Methodology & OCDI-Gov

What is being measured, how, and — the part usually omitted — what the measurement cannot see.

The problem of transferring OCDI to a government

The IOCE Ownership–Control Divergence Index was defined for firms, where both terms have legal referents: ownership is equity held, control is votes exercised. Neither referent exists for a government. Nobody owns a legislature. Applying OCDI to the state without saying what replaces those terms would be an equivocation — the same word doing two different jobs.

So we state the substitution explicitly, and accept that it is a substitution:

Read plainly, OCDI-Gov answers: how far does the money that puts a person in office depart from the people who are entitled to put them there? A score of 0 means the funding distribution mirrors the electorate exactly. A score approaching 1 means the money comes from somewhere the electorate is not.

Why geographic units rather than individuals

Computed over individuals, the measure degenerates. Almost every voter contributes $0, so the two distributions are nearly disjoint and OCDI-Gov saturates near its maximum for every member of Congress — technically correct, analytically useless, and it would flatter the thesis by construction. Aggregating to geography (state in v1.1, congressional district in v1.2) keeps the measure discriminating: it varies substantially between legislators and can, in principle, come out low.

This is a deliberate choice to make the metric harder to satisfy. A measure that always returns "maximum divergence" cannot distinguish a captured legislator from a locally-financed one, and would tell us nothing.

Supporting measures

Data sources — all public, all bulk, no keys

Every figure in this report derives from files anyone can download without credentials, registration, or payment. This is a deliberate constraint: a claim about public capture that rests on proprietary data cannot be independently checked by the public it concerns.

The crosswalk is what makes the report possible: it is the only public mapping that links a member's voting identity (ICPSR, used by Voteview) to their financial identity (FEC candidate IDs). 521 of 537 sitting members resolve to at least one FEC committee.

What this measurement cannot see

Stated plainly, because a methodology page that only lists strengths is marketing:

• Agenda control is invisible. Roll calls only record votes that happened. Bills killed in committee, never drafted, or never scheduled leave no trace here. If money operates by determining what gets voted on, this dataset cannot detect it.
• Candidate selection is invisible. By the time someone appears in FEC data they are already a candidate. The filtering effect of money on who can plausibly run happens upstream.
• Dark money is partially invisible. 501(c)(4) organisations may spend on politics without disclosing donors. We can see their expenditures, not their sources.
• Correlation is not transmission. A legislator may attract money because of positions they already hold. Direction of causation is not identified by any analysis on this page.
• Unitemized small donations are compressed. FEC itemisation thresholds mean contributions under $200 appear mostly in summary totals, not transaction records.
• v1.1 is federal only. State, local, lobbying, and media-ownership layers are not built.

Roadmap

Ballot-measure context is informed by Bolts Magazine's "What's on the Ballot". Bolts content is all rights reserved; this report links and cites it and does not reproduce its text.