CD2 Gameplan: July 24 to August 18, 2026

Austin Rogers, FL-02 Republican primary. Prepared 2026-07-24, 25 days out.

Two things frame everything below.

a. 81% of the Panama City market never touches a mail ballot. They vote early (Aug 8-15) or on Election Day. Your vote arrives in an eleven-day window that opens two weeks from now.

b. Panama City has to be 56.1% of every ballot cast from here forward, while it is only 46.7% of what is banked so far. That single number is the scoreboard. It is reported every morning at 7:43 by the VBM tracker.

1. What the research actually supports

Every tactic below is rated against published evidence, not vendor claims. Sources in Section 6.

1a. Turnout work beats persuasion work

This is the strongest finding in the field and it should drive your allocation.

• Face-to-face canvassing raised turnout roughly 9 points in Gerber and Green's New Haven experiment, and produced the largest effects of any method across six later local-election trials.
• Kalla and Broockman pooled 49 field experiments and found the average persuasive effect of campaign contact in general elections is approximately zero.

The critical qualifier, and it works in your favor: their null is for general elections, where party ID gives voters a shortcut that overrides everything a campaign says. A same-party primary removes that shortcut. Persuasion is genuinely possible here in a way it is not in November. But turnout remains the surer buy.

Implication. Spend on identifying and moving your people, not on converting Gross's. Rogers already leads 4-of-4 super voters 26-13. Those people are the asset.

1b. Negative advertising is weaker than it feels

Lau, Sigelman and Rovner's meta-analytic reassessment found no reliable net vote gain from going negative, and some evidence the attacker pays a price. Negative material is more memorable than positive material, which is exactly why campaigns overrate it.

Implication. The contrast on Gross is real and documented, and it should run. But it belongs in a PAC vehicle rather than in Rogers's own mouth, and it should not crowd out the positive biography that is currently producing his 4.3:1 image ratio. The Scenario Lab's attack magnitudes are calibrated to an informed-ballot poll question and should be read as an upper bound.

1c. Endorsements: modest on their own, useful as a signal

• Local newspaper endorsements move roughly 1.3 points of vote share, decisive only in very close races.
• Learning you agree with a candidate on an issue moves about 14 points, an order of magnitude more than a name-based endorsement.
• Endorsement effects are largest in low-information contexts where voters lack a party cue, which describes this race exactly.

Implication. Chase endorsements for what they carry, not for the name. An endorsement that arrives with a reason attached ("he is the only one who has actually lived here") does the work of an issue cue. A bare logo does not. Local sheriffs, county commissioners, and pastors outperform distant figures per unit of credibility because voters can place them.

1d. Social media: format matters more than platform

• Short-form vertical video outperforms static images.
• Behind-the-scenes content is consistently underused by campaigns, which cluster on formal rally and canvassing imagery.
• Whether policy or personal content wins engagement is genuinely unsettled in the literature. Do not let a vendor tell you otherwise.

Implication. Post proof-of-work, not production. See Section 4.

2. The calendar

Phase 1: July 24 to August 7 (15 days) - build the universe

Goal: every persuadable and every soft supporter identified and touched once before early voting opens.

a. Door program launches immediately. Priority turf is priority_turf.csv: 113 precincts, 53,437 doors, PC DMA plus Wakulla. Top 25 precincts are 48% of it. This is the single highest-return activity available to you.

b. Bay is the anchor. 16,065 in-person voters, 25% of the entire district electorate. Precincts 20, 19, 17, 15, 18 are a third of the county's universe.

c. Wakulla is the sleeper. Precincts 2 and 7 are the #2 and #10 turf district-wide, at Panama City-level door density. Wakulla turns out at 40.8% and does not vote by mail. Nobody else is working it.

d. Endorsement rollout begins. One per week, each with a reason attached, front-loaded so they are absorbed before ballots are cast.

e. The contrast runs through the PAC, weighted to rural Tallahassee counties, where Gross's image is still intact and the purity and localism frames test strongest. He is already under water in Panama City (27/35); there is less to take there.

Phase 2: August 8 to 16 (9 days) - early voting

This is the election. Treat Aug 8 as election day and everything after as overtime.

a. Shift the entire door program to chase mode: contacted-and-supportive voters who have not yet voted.

b. Wakulla and Bay run parallel operations. They are different media markets but the same ground game.

c. Daily tracker read: is PC's share of the remaining vote converging toward 56.1% or drifting?

d. Rogers's public schedule should be visibly inside the district every day, and posted (Section 4).

Phase 3: August 17 to 18 - close

a. All remaining volunteer capacity to the top 25 precincts. b. Election Day is a turnout operation, not a persuasion operation. c. Polls close 7:00 PM CT / 8:00 PM ET. Note the district spans both zones - Bay and the western counties close an hour later than Leon.

3. Where the money goes

Ranked by evidence quality, best first.

Broadcast TV is the most expensive way to buy reach you can get on cable for roughly a third of the cost in this market.

4. Social media plan

Post as proof of work. The through-line is: Austin is here, and the other guy is not. That is the same contrast as the carpetbagger attack, delivered positively and in Rogers's own voice, which is where positive content belongs.

4a. Weekly cadence

4b. Rules

a. Vertical video by default. Static images only for endorsement cards and event promos. b. Location in every caption. "Knocked doors in Wewahitchka this morning" is the message. c. Never post a door photo without a person in it. An empty street is not proof of work. d. Shoot on a phone. Overproduced video reads as bought, and bought is Gross's problem, not yours. e. Do not run the attacks from Rogers's own accounts. The PAC carries the contrast; the candidate carries the biography. This is what keeps the 4.3:1 image ratio intact.

4c. What to show in the last ten days

Early voting locations, hours, and lines. "I just voted early in Panama City, here is where you can" is the highest-value post available Aug 8-15, because it converts your existing supporters rather than trying to make new ones.

5. What would change the plan

a. Trump endorsement. Ends the race. Nothing else in the model comes close. No action available beyond staying endorsable. b. Gross goes negative on Rogers. The model's single worst scenario, dropping Rogers to roughly 33%. A weak answer is not enough; it takes a PAC-scale response. Have it drafted now, not after. c. Gulf and Washington reporting. 994 mail ballots currently reporting zero returns in your stronger market. The tracker watches this nightly. d. PC share of remaining vote drifts above 56.1%. Means the in-person program has to make up more ground in less time. Watch it daily.

6. Sources

• Gerber, A. and Green, D. (2000). The effects of canvassing, telephone calls, and direct mail on voter turnout. APSR.
• Green, D., Gerber, A. and Nickerson, D. (2003). Getting out the vote in local elections: results from six door-to-door canvassing experiments. Journal of Politics 65(4).
• Kalla, J. and Broockman, D. (2017). The minimal persuasive effects of campaign contact in general elections: evidence from 49 field experiments. APSR 111(1).
• Lau, R., Sigelman, L. and Rovner, I. (2007). The effects of negative political campaigns: a meta-analytic reassessment. Journal of Politics 69(4).
• Boudreau, C., Elmendorf, C. and MacKenzie, S. (2015). Informing electorates via election law: an experimental study of partisan endorsements and nonpartisan voter guides in local elections. Election Law Journal.
• Research on local newspaper endorsement effects (~1.3 pt vote share), APSA Preprints, 2023.
• Boulianne, S. and Larsson, A. (2023). Engagement with candidate posts on Twitter, Instagram, and Facebook. New Media & Society.
• Cross-platform emotions and audience engagement in social media political campaigning. Political Communication (2022).

Campaign data sources: co/efficient poll July 14-16 2026 (n=624, MoE 3.91); Guidant March 2026; CD2 voter file (252,554 registered Republicans); Florida DOS daily VBM/EV file, election 49893; FEC Q2 filings.