Kevin Collins

@kwcollins.bsky.social

Co-founder, Survey 160. Loves survey and voter participation. Researcher. Democrat. YIMBY. Husband and Dad. Opinion haver and measurer. Outdoors and cooking enthusiast. He/Him

PSA: Primary polling is prone to huge errors and always has been. Voters often tune in late, see several candidates as acceptable, and many change their minds up to the last minute. Turnout varies enormously, meaning weighting is prone to a lot of imprecision.

Sure maybe polling is over, or maybe don't rush to have too many takes about a Democratic race where almost all of the publicly available polling is either strategically released by one of the candidates or from Republican firms?

Bild

Two very interesting papers on how minorities respond to exclusion (having IDs not represented in info + categories presented to them). Somewhat different treatments and outcomes, and diff populations, but I think very parallel in spirit -- one account of mobilization, the other demobilization

BildBild

When the example questions produced by your "AI Research Agent" that you show as the example on your website are ... very poorly written ... it's a sign you need *more* researchers on your team rather than trying to replace them .

OUT TODAY! 🥳 A straightforward, non-technical guide to thinking like a quantitative social scientist. Written for students, journalists, and curious citizens. A lot of thought and heart went into this. I genuinely hope folks enjoy it!

BildBild

Unpopular view perhaps (but I think I’m an early enough adopter to have some credibility) but I think it would be better for society for people whose ideologies is different from mine to build their own communities here than it is for them to marinate in Musk’s algorithmically chosen content

Christopher Mims@mims.bsky.social · yesterday

Nothing more Bluesky than the CEO of the company being like “We still plan on fulfilling our founding mission“ and all the OG posters absolutely losing their minds

I picked 12 cups of figs today from a fig tree we planted just three years ago, and it's just the start of our harvest. A Chicago Hardy fig is a great investment in the DC climate, if you're thinking about it

My main point got buried in the middle of this thread, but @today.yougov.com if you're going to sell synthetic versions of panelists, let the panelists compare our answers to the predictions of our digital "twins". Use the data to improve your product, and if it works well, it's a marketing feature.

Kevin Collins@kwcollins.bsky.social · 2d ago

YouGov, Qualtrics, Ipsos and just about most other companies with an online panel asset, have created or at least experimented with synthetic replicas (because it's another way to monetize the data)

YouGov, Qualtrics, Ipsos and just about most other companies with an online panel asset, have created or at least experimented with synthetic replicas (because it's another way to monetize the data)

Someone please create a generative AI powered ombudsman for billionaires that instead of being sycophantic, asks you "is this really a good idea, here's some reasons you should reconsider"