Method · Version 1 · Settled 29 August 2026
These rules were settled on 29 August 2026 and this page went up before any of the record did. A method written after the results isn't a method, it's an excuse.
Step one
If there's no national poll that fits, the topic doesn't go on the list, and we say that rather than guess.
Every topic on the record links to the exact question it came from. Most of them are in this questionnaire, which carries the wording, the figures, the crosstabs and the methodology for the February and March 2026 surveys.
A question that attributes a policy to a named politician isn't measuring the policy. Those numbers don't qualify.
We don't drop a topic because its survey is old, and we don't have a cut-off. Opinion on most of these things holds steady for years at a time, and then moves quickly when something happens — an older survey on a settled question is better evidence than a recent one taken in the middle of a news event.
The date is the point. A figure measured before something happened is the only way anyone can later see what people thought beforehand. When a topic gets a newer survey, the older figure stays on the page underneath it with its own date, so what you get over time is the movement rather than a single number.
Step two
Not introduced. Not signed on to. Not voted on and failed. Something that took effect and does something toward what Americans said they wanted.
We applaud lawmakers who try to write laws that move toward what the polling shows people want. But if nothing on a topic became law, nothing is reaching anyone. This site is about what is actually moving the country toward what Americans said they wanted. Outside the blog, it is not a place for ideas at the national level that go nowhere and benefit nobody.
If a law moves in the stated direction by even the tiniest amount, it counts. We would rather be accused of being too easy on Congress than too hard. If they can't clear a low bar, that's the finding.
We judge the text of a law, not how well it was carried out. Execution is a separate thing this can't assess.
Not Congress alone. A bill only matters here once it has become law.
State and local action is out of scope, however much of it there is. National polling asks what the country wants; this measures what the federal government did about it. Where a topic is moving mainly in the states, the topic page says so.
A law outlasts the administration that made it. Executive orders are excluded entirely — one president can undo what another did alone, so it isn't the government getting something done.
Rules written by a government department under a law get published on the topic page under Additional Notes, and are never counted. Childcare is the example: a rule capped what families paid, and another rule removed it. Under a laws-only record that would be invisible, and it was the most important thing that happened to that topic.
Step three
Two bills were written and numbered separately, and Congress combined them into one vote. The test is documentary: does the bill that passed contain the text of a bill that was written separately? We never ask whether two things are related — that's an opinion.
If it does, we check every subject in it. If all of them clear the same 65% bar, the vote counts and each subject appears at its own place in the order. If any subject falls below that bar, or has never been polled, we don't record how anyone voted on it.
People were asked about a two-part idea and Congress legislated only one part. Status: only part of it has a bill. Half of a proposal is not the proposal people supported.
Step four
We never say a bill passed. A measure can pass one chamber and still not be law. It became law or it didn't.
A bill is still waiting only if something has happened to it in the record since our last update, with six months allowed from the day it was introduced so new bills aren't called dead early. Everything expires when this Congress ends in January 2027 regardless.
Step five
Step six
Never a result before the method that produced it.
Step seven
About a quarter of American adults aren't registered — younger, more likely to rent. On housing questions that's the group a policy affects most. Both kinds of survey qualify here, but a topic measured among registered voters only carries a line under it saying so, linked to this section. We don't know which way it moves the number, so we don't guess.
A bill carrying hundreds of subjects will always contain something nobody has polled to this standard, so we never record how members voted on those. That takes several topics off the table permanently — tax changes and spending bills essentially only move this way.
National surveys of all adults, with methods and full tables published, are what this project can use. If there's a question on this list you think deserves a proper national survey, we'd rather cite yours than nobody's.
Changes to this page
An earlier draft required two independent national pollsters to agree before a topic qualified. That requirement is gone. One pollster now qualifies a topic, as long as it publishes how it asked and shows its full tables, and as long as we read the number off those tables.
The reason: our rule that one source isn't a source covers claims about what happened — how someone voted, what a bill says. A national survey isn't a claim about an event. It's the measurement, published with its dates, its wording, its sample and its tables. You check it by reading it, which is now required.
Made before anything was published, and written here rather than left out.
Declared interests
Anyone running something like this has stakes. They only mean anything if they're written down before they could matter.
This topic touches my family directly. Writing it down now rather than after the next update.
My prediction: it drops below 65% and comes off the list at the next update. It's already the weakest topic on the page — last in order, and the only one with a party below 60%. If it drops, it gets removed the way any topic would, and the removal shows up in the change log with nothing added. If it holds, it stays for the same reason. Either way, measure the outcome against this.
— Mark
After publication