# Money Map — Methodology **Dataset:** `data/money/money-index.json` **Build date:** 2026-10-01 · **Cycle covered:** 2023-01-01 → present · **Maintainer:** SiLobbyist data layer This document describes exactly how the Money Map's campaign-finance figures were produced, what each field means, and — critically — what the data cannot support. It is written for a working lobbyist: trust the numbers only as far as this document says you can. --- ## 1. Sources **Authoritative source:** Cal-Access, the California Secretary of State's campaign finance and lobbying disclosure system. Every dollar in the Money Map originates as a filed disclosure (Form 460/461/496/497 family — see §2). **Access path used for this build:** the public record feed published by **CalMatters Digital Democracy** (`calmatters.digitaldemocracy.org`), a nonprofit newsroom database that ingests filed contribution records from the Secretary of State. Per its published methodology, Digital Democracy pulls records directly from the SOS and hand-categorizes contributors. **Why not Cal-Access directly?** At build time (2026-10-01) both direct routes were unavailable to automated collection: - `cal-access.sos.ca.gov` (web search) sits behind Incapsula bot mitigation; the filing search returns a JavaScript challenge page to non-browser clients. - The official bulk export (`https://campaignfinance.cdn.sos.ca.gov/dbwebexport.zip`, linked from `sos.ca.gov/campaign-lobbying/.../raw-data-campaign-finance-and-lobbying-activity`) returned empty `HTTP 200` responses (0 bytes) to automated requests; the documentation ZIP on the same host downloaded normally, confirming host-level filtering of the data file. The Digital Democracy feed exposes the same underlying filed records (donor name as filed, transaction date, amount, transaction type) and is itself a public, inspectable source: every legislator entry below links to its source page. **What this means for traceability:** each figure is traceable to `(source_url, transaction date, donor name as filed)`. SOS **filing ID numbers** are not exposed by this feed, so the `filings` array from the original schema is **omitted by design** — documented in §8. No figure in this dataset is estimated, rounded beyond cents, or invented. --- ## 2. Filing types behind the figures California campaign disclosure, in brief (per the SOS Cal-Access table documentation and FPPC filing schedules): | Form | What it is | Money Map use | |---|---|---| | **Form 460** (Recipient Committee Campaign Statement) | The workhorse: filed by candidate and primarily-formed committees each semi-annual/quarterly/pre-election period. Schedules A (monetary contributions received), C (non-monetary), etc. | Primary source of per-legislator contribution records (`RCPT` table in Cal-Access) | | **Form 461** (Independent Expenditure Committee & Major Donor Committee Statement) | Filed by committees making independent expenditures and by major donors | Feeds IE records | | **Form 496** (24-hour Late Independent Expenditure Report) | Filed within 24 hours when an IE of $1,000+ is made in the 90 days before an election | IE spike detection (not separately broken out in this build) | | **Form 497** (24-hour Late Contribution Report) | Filed within 24 hours for contributions of $1,000+ received in the 90 days before an election | Late-money detection (not separately broken out in this build) | | **Form 803** (Behested Payment Report) | Filed by the *payor* (not the official) when payments are made at the behest of an elected official — e.g., to a charity or cause the official solicited | **Not in this dataset** — see §7 | The feed collapses these into transaction types: `Candidate Donations`, `Independent Expenditures`, `Party Committees`, `Gift`, `Sponsored Travel`. --- ## 3. Record fields (as received) Each raw record carries: - `giver.label` / `giver.specific_label` — donor/committee name **as filed** (`specific_label` preferred; it preserves the filer's exact text) - `receiver.label` — the legislator the money is attributed to - `amount` — dollars and cents, as filed - `date` — transaction date (`YYYY-MM-DD`) as filed; **empty for IE records** in this feed - `type` — one of the transaction types above - `description` — schedule detail, e.g. `Itemized Monetary:`, `Unitemized Monetary:`, `Itemized Non-Monetary: ` --- ## 4. Aggregation and normalization rules 1. **Cycle filter.** Only records with `date >= 2023-01-01` enter legislator totals and donor rankings. Records with empty dates are excluded from cycle aggregates (this affects IE records in particular — see §7). 2. **"Total raised" definition.** `total_raised` = sum of `Candidate Donations` records with monetary descriptions (itemized + unitemized) in-cycle. Excludes loans, non-monetary/in-kind contributions, and committee transfers. This matches the colloquial "money raised" but is **not** the committee's cash-on-hand. 3. **Donor roll-up.** Contributions are grouped per legislator by **normalized** donor name: uppercased, accent-stripped, punctuation removed, common entity suffixes (`LLC`, `INC`, `CORP`, `PAC`, `ASSN`, …) stripped for comparison only (display always uses the filed name). Distinct dates are collected per donor→legislator pair into `filing_dates`. 4. **Unitemized contributions.** California committees report sub-$100 (per-election sub-threshold) contributions in aggregate. These buckets are **counted in `total_raised` but excluded from donor rankings** — they are not attributable donors, and ranking them would mislead. 5. **Top-donor cut.** Top 12 donors per legislator by in-cycle monetary total; top 200 donors overall by giving across all legislators in the dataset. 6. **No double counting across legislators.** A donor giving to 5 legislators appears once in `donors[]` with per-recipient breakdown; legislator entries each show their own slice. 7. **Amendments.** If a committee amends a filing, the feed may carry both the original and the amended record. Byte-identical duplicate feed records are collapsed to one (conservative deduplication); near-duplicates that differ in any field are kept, which can modestly inflate a donor's apparent total. Large figures should be verified in Cal-Access before a client sees them. 8. **Vacant seats.** Two seats were vacant at build time (AD-3, SD-10), so the dataset covers 118 seated legislators, not 120. --- ## 5. Donor-type labels (heuristic) Each donor carries a `type` label — `labor`, `corporate`, `trade`, `pac`, `individual`, `party`, `tribal`, `candidate`, or `other` — assigned by **heuristic classification from the filed name** (keyword rules, e.g. union locals → labor, "LLC"/"Inc" → corporate). These are **not official designations** and are wrong often enough to matter: spot-check before citing, and the Money Map UI marks every badge as heuristic. The brain never states a donor's type as fact without that qualifier. ## 6. Lobbyist-client overlap (name-match heuristic) Donors are flagged `lobbyist_client_overlap` when the normalized donor name exactly matches a registered lobbyist client name in `data/lobbyists.json` (1,190 client names). This is a **normalized name-match heuristic**: it catches "BLUE SHIELD OF CALIFORNIA" → "Blue Shield of California" but misses DBA/trade-name variants, and a match does **not** establish that the lobbyist directed the contribution. Absence of a flag proves nothing. Never present a flag as evidence of coordination, and never make unverified negative claims about any donor or official. ## 7. What the feed does NOT contain (by design) - **Independent expenditures:** 60 IE records are included, but the feed exposes **no transaction dates and no support/oppose stance** for them. Amounts and spenders are real filed figures, but they **cannot be cycle-attributed** here and must never be read as support vs. oppose. For stance and timing, pull the committee's Form 496/497 filings in Cal-Access directly. - **Behested payments:** FPPC Form 803 data is **not in this feed** — the `behested` array is empty by design. Use the SOS/FPPC behested-payments search separately. (A separate SOS-sourced behested extract exists in `data/money/behested_raw.json` — 4,018 records, 2024–2026 — for a future merge; it is not yet wired into the index.) - **Gifts and travel:** 50 records, cycle-filtered from 2023-01-01. These are **personal disclosures** (Form 700 family), not campaign support — the UI frames them as influence signal, never as contributions. - **Filing ID numbers:** not exposed by the feed (see §1). Traceability is via `source_url` + transaction date, per legislator. ## 8. What the data cannot support - Cash-on-hand or burn rate (totals are contributions received, not balances). - Attribution of motive: a contribution is a filed fact; *why* it was given is not in this dataset. - Cycle attribution for IE records (no dates in feed). - Support/oppose reading of IE records (no stance in feed). - Completeness for late-2026 filings: the feed was last updated ~May 2026; 24-hour reports (Forms 496/497) near the general election lag. Rebuild before the election. - Donor rankings below the top-200 statewide / top-12 per legislator cut (long-tail splits from name variants are possible — verify large figures). ## 9. Version history - **2026-10-01 (v1, current):** CalMatters Digital Democracy feed build — 118 seated legislators, $221.3M in-cycle filed monetary contributions (2023-01-01 → present), top-200 donors, 60 IE records, 50 gift/travel records. Raw feed snapshots preserved outside the repo's data tree. - **2026-10-01 (superseded):** an SOS Power Search build covering 7 of 120 legislators was produced during a partial SOS outage and then set aside — its per-legislator contribution data was too thin, though its complete IE (8,047 rows) and behested (4,018 records) extracts are preserved in `data/money/ie_raw.json` and `data/money/behested_raw.json` for a future merge. The Power Search methodology draft was not kept.