A weekly digest and podcast for progressive political data scientists running GOTV and persuasion experiments. What moved voters this week, what the evidence says, and what it means for your program.
Delivered Monday mornings · No spam · Unsubscribe anytime
Each edition pulls from peer-reviewed journals, practitioner substacks, research org newsletters, and live polling data — summarized and discussed by two hosts who care about effect sizes as much as winning elections. Digest in the email, full podcast attached.
In creating this project, I collaborated with Claude to assist with research, summarization, drafting, and editing. The podcast hosts, Alex and Jordan, are not real people — they are AI emulations voiced with text-to-speech. I affirm that all AI-generated and co-created content underwent thorough review and evaluation. The final output accurately reflects my understanding, expertise, and intended meaning. While AI assistance was instrumental in the process, I maintain full responsibility for the content, its accuracy, and its presentation. Listeners are encouraged to fact-check any scientific or statistical claims in the digest and podcast against the original sources cited in each edition. This disclosure is made in the spirit of transparency and to acknowledge the role of AI in the creation process.
We're operating in a fog. This week's research landscape reveals a dangerous gap between what campaigns measure and what actually matters. While Trump's approval craters to 37% and economic pessimism deepens, the field still can't tell you which 8% of voters are actually persuadable — or whether your $45K digital spend moved a single vote. The data exists. The experiments don't.
We know Trump's approval is underwater. We know 74% think the economy is getting worse. We know 8% of adults are persuadable and 1.7% will decide races. We don't know what to do about any of it.
Randomized persuasion messaging test among "decided but volatile" voters. Randomly assign to:
Measure: actual turnout via voter file. Sample: 20,000+ per arm. Cost: ~$50K text-based.
Every edition pulls from practitioner newsletters, research orgs, academic journals, and private email newsletters — summarized by Claude, scripted by two hosts who care about effect sizes.