How it works
Methodology and caveats
Everything here is designed to be explained in a sentence. All numbers in this prototype are synthetic.
Station areas
Each station area is a half-mile circle around a station, about a 10-minute walk. Station locations and line alignments here are approximate.
Control corridors
To tell displacement from normal change, we compare station areas with control corridors: dense streets like Lyndale Ave S, Central Ave NE and Rice St that did not get frequent transit. If pressure rises near transit but not on similar streets, transit is a likely factor.
Base mobility
Households move every year. Base mobility is the region’s typical yearly move rate for each group. We only flag displacement when moves out exceed that baseline.
| Group | Moves per year |
|---|---|
| Under 50% AMI | 18.4% |
| 50–80% AMI | 15.1% |
| 80–120% AMI | 12.2% |
| Over 120% AMI | 9.6% |
| Asian | 13.8% |
| Black or African American | 19.6% |
| Hispanic or Latino | 17.9% |
| American Indian | 20.4% |
| White | 11.7% |
| Two or more / other | 16.5% |
Project phases
Every year is tagged with a phase: pre-development, planning, construction or operation, based on approximate milestone years. Pressure often rises during planning, long before opening.
Change types
Each station area gets one change type per year, based on three transparent thresholds.
| Change type | Meaning | Rule (current sliders) |
|---|---|---|
| Stable | Few households at risk and no strong market pressure. | low-income share < 30% AND rent growth < 66th percentile |
| Vulnerable | Many low-income households live here, but strong pressure has not arrived yet. | low-income share ≥ 30% AND rent growth < 66th percentile AND net loss < 20 |
| Early displacement | Many low-income households and one warning sign: fast rent growth or a net loss of low-income households. | low-income share ≥ 30% AND (rent growth ≥ 66th percentile OR net loss ≥ 20 low-income households), but not both |
| Active displacement | Many low-income households, fast rent growth and a net loss of low-income households, all at once. | low-income share ≥ 30% AND rent growth ≥ 66th percentile AND net loss ≥ 20 low-income households |
| Advanced change / Exclusion | Rents are rising fast where few low-income households remain. | low-income share < 30% AND rent growth ≥ 66th percentile |
Test the definition
Move the sliders to see how sensitive the categories are. Inspired by MITRE’s Gentrification Indicators Dashboard.
% of households under 50% AMI
percentile of station areas
households per year
Changing the definition moves 0 station areas between categories.
- Stable48 areas
- Vulnerable0 areas
- Early displacement7 areas
- Active displacement1 areas
- Advanced change / Exclusion27 areas
Pressure and vulnerability
We split the index in two, like Seattle’s dashboard. Pressure measures forces of change; vulnerability measures how exposed residents are. Each component is ranked as a percentile among all station areas for the year, then weighted.
| Component | Group | Weight |
|---|---|---|
| Rent growth (3 years) | Pressure | 45% |
| New development | Pressure | 25% |
| Income change (3 years) | Pressure | 30% |
| Low-income share | Vulnerability | 35% |
| Renter share | Vulnerability | 20% |
| Rent burden | Vulnerability | 25% |
| Households of color | Vulnerability | 20% |
Pressure and vulnerability
Pressure →
- Top right: High pressure, high vulnerability: act early
- Top left: High vulnerability: protect residents now
- Bottom right: High pressure: watch for exclusion
The bivariate map splits each score into thirds (low, medium, high) to make a 3×3 grid.
Confidence
Every value carries a confidence level based on how many households stand behind it: High (500 or more), Medium (50 to 499), Low (under 50).
Where households went
The flow view groups tracked moves by destination. Destinations describe where tracked households moved. These records cannot establish why a household moved or whether the move was voluntary. Destinations in this prototype are synthetic.
Privacy
Only aggregated, de-identified counts are shown. Any cell under 11 households is hidden and labeled “suppressed”.
Limitations
- Moves do not tell us why a household left; resident engagement fills that gap.
- Half-mile circles ignore barriers like rivers and highways.
- Small areas are noisy year to year; look at multi-year trends.
- This prototype uses synthetic data. Real results will differ.