Election Forecast Waits as New Mexico Votes Go Unreported

The New York Times’ live forecast for New Mexico’s Third Congressional District had no votes to count and no county results to show. Its election model waited on estimates as the race between incumbent Democrat Teresa Leger Fernandez and Republican Martin Zamora unfolded. The page listed Fernandez ahead in its latest poll, but that was a pre-election survey, not a ballot tally.

A Forecast Is Not a Vote Count

Research & Polling Inc. surveyed voters from Oct. 1-9 and found Fernandez ahead by 14 percentage points: 53% to 39% for Martin Ruben Zamora. Those figures describe the poll, not election results. With no votes reported on the live page, the forecast hadn’t begun weighing actual returns against its earlier expectations.

The model estimates a winner in real time by combining pre-election data, reported votes and estimates of votes still to come. At the start of election night, it relies on polls, past election results and demographic data. Once returns arrive, the model compares actual results with its pre-election estimates, then uses that comparison to model the remaining vote. County results gradually replace the earlier estimates.

The picture can shift before all ballots have been reported. The page said a steady margin estimate as new results arrive makes the forecast more trustworthy. But the information available at the time described here offered no reported votes to test the polling lead against. A projected advantage and a counted result aren't the same thing.

The Count Moves on Its Own Clock

New Mexico’s absentee-ballot processing must either conclude or pause and resume the next day by 1 a.m. Eastern time. The page also noted that in the last two general elections, nearly all unofficial results were reported by noon Eastern time the day after the election. That timing adds another moving part to the forecast: ballots being processed and results being reported don't necessarily arrive together.

Chaves, Colfax, Curry, Eddy and Harding each showed no votes and 0% reported in the county listings. Those figures describe what the page had received, not proof that nobody had voted in those counties. Until results arrive, the model's early inputs remain pre-election polls, past results and demographic data—the machinery of prediction standing in for the count.

Margins Without a Current Return

The page listed a 2024 House margin of D+13 and a 2024 presidential margin of D+5. Its comparison map uses margins from the 2024 presidential election and displays only areas that have reported almost all their votes. So the map isn't a live picture of every place in the district; areas without near-complete reporting don't appear in it.

The page provided no report of grassroots organizing, mutual aid, or community action, and no legislative proposal to change the process. It also supplied no account of how the candidates’ policies would affect people in the district. What it did offer was a polling lead, a model waiting for returns, and rules and timelines governing ballot processing. For now, the live forecast has estimates. The counties listed have no reported votes.