Methodology

How a finding gets published

The point of stating a method is that a reader can hold the work to it. This is the standard, including the places it currently falls short.

The five-part test

Each policy question is worked through in the same order:

The stated or observable objective of the policy.
The statutory or budgetary change that was actually made.
The first-order fiscal effect.
The second-order effect on municipalities, households or institutions.
The measurable outcome.

Where the record supports only the first three steps, the downstream claim is treated as unresolved rather than filled in with inference.

Claim, evidence, local impact, counterargument, conclusion

If an analysis cannot get through all five, it is not finished. The counterargument step is the one most often skipped elsewhere, and it is the reason several findings here are weaker than the headline version of the same argument would be.

Every analysis states what it does not show

Every page carries two panels: what the evidence shows, and what it does not. The second is not a disclaimer. It is a list of the specific inferences the evidence will not carry.

This is enforced rather than encouraged. The publishing system refuses to build an analysis whose limits are not stated, so there is no version of this site where that section is quietly dropped for a strong story.

Do not overclaim

If the evidence says a policy may increase municipal costs, it does not say it raised your taxes.
If a programme costs a sum statewide, that is not what your town paid.
If causation has not been established, it is not claimed.
A statewide total can rise while a particular town loses money. Both facts get stated.

Sources

Primary sources are preferred: General Court bill text, roll calls and fiscal notes; Legislative Budget Assistant audits; Department of Revenue Administration tax rate and valuation reports; Department of Education adequacy and cost figures; municipal budget documents.

Where a figure reaches this site through a news organisation rather than directly from the agency, the source list says so. Where an analysis organisation's model is used, it is labelled as an estimate rather than a measurement.

Status, and what it means

Published. Sourced and checked. Indexed, in the feed, and stands as the publication's claim.
In review. Sourced, but a load-bearing claim has not yet been re-checked against the primary record. These pages are readable and clearly marked, and they are kept out of the feed, the sitemap and search-engine indexes until the check is done.

Analyses currently in review, and exactly what each one is waiting on, are listed on the corrections page.

Commentary is labelled

The State Ledger occasionally publishes argument rather than evidence. Those pieces carry the Commentary type and a label on the page. Mixing argument with analysis without saying which is which is how a research desk stops being worth reading.

How pages are connected

Each analysis names the issues, places, people, programs and bills it is about. Every issue, place, person and bill page is generated from those names, so one analysis appears on each page it is filed under without being copied. A piece filed under a town is also county and state evidence. A statewide piece is not automatically about any town.

Town and official pages keep three things apart: analysis filed under them, pieces that only mention them, and county or statewide coverage that applies to them. A mention is not a finding. Naming a legislator does not mean they sponsored or voted for anything, and naming a town does not mean a local effect was measured. Search matches full text, so it can return pieces a town or official page lists only as mentions.

"Read next" links on an article are chosen by the editor where a reading order matters, and otherwise generated from shared officials, bills and places. Sharing only the statewide tag is never treated as a reason to read one piece after another.

Three questions worth remembering

How would we measure that?
What happened here, in this county?
What evidence would change your mind?

Questions about any of this: aaron.gomo@gmail.com.