Loading…

PSA: The 2024 election poll I tracked had a 6 point swing that nobody caught | Debate the Data - Vivisector

PSA: The 2024 election poll I tracked had a 6 point swing that nobody caught

I compared two big polls from the same week in October, one from Pew and one from YouGov. Pew had candidate A up by 4 points, but YouGov showed candidate B ahead by 2. Turns out the Pew sample had way more older voters while YouGov balanced for age better. Has anyone else noticed how different poll weighting can flip a result?

4 comments

4 Comments

anthonynelson3mo ago

Yeah, age weighting is the silent killer nobody talks about.

abbyp613mo ago

Man, tell me about it. Once I actually looked into how they apply those weights based on who answers the phone it made so much sense why polls can be so different. I had the same realization a few years back when a local election poll here was way off, and I dug into their methodology. The thing that helped me was just picking one or two pollsters I trust and only really paying attention to their work instead of trying to compare all of them. It cuts down on the noise a lot and you start to see patterns in their weighting that actually make sense. Not saying I trust any of them completely now, but I feel way less lost when I read their reports.

hannahcraig3mo ago

Have you checked how the polls you normally follow handle age weighting? I used to think all major polls were basically the same, but this really opened my eyes to how much methodology matters.

val_ramirez3mo ago

All major polls were basically the same" - man, I used to think that too. I figured pollsters just call people and write down answers, not realizing they're back there doing algebra with age brackets and response rates. Now I just assume every new poll that comes out is secretly lying to me in a slightly different way.