Rhode Island

The 2024 workbook carries 39 presidential-only precinct groups and four federal-only ones, each showing zero registered voters and a non-zero ballot count — 10,957 ballots in total from people the register does not contain. Summing precincts by municipality drops them; summing them into a municipality invents registrations that do not exist; and dividing total ballots by total registrations produces a turnout of 66.46% where the municipal figure is 65.07%. A schema that requires every ballot to have a registered voter and a municipality cannot hold eleven thousand real votes.

Rhode Island's counties exist as geography and as court venues and nothing else — the Census counts zero county governments, a distinction it shares only with Connecticut. There are no county commissioners and no county election administrator. Any pipeline that expects a county tier between the state and the municipality has nothing to put in it, and any contact model keyed to a county election office has no office to key to.

39 municipalities, 84 special districts, 4 school districts, no county governments and no townships. That is fewer local governments than a single mid-sized county elsewhere in this corpus — Colorado has 3,715 and Illinois more still. The school tier is four units for the entire state, so almost every city and town runs its schools as a municipal department rather than an independent district. Modelling the local layer as counties containing municipalities containing school districts produces three tiers where the state has one and a half.

Election day is 56.0% of ballots, early in-person 33.2% and mail only 10.8%. Early in-person is three times the mail share here, which inverts the pattern in Vermont, where two-thirds of ballots arrive by mail and early in-person is not separately reported. The municipal range is 18.3% to 58.4%. Any operations plan, contact window or turnout model that treats early voting and mail voting as one bucket describes neither state, and one calibrated on a mail-heavy state gets Rhode Island's timing wrong by weeks.

The municipalities with the lowest turnout also have the lowest early-voting share: Central Falls at 44.0% turnout and 21.3% early, Providence at 49.6% and 18.3%, against Jamestown at 78.4% and 58.4%. The two variables are not independent, so a model that estimates method mix from a statewide average and applies it to a low-turnout jurisdiction overstates early voting exactly where turnout is weakest, compounding the error rather than cancelling it.

vote.sos.ri.gov returns a Cloudflare challenge and sos.ri.gov returns 403, so the state's headline election portal is unreachable to a fetcher. The Board of Elections at elections.ri.gov is a separate agency on a separate host and publishes the precinct-level workbooks openly. Two agencies both legitimately publish Rhode Island election data, they are not the same publisher, and choosing the better-known one yields nothing. Provenance here has to name the agency, not the state.

The Reg_Voters sheet gives a total per precinct and nothing else — no party, no status, no age. Its Election Day, Mail and Early Voting columns exist but are zero throughout, because they describe ballots rather than registrations, and a reader who takes the sheet's column headers at face value concludes that no Rhode Islander is registered to vote by mail. The party breakdown that most states publish alongside registration is simply not in this artifact.

There is no unincorporated territory in Rhode Island and no township tier — 8 cities and 31 towns tile the whole state, and every voter who has a municipality has one of those 39. This is the family invariant that Massachusetts, Connecticut and Vermont share, and it means the municipality is the general-purpose government rather than an island inside a county. Address resolution that falls back to a county when no municipality matches will never fire correctly.