Policy Maps

How our policy-impact maps work

Short version: we take a policy's stated focus, trace it through public indicators, and show you a map — always labeled with how confident we are, never dressed up as more certain than it is.

The pipeline, step by step

  1. 1. Start with the policy

    Every policy on this site comes from a candidate's own published platform, tagged by category and by which "equity dimension" it's aimed at — health access, housing affordability, fiscal equity, and so on.

  2. 2. Look up the dimension

    We cross-walk that equity dimension to one or more public indicators — Census/ACS income and poverty data, CDC PLACES health estimates, EPA EJScreen environmental burden scores. Some dimensions don't have a good indicator yet; we say so instead of guessing.

  3. 3. Find the best available geography

    We prefer census tract data — the smallest geography we can map. When only county-level data exists, we use it, but we mark every tract that inherited a county value instead of its own — hatched and faded, never presented as tract-precise.

  4. 4. Rate the confidence

    Every map gets a T1–T5 tier. T1 is data measured directly for that tract. T5 is an indirect proxy standing in for something we can't measure directly yet. The tier is never hidden — it's printed on the map's own caption.

  5. 5. Write the caption honestly

    Every map's caption follows one formula: "This map shows [indicator], a [tier] estimate, as a proxy for who [policy]'s [dimension] focus may affect — not a modeled count of people impacted." We never say how many people a policy would affect. We show a plausible pattern, not a headcount.

Three ways this kind of map can mislead people — and what we do instead

Reading a proxy as a headcount

A shaded tract does not mean "this many people are affected." It means "this tract's value on a related indicator is high or low relative to other tracts." Our captions say "estimate" and "proxy," never "affected population" or "people impacted" — a real headcount would require eligibility microsimulation we don't build in v1.

County data looking as precise as tract data

When a tract only has its county's value because no tract-level data exists, that tract is visually different on every map — hatched pattern, reduced opacity — never painted with the same solid fill as a tract with its own real data.

Comparing two policies that aren't really comparable

When you overlay two candidates' maps, we check how many confidence tiers apart their underlying indicators are. If they're more than one tier apart, we flag the comparison as "not directly comparable" instead of letting two different-confidence maps imply a false equivalence.

The five confidence tiers

Where the data comes from

Indicator values are illustrative placeholders in this v1 build, generated to demonstrate the methodology pipeline on real Dane County tract geometry — not yet a live pull from measured data. Every map and every dataset file says so explicitly. A future wave replaces this with live-baked indicator values, sourced from the Policy Data Infrastructure API where it's populated, and from Census/ACS, CDC PLACES, and EPA EJScreen directly where it isn't.

What's coming next (not built yet)

A grounded "talk to the data" assistant — ask a question in plain language, get an answer sourced to these same indicators — is planned for a later wave. It isn't part of this build; you won't find a chat box here yet.