Winning an election and running a strong campaign are not always the same thing (in fact the two often do not go hand in hand). A candidate can win comfortably while performing below what a state's voting history would suggest, and another can lose while far exceeding it. That distinction is what our measure, PAAR, or Percent Above Anticipated Result, is built to capture.
For each of this year's 35 Senate races, we estimated what a generic candidate from each party should receive, using past Senate and Presidential results, demographics, and incumbency. After November 3, we'll compare those benchmarks to what level of support candidates actually get.
The takeaway so far is fairly straightforward: the competitive races start from very different places.
Michigan is almost evenly divided, with Democrats ahead by just 0.2 points.
Maine starts with a 3.5-point Democratic advantage in a Republican-held seat.North Carolina is the clearest split between our model and the ratings. Our baseline gives Republicans a 3-point edge, while Cook rates the race Likely Democratic.
Ohio, Texas and Iowa begin with Republican advantages of roughly 10 to 13 points.
Georgia's 2.4-point Democratic advantage depends heavily on incumbency.
It's worth noting that these are benchmarks, not predictions of who will win. The more interesting question comes once the votes are counted: which candidates beat the terrain they were handed, and which were bad stewards of their donnor’s support?
Our last post compared 2026 with 2018 at the national level (https://data-calling.beehiiv.com/p/here-we-go-again). This one goes state by state, and the full analysis, including expected vote shares for all 35 races, is on our site: https://www.lpstrategies.com/data/2026-senate-anticipated-results-analysis
The Lincoln Park Strategies Team
