Skip to content

535 Final forecast · Nov 3
Dispatches

More

← All dispatches

Movement note AZ-2

Model moves AZ-2 toward Crane, cutting Nez to a 42% chance

Our estimate for the Democrat fell 9.5 points in seven days, over a stretch that included one district survey showing Crane ahead.

Written by the newsroom · anthropic/claude-opus-5 · August 27, 2026 at 6:03 PM EDT

Correction: the forecasts in this article predate a polling-data repair. Some imported polls omitted candidates who shared a party, and ranked-choice rounds were not consistently selected. The current forecasts use corrected results and recalculated pollster adjustments. See current forecasts.

Our model now gives Jonathan Nez, the Democrat, a 42 in 100 chance of winning Arizona's 2nd District, down from 51% in our August 20 run. That is a move of 9.5 points of Democratic win probability across seven days, toward the Republican incumbent, Eli Crane.

The mean margin moved with it. Our previous run had the race at D+0.3 on average; this one has it at R+2.0. The spread is roughly unchanged and remains wide: our 5th-to-95th percentile range runs from R+18.3 to D+14.5, which is to say the model treats a comfortable win by either candidate as well within reach.

One district poll appears in the payload. Peak Insights surveyed 400 likely voters August 17-20 and put Crane at 53% and Nez at 42%. That survey finished fielding on the same day as our earlier run, so we cannot say from the payload alone how much of it was already priced in. It is the only AZ-2 polling we have, and its direction matches the direction the model moved. Some of the change may also reflect inputs beyond this district; the payload does not decompose the shift, and we are not going to guess at the split.

For context, our national average of 24 generic-ballot polls over the past 45 days stands at D+6.6, and our model gives Democrats a 94% chance of winning at least the 218 seats needed for House control, with a mean of 240.9 seats. That figure carries a caveat of our own making. Our simulation correlates error nationally, by census division, by state and within each district, but it has no term linking demographically similar districts across regions, because we lack the cluster data to build one. The practical effect is that our House control probability is marginally firmer than a fuller model's would be.

Polls cited

  • Peak Insights · Aug 20, 2026 · n=400 · AZ-2

Every constant behind these numbers is on the methodology page.