Sponsored by A Fight Worth Having (super PAC, FEC C00933085); conducted by RedBox Research, AFWH's polling unit.
Population: Likely voters (AFWH house estimate) in the declared poll frame (FL); weighted to the November 3, 2026 general electorate as the state's 2026 turnout verdict defines it: each margin from its own source, age group: the turnout model's predicted 2026 electorate, trained only on elections since 2016; congressional district: the turnout model's predicted 2026 electorate, trained only on elections since 2016; gender: the turnout model's predicted 2026 electorate, trained only on elections since 2016; party registration: midpoint of count completion at 2022 after-Friday pace and 2022 after-Friday conversion of 2026 not-yet-voted registrants, from Florida Division of Elections ballots by registered party as of the Division of Elections report of October 30, 2026, paired with the same pre-election day in 2022. Its party registration target was revised on October 31, 2026, after fieldwork and after earlier results had been seen; the verdict's other targets were set in advance. Race is weighted to the 2022 general electorate's composition by the race voters recorded on the state's voter file, with a verified surname-and-geography estimate for voters who recorded none. The AFWH house estimate: the average of the weighting recipes held to the 2024 vote that passed our checks for this poll (recalled-vote recipe, recalled-vote recipe plus turnout groups, recalled vote among verified 2024 voters and recalled vote among verified 2024 voters plus turnout groups), each on our likely-voter model's turnout step, raked to the poll's electorate targets and to a 2024 presidential vote target of Harris 44.0% of the Harris and Trump vote through recalled 2024 vote and the Census college share of the geography's registered citizens. (FL).
Likely-voter rule: No question screened respondents out of the interview. Each respondent's voter-file turnout score was updated by their answer to the vote-intention question (Q13), with an odds ratio of 2.42 between the lowest and highest answers. Respondents who said they had already voted (Q1) counted as certain to vote; a missing answer kept the file score; anyone else whose file score was zero got no weight. Each respondent's weight was multiplied by the updated score before raking to the targets.
Each person counts by their chance of voting: the turnout score our voter-file model gives them, adjusted by how likely they said they are to vote.
Sample frame: 13,427,545 voter-file records in universe "FL 2026 general" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 1,077; 223,636 sent a text invitation.
Sample frame: 458,052 voter-file records in universe "FL 2026 general — FL-13 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 107.
Sample frame: 445,455 voter-file records in universe "FL 2026 general — FL-14 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 91.
Sample frame: 431,922 voter-file records in universe "FL 2026 general — FL-25 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 90.
Sample frame: 378,186 voter-file records in universe "FL 2026 general — FL-27 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 76.
Sample frame: 443,967 voter-file records in universe "FL 2026 general — FL-22 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 89.
Sample frame: 458,052 voter-file records in universe "FL 2026 general — FL-13 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 107; 16,401 sent a text invitation.
Sample frame: 445,455 voter-file records in universe "FL 2026 general — FL-14 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 91; 15,041 sent a text invitation.
Sample frame: 431,922 voter-file records in universe "FL 2026 general — FL-25 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 90; 14,220 sent a text invitation.
Sample frame: 378,186 voter-file records in universe "FL 2026 general — FL-27 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 76; 11,051 sent a text invitation.
Sample frame: 443,967 voter-file records in universe "FL 2026 general — FL-22 geographic fill for the final-week statewide poll" (State: FL; Office: senate; Turnout model: AFWH's Florida turnout model, trained on the 2022 general election); random draw of 89; 11,854 sent a text invitation.
The estimate pools 6 draws: FL 2026 final-week statewide and leadership — fresh sample (2,501 interviews), FL 2026 final-week statewide — FL-13 geographic fill (142), FL 2026 final-week statewide — FL-14 geographic fill (92), FL 2026 final-week statewide — FL-25 geographic fill (60), FL 2026 final-week statewide — FL-27 geographic fill (59) and FL 2026 final-week statewide — FL-22 geographic fill (44), 2,898 in all. They are weighted as one sample stratified by congressional district: each district's respondents start at the district's share of the likely electorate, each in proportion to their likelihood of voting, and the weights are then raked to the targets.
Modes: Text message invitation to a self-administered web survey. Languages: English and Spanish (278 interviews answered in Spanish, 3 of them partly in English).
Interviews completed October 30, 2026 to October 31, 2026.
n = 2,898 completed interviews; 292,203 sampled voters were sent a text invitation; margin of error ±2.1 percentage points at 95% confidence including a design effect of 1.37.
Effective sample size 2,120; the largest weight is 28.9 times the smallest.
Weighting: a weighted average of several raking recipes' weights to party registration (Conservative 0.1%, Democratic 32.8%, Green 0.0%, Libertarian 0.2%, no party 19.0%, other parties 2.2%, Republican 45.6%); gender (women 53.3%, men 46.5%, gender not recorded 0.2%); age group (18-29 6.8%, 30-44 15.1%, 45-54 13.6%, 55-64 20.1%, 65+ 44.3%); 2024 presidential vote (recalled) (Harris 40.0%, did not vote or voted for someone else 9.2%, Trump 50.8%); congressional district (FL-01 4.0%, FL-02 3.9%, FL-03 3.8%, FL-04 3.6%, FL-05 4.2%, FL-06 4.4%, FL-07 3.9%, FL-08 4.0%, FL-09 3.3%, FL-10 2.5%, FL-11 4.3%, FL-12 3.5%, FL-13 4.1%, FL-14 3.5%, FL-15 3.8%, FL-16 4.3%, FL-17 4.5%, FL-18 3.1%, FL-19 4.0%, FL-20 2.7%, FL-21 4.3%, FL-22 3.4%, FL-23 2.9%, FL-24 2.1%, FL-25 3.5%, FL-26 2.6%, FL-27 2.9%, FL-28 2.8%); education (college degree) (college degree 39.7%, no college degree 57.5%, not recorded 2.8%); race and ethnicity (voter-file model) (Asian or Pacific Islander 1.8%, Black 10.6%, Hispanic 14.0%, other races 2.2%, White 71.4%), from the 2026 electorate targets of the state's 2026 turnout verdict; base weights are the share of the likely electorate in each congressional district, divided among the district's respondents in proportion to their likelihood of voting.
Estimate: Each respondent's weight combines two families of weighting recipes anchored to the 2024 presidential vote: one to the 2024 target, one to the 2024 vote of respondents the voter file shows voted, each named candidate held to its share of that target. The 2024 presidential vote target is not the certified result: it is Harris 43.6%, Trump 55.4% and everyone else 0.9% of the 2024 vote (Harris 44.0% of the Harris and Trump vote). Source: The 2024 presidential vote of this year's projected Florida electorate, from voter-file 2024 vote history and certified county and precinct results (Harris 42.7% of the two-party vote among projected voters with a 2024 vote record), adjusted for the rate at which voters now misremember their 2024 vote, as measured by re-interviewing the same panelists (G. Elliott Morris / Verasight, October 22, 2026): Harris 44.0% of recalled two-party vote. The 2024 presidential vote target was set on November 1, 2026, after fieldwork and after earlier results had been seen. Within a family, the recipes that pass the partisan checks are averaged, each recipe's weights scaled to a mean of 1 first. Each family counts in proportion to how often its average fits voter-file facts no recipe weights to best (2020 and 2022 turnout, registration year and county, and where the state's file of returned ballots is held, the early vote by party) over 500 resamples of respondents: 2024 recall (2 recipes), 1.6%; corrected 2024 recall of file-validated 2024 voters (2 recipes), 98.4%. Each recipe rakes the likely voters to its targets, its weights trimmed at the 1st and 99th percentiles; recipes off the electorate's party mix by more than 3 points or the 2024 target by more than 5 were left out. Recipes without a 2024 vote anchor were computed as sensitivities only. A question asked of a random part of the sample is read on the same weights of the respondents who answered it.
Weight trimming: Each recipe in the estimate raked the design weights to its targets with no bound on the adjustment, then trimmed its final weights at their 1st and 99th percentiles and raked them again; averaging the recipes trimmed no weight.
Processing: completed interviews only; partial interviews are excluded. Quality checks flagged 2842 opened surveys not completed at the time of the check, 3 interviews with a recorded duration below 35% of the poll's median interview time. 150 estimates reported, 0 withheld under the reporting rules.
Limitations: The margin of error (2.1 points at 95% confidence, including the design effect) describes sampling error for a result near 50% among all respondents. Results for subgroups carry larger margins, and question wording, coverage and nonresponse can also affect results. The margin of error on the gap between two candidates is roughly twice the poll's margin of error; each race's own figure is given in the fine print with its result. The sponsor is a political committee (a super PAC). Readers should weigh results with that sponsorship in mind. Respondents could be reached only through a telephone number matched to their voter record; voters without a usable number were outside the sample frame. Respondents answered online; people unable or unwilling to complete a web survey are represented only through weighting. Voters who answered an earlier wave were excluded from this sample, so no respondent was re-interviewed. Where that exclusion left a sampling cell with fewer voters than planned, every remaining voter in the cell (676 voters in 285 cells) was selected with certainty and weighted accordingly. 61 respondents had no recorded age group and were not adjusted on that variable. 233 respondents had no recorded congressional district and were not adjusted on that variable. 157 respondents had no recorded race and ethnicity (voter-file model) and were not adjusted on that variable. 12 estimates rest on 50 to 99 respondents and are shown marked “interpret with caution.”
Questions, as written (FL 2026 final-week statewide and leadership, version 5):
1. How likely are you to vote in the November 3, 2026 general election in {place}? [Already voted / Extremely likely / Very likely / Somewhat likely / Somewhat unlikely / Very unlikely / Extremely unlikely / Don't know / No opinion]
2. Who did you vote for in this year's general election for U.S. Senate? [Ashley Moody, Republican / Angie Nixon, Democrat / Neil J. Gillespie, No Party Affiliation / Someone else / Did not vote in this race / Prefer not to answer] (Asked only of respondents who answered Q1 “Already voted”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
3. If the election for U.S. Senate were held today, who would you vote for? [Ashley Moody, Republican / Angie Nixon, Democrat / Neil J. Gillespie, No Party Affiliation / Someone else / Undecided] (Asked only of respondents who were asked Q1 and did not answer “Already voted”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
4. If you had to decide today, which candidate would you lean more toward? [Ashley Moody, Republican / Angie Nixon, Democrat / Neil J. Gillespie, No Party Affiliation / Someone else / Still undecided] (Asked only of respondents who were asked Q1 and did not answer “Already voted” and answered Q3 “Undecided”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
5. Are you definitely or probably going to vote for that candidate? [Definitely / Probably / Don't know / No opinion] (Asked only of respondents who were asked Q1 and did not answer “Already voted” and answered Q3 “Ashley Moody, Republican” or “Angie Nixon, Democrat” or “Neil J. Gillespie, No Party Affiliation”)
6. Who did you vote for in this year's general election for Governor? [Byron Donalds, Republican / David Jolly, Democrat / Scott Eckhard Jewett, Libertarian / Charles Burkett, No Party Affiliation / Frank J. Russo, No Party Affiliation / Moliere "Moe" Dimanche, No Party Affiliation / Dean Ocean Abrams, No Party Affiliation / Jeffrey Peter "Dr. Jeff" Datto, No Party Affiliation / Someone else / Did not vote in this race / Prefer not to answer] (Asked only of respondents who answered Q1 “Already voted”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
7. If the election for Governor were held today, who would you vote for? [Byron Donalds, Republican / David Jolly, Democrat / Scott Eckhard Jewett, Libertarian / Charles Burkett, No Party Affiliation / Frank J. Russo, No Party Affiliation / Moliere "Moe" Dimanche, No Party Affiliation / Dean Ocean Abrams, No Party Affiliation / Jeffrey Peter "Dr. Jeff" Datto, No Party Affiliation / Someone else / Undecided] (Asked only of respondents who were asked Q1 and did not answer “Already voted”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
8. If you had to decide today, which candidate would you lean more toward? [Byron Donalds, Republican / David Jolly, Democrat / Scott Eckhard Jewett, Libertarian / Charles Burkett, No Party Affiliation / Frank J. Russo, No Party Affiliation / Moliere "Moe" Dimanche, No Party Affiliation / Dean Ocean Abrams, No Party Affiliation / Jeffrey Peter "Dr. Jeff" Datto, No Party Affiliation / Someone else / Still undecided] (Asked only of respondents who were asked Q1 and did not answer “Already voted” and answered Q7 “Undecided”) (Candidate order: as on the Florida ballot (Fla. Stat. 101.151(3)(a)-(b)))
9. Are you definitely or probably going to vote for that candidate? [Definitely / Probably / Don't know / No opinion] (Asked only of respondents who were asked Q1 and did not answer “Already voted” and answered Q7 “Byron Donalds, Republican” or “David Jolly, Democrat” or “Scott Eckhard Jewett, Libertarian” or “Charles Burkett, No Party Affiliation” or “Frank J. Russo, No Party Affiliation” or “Moliere "Moe" Dimanche, No Party Affiliation” or “Dean Ocean Abrams, No Party Affiliation” or “Jeffrey Peter "Dr. Jeff" Datto, No Party Affiliation”)
10. Who did you vote for in the 2024 presidential election between Donald Trump and Kamala Harris, or did you not vote? [Donald Trump / Kamala Harris / Someone else / I did not vote / Don't know / No opinion]
11. How much have you thought about this election? [A great deal / Some / A little / Not at all / Don't know / No opinion / Prefer not to answer]
12. When it comes to the November election, would you say you are very enthusiastic about voting, somewhat enthusiastic, not very enthusiastic, or not at all enthusiastic? [Very enthusiastic / Somewhat enthusiastic / Not very enthusiastic / Not at all enthusiastic / Don't know / No opinion] (Asked only of respondents who were asked Q1 and did not answer “Already voted”)
13. On a scale from 0 to 9, where 0 means you definitely will not vote and 9 means you definitely will vote, how likely are you to vote in this election? [0 / 1 / 2 / 3 / 4 / 5 / 6 / 7 / 8 / 9 / Don't know / No opinion / Prefer not to answer] (Asked only of respondents who were asked Q1 and did not answer “Already voted”)
14. When did you cast your ballot in the November 3, 2026 general election? If you voted by mail, please tell us when you mailed or returned your ballot. [On or before October 23 / October 24 or later / Don't know / No opinion] (Asked only of respondents who answered Q1 “Already voted”)
15. How did you cast your ballot in the November 3, 2026 general election? [In person at an early voting location / By mail / By returning a mail ballot to a drop box or election office / In person on Election Day / Another way / Don't know / Prefer not to answer] (Asked only of respondents who answered Q1 “Already voted”)
16. How do you currently plan to cast your ballot in the November 3, 2026 general election? [In person at an early voting location / In person on Election Day / By mail / By returning a mail ballot to a drop box or election office / Another way / Not sure yet / I do not plan to vote] (Asked only of respondents who were asked Q1 and did not answer “Already voted”)
17. When do you currently plan to cast or return your ballot? If voting by mail, please tell us when you plan to mail or return it. [Before Election Day, November 3 / On Election Day, November 3 / After Election Day, November 3 / Not sure yet / I do not plan to vote] (Asked only of respondents who were asked Q1 and did not answer “Already voted” and were asked Q16 and did not answer “I do not plan to vote” or “In person on Election Day”)
18. Thinking back to when you cast your ballot, how did you feel about voting in this election? [Very enthusiastic / Somewhat enthusiastic / Not very enthusiastic / Not at all enthusiastic / Don't know / No opinion] (Asked only of respondents who answered Q1 “Already voted”)
19. What is your age? [18 to 29 / 30 to 44 / 45 to 54 / 55 to 64 / 65 or older / Prefer not to answer]
20. What is your gender? [Man / Woman / Nonbinary or another gender / Prefer not to answer]
21. Are you Hispanic or Latino? [Yes / No / Prefer not to answer]
22. How would you describe your racial background? [White / Black or African American / Asian or Pacific Islander / Native American or Indigenous / Middle Eastern or North African / Other or multiracial / Prefer not to answer]
23. What is the highest level of school you have completed? [Less than high school / High school graduate / Vocational or trade school / Some college, no degree / Associate's degree / Bachelor's degree / Graduate or professional degree / Prefer not to answer]
24. Do you consider yourself a Democrat, a Republican, an independent or a member of another party? [Democrat / Republican / Independent / Another party / Don't know / No opinion]
25. Would you call yourself a strong Democrat or a not very strong Democrat? [Strong Democrat / Not very strong Democrat / Don't know / No opinion] (Asked only of respondents who answered Q24 “Democrat”)
26. Would you call yourself a strong Republican or a not very strong Republican? [Strong Republican / Not very strong Republican / Don't know / No opinion] (Asked only of respondents who answered Q24 “Republican”)
27. And as of today, do you lean more to the Democratic Party, the Republican Party, or neither? [The Democratic Party / The Republican Party / Neither / Don't know / No opinion] (Asked only of respondents who answered Q24 “Independent” or “Another party” or “Don't know / No opinion”)
28. If Democrats win control of the Senate in the 2026 election, which of these people would you prefer to see as Senate Majority Leader? [Chuck Schumer / Bernie Sanders / Someone else / No opinion] (answer options in a random order for each respondent; options such as “Not sure” are not rotated)
29. If Democrats win control of the House of Representatives in the 2026 election, which of these people would you prefer to see as Speaker of the House? [Hakeem Jeffries / Alexandria Ocasio-Cortez / Someone else / No opinion] (answer options in a random order for each respondent; options such as “Not sure” are not rotated)

Methodology appendix
Weight trimming
  Bounds: none: each respondent keeps their design weight and the weighting adjustment is not bounded.
  Weights trimmed: 0 (strategy: each recipe trimmed before the recipes were averaged).
  Design effect before trimming: not recorded.
  Design effect after trimming: 1.367.
Data-defect shadow arm (pollster.shadow-arm.ddc.v1; a pre-registered shadow estimate, not the headline)
  Expected electorate (N): 7,267,006, from the frozen turnout projection.
  U.S. Senate: Angie Nixon, Democrat minus Ashley Moody, Republican among decided respondents (n=2711): headline basis -3.4, shadow arm -6.3 (correlation -0.000574, the state's 2020 value, shrunk toward the pooled value).
  Governor: David Jolly, Democrat minus Byron Donalds, Republican among decided respondents (n=2762): headline basis -1.9, shadow arm -4.7 (correlation -0.000574, the state's 2020 value, shrunk toward the pooled value).
  Assumption: The correlation was measured on CES, an opt-in online panel, under our weighting recipe. Our voter-file estimator has no general-election history, so carrying CES's correlation over is an assumption.
  Assumption: There is no midterm correlation source. The 2020 presidential correlation is applied to 2026 midterm contests.
  Assumption: The correction produces no weights, so it has no design effect of its own. Its uncertainty is not estimated here.
A + recall shadow arm (pollster.shadow-arm.a-plus-recall.v1; a pre-registered shadow estimate, not the headline)
  Recalled 2024 vote: 2898 respondents answered; Harris:Trump set to the certified 2024 ratio in FL (43.4% Harris of the two-party vote).
  U.S. Senate: Angie Nixon, Democrat minus Ashley Moody, Republican among decided respondents (n=2711): headline basis -3.4, A + recall -0.7 (shift +2.7).
  Governor: David Jolly, Democrat minus Byron Donalds, Republican among decided respondents (n=2762): headline basis -1.9, A + recall -0.2 (shift +1.7).
A + education shadow arm (pollster.shadow-arm.a-plus-education.v1; a pre-registered shadow estimate, not the headline)
  Weighted college share 40.9% (n=2812) vs CPS November 2024 registered citizens in FL 40.9% (SE 1.3): gap +0.0 points; the arm computes only beyond 5.
  Status: not triggered. the weighted college share 40.9% is within 5 points of the CPS registered-citizen share 40.9% in FL (gap +0.0)
A + post-ballot partials shadow arm (pollster.shadow-arm.a-plus-partials.v1; a pre-registered shadow estimate, not the headline)
  Post-ballot partials: 0 of 2501 completes (0.0%); breakoff gap by file party: 0.0 points. The arm computes only above 10% or 3 points.
  Status: not triggered. post-ballot partials are 0.0% of completes (rule: above 10%) and the breakoff gap by file party is 0.0 points (rule: above 3)
LV fitted shadow arm (pollster.shadow-arm.lv-fitted.v1; a pre-registered shadow estimate, not the headline)
  Status: awaiting validation. 2898 of 2898 respondents carry a draw-time file score; 0-9 intention groups: 9 2774, 0-4 42, 5-6 17, 7-8 55, no answer 10.
  Coefficients are fitted after the election on these respondents matched to the post-election voter file, by the rule fixed in advance.
Per-state likely voters check (pollster.check.plv-state.v1; the headline recipe with the national and the state's own intention ratio side by side)
  FL: own ratio 1.66 against the national 2.42; back-test: did not beat the national ratio.
  U.S. Senate: Angie Nixon, Democrat minus Ashley Moody, Republican among decided respondents (n=2711): headline basis +6.8, Per-state likely voters +6.9 (shift +0.1).
  Governor: David Jolly, Democrat minus Byron Donalds, Republican among decided respondents (n=2762): headline basis +7.1, Per-state likely voters +7.2 (shift +0.1).
A + covariate undecided allocation shadow arm (pollster.shadow-arm.undecided-covariate.v1; a pre-registered shadow estimate, not the headline)
  Status: unavailable. no respondent answered approval; the undecided-allocation model passed only with every feature family
Turnout level scenarios (carried beside the estimates; not the headline, no weight changes)
  Production level: 54.2% of 13,427,545 active registrants, 7,272,452 ballots.
  Basis-matched level: 54.2% of 13,427,545 active registrants, 7,272,452 ballots.
  FL file: 16,171,659 on the roll, 13,427,545 active; AFWH's Florida turnout model, trained on the 2022 general election (version 18). Both are scored against the certified total ballots cast after the election.
AAPOR outcome rates (Standard Definitions, 10th edition)
  Final dispositions of the 755,831 drawn cases, as of November 1, 2026, 8:25 AM ET (when the release's weighting run was made):
    1.1 Complete interview: 2,898
    2.1133 Replied STOP to the invitation: 19,449
    2.12 Break-off: started the survey (web link or phone) and stopped before completing it: 3,595
    3.20 Invitation delivered, no response: 261,593
    3.219 Invitation not delivered: 4,666
    3.23 Not attempted: no invitation sent and no call dialed: 463,033
    3.9 Invitation sent, no delivery status returned: 2
    4.81 Duplicate listing of a voter already in the sample: 595
  RR1 0.4%; RR3 0.4%; COOP1 11.2%; REF1 3.1%; CON1 3.4%.
  e = 1.0000: proportional allocation (the default of AAPOR Survey Outcome Rate Calculator 5.1 (2023), List Samples sheet) over the 25,942 cases whose eligibility was settled by contact (25,942 eligible, 0 screened out). Duplicate listings are left out because the frame check removed them from every drawn case. The questionnaire has no eligibility screener, and the Standard Definitions treat an unreached named-list case as eligible when none is required.
  Each voter's final disposition is the most informative outcome across every text and call (AAPOR Standard Definitions, 10th edition, pp. 9 and 13): a complete in either mode, then a screen-out, then a refusal or break-off, then the most informative unreached outcome.
  Text message to the web survey: 292,203 cases attempted. RR1 1.0%; RR3 1.0%; COOP1 11.2%; REF1 7.9%; CON1 8.9%. Dispositions: 1.1 2,898; 2.1133 19,449; 2.12 3,595; 3.20 261,593; 3.219 4,666; 3.9 2.
  RR1 by sample frame: FL 2026 final-week statewide — FL-22 geographic fill 0.1% (44 of 30,099); FL 2026 final-week statewide — FL-14 geographic fill 0.2% (92 of 44,899); FL 2026 final-week statewide — FL-13 geographic fill 0.3% (142 of 54,852); FL 2026 final-week statewide and leadership — fresh sample 0.7% (2,501 of 338,397); FL 2026 final-week statewide — FL-27 geographic fill 0.1% (59 of 39,773); FL 2026 final-week statewide — FL-25 geographic fill 0.0% (0 of 39,806); FL 2026 final-week statewide — FL-25 geographic fill 0.2% (60 of 39,839); FL 2026 final-week statewide — FL-13 geographic fill 0.0% (0 of 54,969); FL 2026 final-week statewide — FL-22 geographic fill 0.0% (0 of 27,742); FL 2026 final-week statewide — FL-27 geographic fill 0.0% (0 of 39,986); FL 2026 final-week statewide — FL-14 geographic fill 0.0% (0 of 44,874)
  RR1 by file party: Conservative 0.3% (6 of 1,946); Democratic 0.5% (994 of 214,122); Green 0.5% (3 of 552); Libertarian 0.7% (13 of 1,984); no party 0.3% (527 of 207,418); other parties 0.5% (110 of 23,520); Republican 0.4% (1,245 of 305,694)
  RR1 by age band: 18-24 0.2% (116 of 66,561); 25-34 0.2% (173 of 100,608); 35-44 0.3% (311 of 113,029); 45-54 0.4% (439 of 112,951); 55-64 0.5% (676 of 135,817); 65+ 0.5% (1,183 of 223,452); not recorded 0.0% (0 of 2,818)
  RR1 by race: not recorded 0.2% (59 of 36,654); Native American 0.4% (16 of 3,812); Asian 0.2% (43 of 22,626); Black 0.3% (203 of 63,790); Hispanic 0.3% (424 of 147,339); multi 0.4% (21 of 5,383); other races 0.3% (65 of 20,874); White 0.5% (2,067 of 454,758)
Disclosure supplement (AAPOR Code, June 2026, Section III items stated from the records)
  A1 Data collection strategy: A survey of FL likely voters for the November 3, 2026 general election. Collection: Text message invitation to a self-administered web survey.
  A2 Who collected the interviews: RedBox Research, AFWH's polling unit, drew the sample and collected the interviews itself; no outside fieldwork vendor collected interviews.
  A3 Wording shown or read before the questions:
    Invitation text (wave 1), 198,421 sent: "{greeting}, RedBox Research: a {survey_minutes} MIN poll on Florida's Senate and governor races. {survey_url} STOP to end. HELP for help."
    Invitation text (wave 1), 52,998 sent: "RedBox Research survey: {survey_url} Reply STOP to opt out, HELP for help."
    Invitation text (wave 1), 40,784 sent: "{saludo}, RedBox Research: encuesta de {survey_minutes} min sobre las elecciones de Florida. En español: {survey_url_es} / A {survey_minutes} MIN Florida election poll in English: {survey_url} STOP para salir/to end. HELP ayuda/help"
    Reminder text (wave 2), 178,228 sent: "Hi {first_name}, RedBox Research is still surveying Florida voters ({survey_minutes} MIN): {survey_url} Reply STOP to end, HELP for help."
    Reminder text (wave 2), 53,763 sent: "Hola {first_name}, RedBox Research: ¿nos ayudas con la encuesta de {survey_minutes} min sobre Florida? En español: {survey_url}?lang=es / Florida poll in English: {survey_url} STOP para salir/to end. HELP ayuda/help"
    Reminder text (wave 2), 8,138 sent: "Hi {first_name}, only Boomers take polls. Polls matter. Be counted ({survey_minutes} MIN): {survey_url} RedBox Research. STOP to end, HELP for help"
    Reminder text (wave 2), 2,223 sent: "Hi {first_name}, your answers are saved. Finish the RedBox Research Florida poll: {survey_url} Reply STOP to end, HELP for help."
    Reminder text (wave 3), 25,501 sent: "Hi {first_name}, only Boomers take polls. Polls matter. Be counted ({survey_minutes} MIN): {survey_url} RedBox Research. STOP to end, HELP for help"
    Reminder text (wave 3), 141 sent: "Hi {first_name}, your answers are saved. Finish the RedBox Research Florida poll: {survey_url} Reply STOP to end, HELP for help."
    Words in braces were filled in for each recipient when the text was sent.
    Introduction: the frozen questionnaire has no welcome screen, so the survey opened on its first question.
    The survey was self-administered on the web: there was no interviewer or AI script.
  A4 Likely voters: No question screened respondents out of the interview. Each respondent's voter-file turnout score was updated by their answer to the vote-intention question (Q13), with an odds ratio of 2.42 between the lowest and highest answers. Respondents who said they had already voted (Q1) counted as certain to vote; a missing answer kept the file score; anyone else whose file score was zero got no weight. Each respondent's weight was multiplied by the updated score before raking to the targets.
  A5a Sample type: A probability sample: every voter was drawn at random, with a recorded seed, from the voter-file universe named in the disclosure. No response was generated by AI (no synthetic respondents were used).
  A5b Sample frame and its date: Voters were drawn from the AFWH voter file, built from the Florida official state voter registration file dated June 30, 2026 (received August 13, 2026), for 755,831 drawn voters. Telephone numbers: of the 292,203 voters sent an invitation or called, 172,996 at a number on the official voter file; 119,176 at numbers L2 appended to the voter record (orders 132640, 132877 and 133125); 31 at a number both the official voter file and L2 held.
  A5c Coverage: 292,203 of the 755,236 drawn voters (38.7%) had a textable number and were sent an invitation; the other 463,033 could not be reached by text and are counted as not attempted (AAPOR 3.23).
  A5e Quotas: Sample draws were sized to 77,622 target cells. No interview was stopped or screened out because a cell was full (no case carries a quota-full or screen-out disposition).
  A5i Incentives and contact: No incentive is recorded for this poll, and Pollster has no step that pays or promises one; the wording of the texts sent is under A3. Contact: an invitation text and up to 2 reminder texts.
  A6 Modes: Text message invitation to a self-administered web survey. Languages: English and Spanish (278 interviews answered in Spanish, 3 of them partly in English).
  A8a Interviews by frame and mode: Completed interviews by frame: FL 2026 final-week statewide — FL-22 geographic fill 44; FL 2026 final-week statewide — FL-14 geographic fill 92; FL 2026 final-week statewide — FL-13 geographic fill 142; FL 2026 final-week statewide and leadership — fresh sample 2,501; FL 2026 final-week statewide — FL-27 geographic fill 59; FL 2026 final-week statewide — FL-25 geographic fill 0; FL 2026 final-week statewide — FL-25 geographic fill 60; FL 2026 final-week statewide — FL-13 geographic fill 0; FL 2026 final-week statewide — FL-22 geographic fill 0; FL 2026 final-week statewide — FL-27 geographic fill 0; FL 2026 final-week statewide — FL-14 geographic fill 0. By mode: web 2,898.
  A8c Margin of error: The margin of error is shown because the sample was drawn at random; with a low response rate it holds only as far as weighting corrects nonresponse.
  A10 Quality checks: Each invited voter received a unique survey link; a link yields at most one interview. Quality flags recorded: 3 interviews with a recorded duration below 35% of the poll's median interview time. Flags are measured, not used to drop cases: all 2,898 interviews completed by the status date are in the weighted estimates.
  A11 Total error: The margins of error describe sampling error with the design effect; they are not widened for other sources of error. The margin of error on the gap between two candidates is roughly twice the poll's margin of error; each race's own figure is given in the fine print with its result.
  A12 AI use: (a) Task performed by AI: none in data collection; no AI interviewer or chatbot was used (Text message invitation to a self-administered web survey). AI in processing: No language model coded, weighted or wrote any published result; every published number and sentence is produced by deterministic code from the recorded responses. (b) Human oversight or validation: not applicable; no AI output enters the data or the published results.
  B1 Panel: No panel: respondents were drawn fresh from the voter file for this poll.
  B5 Dispositions:
    Standard: AAPOR Standard Definitions, 10th edition (2023), list samples: section 1.3 (phone surveys of named persons) and section 1.6 (SMS surveys of named persons).
    Final dispositions of the 755,831 drawn cases:
      1.1 Complete interview: 2,898
      2.1133 Replied STOP to the invitation: 19,449
      2.12 Break-off: started the survey (web link or phone) and stopped before completing it: 3,595
      3.20 Invitation delivered, no response: 261,593
      3.219 Invitation not delivered: 4,666
      3.23 Not attempted: no invitation sent and no call dialed: 463,033
      3.9 Invitation sent, no delivery status returned: 2
      4.81 Duplicate listing of a voter already in the sample: 595
    Outcome rates: RR1 0.4%; RR3 0.4%; COOP1 11.2%; REF1 3.1%; CON1 3.4%.
    560,197 text invitations sent, 0 calls dialed.
    Text message to the web survey: 292,203 cases attempted. RR1 1.0%; RR3 1.0%; COOP1 11.2%; REF1 7.9%; CON1 8.9%.
    Each mode counts only the people it tried to reach and only the interviews completed in that mode: a voter texted who then finished by phone is a text non-complete and a phone complete. Each mode's RR3 uses the release's e.
    Each voter's final disposition is the most informative outcome across every text and call (AAPOR Standard Definitions, 10th edition, pp. 9 and 13): a complete in either mode, then a screen-out, then a refusal or break-off, then the most informative unreached outcome.
    e = 1.0000: proportional allocation (the default of AAPOR Survey Outcome Rate Calculator 5.1 (2023), List Samples sheet) over the 25,942 cases whose eligibility was settled by contact (25,942 eligible, 0 screened out). Duplicate listings are left out because the frame check removed them from every drawn case. The questionnaire has no eligibility screener, and the Standard Definitions treat an unreached named-list case as eligible when none is required.