Ragnar Research Partners https://xitter.cf/t.co/NXqgrxHcs3
Washington, DC
Joined October 2011
- Tweets2.7K
- Following974
- Followers311
- Likes8.9K
This is a great case study that more people should pay attention to. A lot of problems with public data right now and weighting to recall doesn't fix all of them, but it sure helps give a cleaner view of what's actually happening on the ground.
We decided to do two versions of this poll b/c of 2 problems in our sample that we wrote about:
1. LOTS of Rs unhappy with Trump are lying about their 2024 vote.
2. Rural Iowa Rs ignored our survey. We think this could be a Canary in the Cornfields for Iowa polling this year.
2
67
Mike gets it.
As a capitalist I think financial incentives tend to work. With a public poll you literally get what you pay for. That’s it; that’s the purpose of it. It’s almost academic or a game and the consequences of failure are negligible.
Internal polls are a different story. They are a tool by which time and money are allocated. If internal polls are garbage, finite resources are misallocated at the candidate, committee, and PAC levels. Therefore the incentives built up over decades are for internal polls to be good (or at least materially probative). Good and bad polls have very real consequences in the context of a campaign.
Because good private polls are the map on which a campaign is fought, there is also no reason to blast them out to the world. If *you* know how your race is actually shaping up, that’s information you keep from your opponent—and vice versa.
This means when you see self-serving private polls in public, they’re probably not intended to be good. Dems these days rely on small-dollar ActBlue donations from the wine-mom class. Self-serving private polls are very good at ginning that up.
In sum, the polls that matter seem to be the private internal polls which we don’t see. But we can see their effects in terms of spending decisions. That’s what to watch if you want to know what’s probably going on.
4
167
Never trust a poll you didn't pay for. Good Lord this is an insane story.
NEW: The company behind a recent poll that showed Karen Bass with a 12-point lead over Nithya Raman for LA mayor & showed Bass collecting most of Spencer Pratt's prior support has admitted their poll was FAKE and was intended as a "social experiment" to see how a poll could spread in the media ecosystem without verification.
The company, Median Strategies, also conducted a recent poll in Wisconsin that had socialist gubernatorial candidate Francesca Hong up by 20 points. Hong lost her primary.
It was all bogus.
Wild story via @Noah__Goldberg at LAT: latimes.com/california/story…
4
351
Team is locked and loaded. Some real veterans ready for whatever the cycle will bring.
One thing we learned at our all company meeting in DC last week: Ragnar likes to win! From pool or pinball to providing your campaign with accurate data and expert opinions, we're in it for the long haul. Let us know if you're up for a challenge below: 👇
ragnarresearch.com/political
1
118
So Mainers face a tough choice between the serial porta-potty masturbator with a nazi tattoo and the woman who agreed with Mitch McConnell that one time?
We live in amazing times.
Seriously. I get that Dems need to mobilize that bias, but if you use the math for propaganda, you get called out for propaganda.
Ann Selzer was not early.... Turnout last night was 23% and Senator De Witt received more votes in 2022 than total votes cast last night. Also Democrats had nearly 3 volunteers for every voter which is not sustainable for their resources (or lack thereof) in 2026.
1
2
336
This feels like the most obvious thing imaginable, but I’m really glad someone is publicly putting some clear data out there showing that the emperor has no clothes.
The results are in: our new analysis, in collaboration with @gelliottmorris, finds major errors when replacing human respondents with AI-generated “synthetic samples.” Over the next 5 days, we’ll share 5 of the most surprising findings.
Read the full report: report.verasight.io/syntheti…
1
3
112
100% endorsed. Get rid of scammers, leave legitimate research.
🧵Polls are an essential tool of American democracy, helping us understand what the public truly thinks and holding elected leaders to account.
But the upcoming release of @Apple iOS 26 threatens the ability to conduct high quality polls, especially at the state and local levels
It’s been a hot minute since an airline told me to get screwed quite like @AmericanAir just did. After cancelling my flight to DCA yesterday they rebooked me on a 5:00am today. Cancelled that one and told me to wait another day.
3 minutes later @SouthwestAir took care of me.
4
192
Anyone at #aapor25 who wants the real deal on text to web should come see @ChelseaGoodale at 9CT this morning. This is the good stuff: side by side comparisons of live dials and text to web in real election scenarios: what it gets right and what it gets wrong. Suite 9.
This is 100% true. I’ve seen some incredibly smart people get a bad idea stuck in their head and lead to disaster.
That’s what good survey design and data is for. Don’t have a hunch: know.
This quoted post is unavailable.
2
82
Got this in the mail tonight. We stopped ordering from them years ago because of this crap. @TheSpiceHouse is better and doesn’t preach at you while you try to season your food.
Torn between just throwing it out or using it as kindling.
1
61
