Read the labels. Facts here are sourced and tiered. measured observed and reconciled; derived arithmetic on observed values; untested a messaging hypothesis with no polling behind it. Every framing below is untested. What is solid is the ground underneath it, not the wording.
Not an estimate of a pool. These are counted records in the CD2 voter file: registered Democrats and unaffiliated voters who are likely to vote in a general and unlikely to vote in a primary. The August primary told us nothing about them because they were never going to be there.
You asked for public data instead of a focus group, and the better evidence was sitting in this district's own past behaviour. Florida publishes official precinct-level results. The turnout archive publishes who cast the ballots, by party registration. Joining them answers the question the whole strategy rests on: do CD2's Democrats and unaffiliated voters actually vote Republican for Congress?
Those votes could only have come from Democrats and unaffiliated voters. It is also a floor, not an estimate: the arithmetic assumes every registered Republican voted Republican, and some did not, so the real number of crossover voters is larger than this.
Every measurement available has an incumbent in it. Dunn was the sitting member in 2022 and 2024, and a chunk of that gap is his personal vote rather than a transferable Republican advantage. Rogers is not an incumbent. The 2018 and 2020 races sit on pre-redistricting lines and cannot be cleanly compared, and there is no open-seat CD2 general on the current map to test against.
So: assume the gap is smaller for Rogers than the 11.5 to 13.9 points Dunn achieved, and plan for the low end. Even half of the 2022 gap is votes, which is larger than his entire primary margin.
The state-employee argument depends on Tallahassee voters being willing to split a ticket. They demonstrably are. In Leon County the Republican congressional candidate ran — in the most Democratic county in the district, the one where state government and FSU are the two largest employers. measured
That is not proof the state-employee framing works. It is proof the behaviour the framing depends on already exists at meaningful scale, which is the part that could not be assumed.
North Florida is not the rest of the state and it is certainly not the country. It is the least urbanised district in Florida, 49% rural, and its economy runs on four things that are all public or public-adjacent: state government and two universities in Tallahassee, an Air Force base that is a third of Bay County's economy, corrections and rural hospitals across the interior, and a coastal fishing economy the state itself regulates.
That matters more than ideology here. These voters do not experience government as an abstraction. It is their employer, their landlord's insurer, their fishing licence, and the base their spouse works on. The persuasion argument that fits this district is not conservative philosophy. It is competence, proximity, and who actually shows up.
Florida works. Washington doesn't. Send someone who knows the difference.
It asks a Democrat or an unaffiliated voter to change nothing about how they see themselves. It rests on something they have lived rather than something they must be argued into, and it points at the institution they already distrust instead of the party they may still belong to.
This is the insight the rest of the field will miss. A Leon County state worker registered Democrat or unaffiliated has spent six-plus years working for a Republican state government, and in that time their pay went up three years running while the federal government they read about lurched between shutdown threats. They have already lived the comparison. Nobody has to sell it to them; it is in their own pay stub.
Rogers's genuine asset here, and the only place his résumé is a pure positive: he was General Counsel to a U.S. Senator and Chief Counsel on committee. Everywhere else that reads as "Washington insider." To a state employee whose agency depends on federal pass-through money, it reads as someone who knows where the money is and how appropriations move. untested
Do not make this about the Governor personally. He is not on the ballot, his numbers among Democrats and unaffiliated voters are a different question from his numbers among Republicans, and naming him turns an experiential argument into a partisan loyalty test. The message is "Florida works", not "vote for the people who run Florida." Also stay out of the university funding fights entirely. FSU and FAMU employ these households and there is no version of that argument that gains us votes.
Rogers is from Panama City. He was here for Michael and for the seven years of rebuilding after it. Green was raised in St. Petersburg and spent those same years working on programmes in Ukraine, Nigeria, Afghanistan and Iraq. That is not an attack and should never be delivered as one. It is a statement about where two people's attention has been, and in a county still rebuilding it does the work by itself.
The insurance bridge. Cost of living is Green's own central message, and property insurance is the sharpest version of it in this county. Do not cede it. A Republican who talks concretely about Panhandle insurance rates is credible in a way that a general "affordability" message is not.
These counties turn out at high rates in generals and are small enough that a few hundred votes move them. The lived experience is a livelihood regulated, closed, and partially reopened by government decisions made elsewhere. It is the same emotional structure as the state-employee argument pointed the other way: decisions about here should be made by someone from here.
With unaffiliated voters the winning version of this is not ideological. It is the gap between what she tells organised constituencies and what she tells North Florida. That framing works on someone who is unaffiliated precisely because they distrust both parties' messaging discipline.
This is the single biggest risk in this district and the reflex most likely to cost us. Anti-government-employee rhetoric that works in a private-sector district is self-harm in a county whose two largest employers are the state and a university, and in rural counties where the prison and the hospital are the payroll. Target Washington's dysfunction, never the people who work for government.
72% of Republican voters support international assistance in principle, on her own endorser's polling. Attack the scale and the priority, never the concept, or she gets to answer with antimalarials and starving children on ground she has spent a decade learning to defend.
Military family background and a Ukraine record close off the first. The Leon County Sheriff's endorsement closes off the second. She had a child in January 2026 and nothing in this campaign touches that.
Her stock-trading ban and term limits poll extremely well with exactly these voters, and a former Senate counsel is the easier man to paint as the insider. Rogers needs his own answer on congressional stock trading, in his own words, early.
Everything above is a hypothesis. The facts are verified; the framings are not. In rough order of value: