How the lobbying data is built

The data is the public register kept under the Lobbying Disclosure Act, published by Congress at lda.gov. LegiList downloads it from the register's own public API, for every year from 2007 on, and links each filing back to the document the filer submitted.

Three kinds of document are used: registrations (LD-1), which open a lobbying relationship between a registrant and a client; activity reports (LD-2), filed each quarter, with the issues worked on, who lobbied, and the money involved; and contribution reports (LD-203), filed twice a year by registrants and their lobbyists, listing political contributions and certain other payments.

Everything is self-reported by the filers. LegiList shows it as filed and does not correct it; where a figure is left out of totals, it is listed below with the reason.

A filer can amend a report, sometimes several times. Each version stays on the register and on LegiList, and each filing page lists the other versions. Counts and totals use only the current version of each report, so an amendment replaces the original rather than adding to it.

The current version is the one posted most recently for the same registrant, client, year and period (for contribution reports: the same registrant, lobbyist, year and period). One exception: a later original report with nothing in it, such as a no-activity termination filed after the quarter's report, does not replace an earlier report that has activity or money; only an amendment can. Registrations are counted apart from reports.

Where a page says Filings, it counts every document on the register, as lda.gov lists them. Where it says Reports, each report is counted once, in its current version.

A lobbying firm reports the income it received from each client. An organization that lobbies with its own staff reports its own lobbying expenses, using one of three accounting methods it chooses: the Lobbying Disclosure Act's definitions (A), or the tax code's definitions for tax-exempt organizations (B) or for businesses (C). Filing pages show which method was used. Amounts are the filers' own estimates, rounded as the register's rules require; a filer reporting less than the register's threshold gives no figure.

An organization's own expense figure is meant to include what it paid outside firms, so adding the two would count that money twice. Spending totals therefore count, for each organization and year, the larger of what it reported itself and what its firms reported for it. Where the organization's own figure is larger, it already covers the firms; where it is smaller, it plainly did not include them. Pages that show both figures show them side by side and never add them.

Whether a report is an organization's own or a firm's follows the register's own marker for a filer lobbying for itself. Where a different organization files a report that gives a fee received and no expenses, it is counted as a firm's report even if that marker is set on it.

The comparison is made within one organization as LegiList groups it (see client names below). Where a company reports its own lobbying under one name and its firms report for it under another, such as a parent company and a subsidiary, the two are separate organizations and both figures are counted, so totals can be somewhat higher than in sources that combine corporate families.

Amounts disclosed for foreign entities are shown on each filing and not added up: a registration states an expected figure once, a report states a period's figure, and an amendment restates rather than adds.

The register keeps a separate client record for each firm a client hires, and each firm types the name its own way, so one organization can hold dozens of records. LegiList treats records as one organization only when their names match after these clean-ups: capitals and punctuation ignored; legal forms such as Inc., LLC, Corp. and Co. removed; "The" at the start or end removed; a trailing "USA" or "US" removed; a trailing abbreviation removed when it is the name's own initials, as in "(NRECA)"; anything after words such as "formerly", "d/b/a", "and affiliates", "and subsidiaries" or "including subsidiaries" removed; common abbreviations such as Assn. and Natl. written out; and numbers split by a space or hyphen read as one, so "1-800" and "1800" match.

There is no guessing beyond those rules: names that merely look alike stay separate, and affiliated organizations, such as a trade association and its foundation, stay separate. A name such as "Firm X on behalf of Client Y" is kept whole, since the client is the part after "on behalf of". An organization's page gathers every record in its group.

The register has no single identity for a person: a lobbyist gets a new record at each firm. Each record keeps its own page. Where two records are very likely the same person, both pages say so and link to each other, but they are not merged, because the register does not confirm it.

Records are linked only when the full name matches exactly and at least one more thing ties them: the same prior government service described on both (compared on what it says, not its wording, since filers word it differently each time); the same firm filing under both records; or one record ending within a year of the other starting, for a name that is uncommon in the register.

Names filed as placeholders, such as "N/A", are kept off the lobbyist directory; the filings themselves are unchanged.

Contribution reports name the person honored in free text, written differently by each filer. A contribution is credited to a member of Congress only when exactly one member fits the name: the name is read in either order, nicknames and middle names are allowed for, a surname alone is never enough, and the member must have been in office, or a candidate, around the date of the contribution. A name that fits more than one member, or an item naming several people, stays uncredited. So does a former member named with the title of an office held after leaving Congress, such as President or Governor: the contribution honors that later office. A contribution paid to a campaign for a state or local office, such as a run for governor, is not credited either, since it funds that campaign. Each member's page shows the honoree exactly as written so the match can be read.

When the same contribution appears twice on one report, for example equal amounts to the same committee on the same day for a primary and a general election, each is kept. When an organization files a contribution report itself rather than a named lobbyist, the organization is shown as the filer.

Very occasionally a filing's own contents show its amounts are not US dollars spent on lobbying. LegiList leaves those amounts out of totals and rankings, reviewed one at a time and listed here with the reason. A figure is left out only when the filing itself shows it is not what the register's amount fields are for; a figure is never left out for being large or unusual. The filings stay on the site, shown exactly as filed, with a note on the filing page.