1. What counts as “public support”

A support figure enters the ledger only with all five of these attached:

  1. The polling organization, named.
  2. The field dates — when people were actually asked.
  3. The population sampled — adults, registered voters, and likely voters are not interchangeable, and the difference routinely moves a result several points.
  4. The sample size and the reported margin of error.
  5. The question, verbatim.

The last one carries the most weight and gets skipped the most often. Question wording is where support percentages are made and unmade. “Do you favor background checks on gun purchases?” and “Do you favor a federal registry of gun owners?” can be reported as the same issue and return wildly different numbers. Any entry here prints the question so you can judge it yourself rather than trusting the summary.

Polls commissioned by an organization with a stake in the answer are labelled as such in the entry. They are not automatically excluded — an advocacy poll can be methodologically sound — but you are told who paid.

2. What counts as “spending against”

Only money disclosed in a primary filing: Federal Election Commission reports, federal lobbying disclosures under the Lobbying Disclosure Act, or the state equivalent. Every figure links to the filing.

Each entry states its basis, and the ledger never silently blends bases, because they measure different things:

  • Independent expenditures — money spent to advocate for or against a candidate, reported to the FEC.
  • Lobbying — money spent to influence legislation directly, reported quarterly, and attributed to an issue area rather than a specific bill.
  • Contributions — money given to candidates or committees.

Adding these together produces a bigger, more impressive, and less meaningful number. The ledger reports them separately.

The known undercount

Disclosed money is a floor, not a total. Spending routed through organizations that are not required to disclose donors does not appear in any filing, and therefore does not appear in this ledger. Every figure here should be read as “at least this much, on the record” — never as the full amount. A project about hidden money has to be honest that it can only count the visible kind.

3. What counts as an outcome

Outcomes are recorded as documented facts with a citation — a roll call, a committee action, a withdrawn amendment, or the expiry of a bill at the end of a Congress. Three states are used:

Blocked Weakened Enacted

Weakened exists because most policy does not simply live or die. A measure that passes with its enforcement provision removed, or its scope narrowed to a fraction of what was proposed, is a real and common result. Collapsing it into “passed” would hide the thing this ledger exists to show.

4. What this ledger does not claim

This is the part most likely to be misread, so it is stated as plainly as possible.

Correlation is not causation. The ledger documents that money moved and that an outcome followed. It does not claim that the money caused the outcome, and no entry should be cited as if it did.

Legislation fails for many reasons that have nothing to do with spending: which party controls the chamber, the filibuster, floor time, competing priorities, election-year caution, presidential opposition, drafting problems, and the ordinary fact that some widely popular proposals are genuinely bad policy on inspection. A ledger entry is a data point that invites investigation. It is not a verdict.

There is also a selection problem worth naming. A ledger of cases where popular policy failed will, by construction, contain cases where popular policy failed. It is not a random sample of all legislation, and it cannot be used to estimate how often the pattern holds overall. It shows that the pattern exists and documents specific instances of it. Anyone citing it for a base rate is using it wrong.

5. The strongest argument against this project

The intellectual foundation for a project like this one is usually taken to be Gilens and Page (2014), which examined 1,779 policy issues and reported that the preferences of economic elites and organized business groups were associated with policy outcomes, while the preferences of average citizens, once elite preferences were controlled for, had a near-zero independent effect.

That finding is genuinely contested, and this project is not going to pretend otherwise.

The counterargument, in its strongest form

Subsequent re-analyses — including work by Peter Enns, and by J. Alexander Branham, Stuart Soroka and Christopher Wlezien — argued that the preferences of the affluent and of the middle class are so highly correlated that the two groups almost always want the same thing. On this reading, they disagree on only a small minority of issues, and when they do, the affluent do not clearly prevail. The claim is that the original result is substantially an artifact of trying to separate two variables that move together, and that it has been widely over-read in popular coverage.

If that critique is right, the honest version of this project's premise is considerably narrower: not “the rich always win,” but “on the specific issues where public opinion and concentrated money diverge, the money is worth tracking.” That narrower claim is the one this ledger is actually built on.

This is why entries are individually sourced rather than aggregated into a headline statistic. The ledger does not rest on the disputed finding. Each entry stands or falls on its own primary records, and a reader who rejects Gilens and Page entirely can still check whether a particular poll said what it is claimed to have said and whether a particular filing shows what it is claimed to show.

6. Corrections

If an entry is wrong, it gets corrected in public. The correction appears on the entry itself, dated, with the original claim preserved rather than quietly overwritten. Entries are never silently edited.

If you have a primary record that contradicts something here, that is the most useful thing you can send.

7. The Code

The same five rules the parent project operates under apply to every entry in this ledger:

  • If I cannot verify it, I will not post it.
  • If it is designed mainly to make you hate your neighbor, I will not share it.
  • If I am wrong, I will correct it publicly.
  • I will call out manipulation no matter which side benefits.
  • I am not here to tell you what to think. I am here to help you notice when someone else is trying to think for you.

Open the ledger  →  ·  Back to the front