2026 Midterms Debrief

Only figures from state election offices or AP and network race calls appear here. Each row says whether it is preliminary or certified and when it was pulled. Rows will move to certified as states post their canvasses, and each change is dated in the update log.

Two things about this table. First, the House number will keep moving for weeks; we report the call and the estimate, not a seat count we would have to retract. Second, the Jefferson County race is ours, and it is in the table because we will hold it to the same standard as everything else: the official canvass replaces the news reports the day it posts.

This is our read. Each claim names the evidence it rests on. Where the evidence is thin, we say so.

The environment did most of the work, and it was not local. Democrats flipped seven Senate seats in seven different states, including Texas, Kansas, and Ohio, and more than forty House districts across every region. Swings that broad and that uniform are a national verdict, not forty separate campaign failures. Exit polls reported by CBS put the president's approval at 38 percent, on track in early exit polls to be the lowest for a president in a midterm, and the vice president himself attributed the losses to the Iran war. Evidence: the AP flip list; CBS exit polling; the vice president's statement as reported by Time. What this means for a county campaign: you did not lose a sheriff's race because of Iran, but your top of ticket was carrying weight you could not see from the ground, and your plan for 2028 should assume you will not know the environment until late.

Early vote is where Republican campaigns are losing ground they used to count on. We have one hard data point and it is our own: in Jefferson County, Lowery trailed 5,241 to 4,321 after early voting and won by 229 on Election Day turnout in rural precincts. Evidence: Pine Bluff Commercial canvass reporting. That is one race, and we will not generalize from it to the country. But it is the same pattern Republican operatives have described since 2020, and we have not seen a state where the party's early-vote program closed the gap this year. If you run a 2027 or 2028 race, build the early-vote chase as its own program with its own budget line, and measure it weekly against the county's early-vote file.

Label mattered less than contact. Lowery ran as an independent in a county that votes Democrat down the ballot, and won on a program built around voter contact and Election Day turnout. Evidence: the result, and our own budget. We know what share of his spend reached a voter because we set the target and tracked it; most campaigns we review cannot say. The budget article explains how to calculate the share and what a review of it looks like. We would rather you knew your number than took our word for what it should be.

Candidates who had a reason to exist before the cycle turned did better than the ones who were recruited into an environment. This is the claim in this section with the least evidence behind it, and we label it as such. It rests on the pattern of who survived the wave in the districts we watched, which is not a dataset. We include it because it is the decision candidates are making right now: whether to run in 2028. The second article on this site is about what to build before you announce, so that the answer does not depend on the environment.

Things the data does not settle yet. Each one says what would settle it.

1. The final House margin. What would settle it: AP calls on the remaining districts and the state canvasses through early December.
2. Alaska and Mississippi. Both Senate races were still too close to call on Nov. 6. What would settle it: completed counts and any recount, with certification by each state.
3. Whether the Republican early-vote deficit was general or local. What would settle it: state early-vote files matched to results, by county, which will be available in most states by January. We will publish what we find.
4. Whether the losses were turnout or persuasion. Exit polls cannot answer this. What would settle it: voter-file turnout scores once states post vote history, compared with 2022 and 2024.
5. Whether a non-party label helped down-ballot candidates in hostile counties, or only ours. What would settle it: a list of independent and third-line candidates who won county offices in Democratic counties this cycle, which we are compiling.