good · 4.5–6% → Fair · >6% → Poor
BLS LAUS monthly average

Median household income

Higher = better

≥$100k → Excellent · $75–100k → Good · $55–75k → Fair · <$55k → Poor
Census ACS 5-Year

5-year job growth

Higher = better

≥10% → Excellent · 6–10% → Good · 2–6% → Fair · <2% (or decline) → Poor
BLS QCEW

Civic Health

Walk Score

Higher = better

≥90 → Excellent · 70–89 → Good · 50–69 → Fair · <50 → Poor
Walk Score (0–100)

Voter turnout

Higher = better

≥75% → Excellent · 65–74% → Good · 50–64% → Fair · <50% → Poor
MIT Election Data Lab

Third places / 10k

Higher = better

≥5.0 → Excellent · 3.5–5.0 → Good · 2.0–3.5 → Fair · <2.0 → Poor
Parks, cafes, community centers per 10k residents

Housing Supply

Vacancy rate

Optimal 4–7%

4–7% → Excellent (balanced market) · 2–4% or 7–12% → Good · <2% (crisis) or >12% (blight) → Fair/Poor
Census ACS DP04

Renter cost burden

Lower = better

≤25% → Excellent · 25–35% → Good · 35–45% → Fair · >45% → Poor
>30% = HUD cost-burdened threshold

5-year population growth

Moderate growth best

1–5% → Excellent · 0–1% or 5–8% → Good · Decline or >8% → Fair/Poor
Census Population Estimates

What Makes This Different

  • Public school opt-in rate — We measure whether families across income levels choose public schools — not just test scores. High opt-in is a civic health signal.
  • Affordability ratio — We calculate median home price ÷ median income to show if a teacher can afford to live where they teach.
  • Total family burden — We sum housing + childcare + food + utilities to show the real cost pressure on families — not just rent.
  • Climate as actuarial reality — We treat wildfire, flood, and heat as 30-year financial risks — not lifestyle preferences. Insurance market stability is a leading indicator.
  • Civic infrastructure — Libraries, walkability, and voter turnout are measurable proxies for the density of public life — the infrastructure of belonging.

Limitations & Honest Caveats

  • Growing dataset — Currently 30 cities — expanding based on community nominations. Nominate yours.
  • Snapshot in time — Data reflects 2022–2024 sources; cities change, and we update annually.
  • Averages hide variation — A city's median income or crime rate can mask dramatic neighborhood-level differences.
  • Quantitative focus — Culture, arts, community feel, and intangibles are real — and harder to measure. We make no claim to capture them.
  • No perfect city — Every city has tradeoffs. The goal is to make those tradeoffs visible, not to declare winners.
← Back to CitiesView on GitHub