Sunday, November 1, 2026

if the polling can't actually make an accurate prediction, does that make the polling useless?

not exactly.

it makes cheap polling useless. unfortunately, almost all media polling today is done on the cheap. in the modern united states, you need large sample sizes to get accurate polling results. i haven't seen a single accurate poll so far this cycle. they publish polls with less than 1000 responses and 5% margins of error in races that are likely going to be determined by a point or two. these kinds of polls are in fact essentially useless, especially if they don't use sampling.

you can't just add up sample sizes, either. the math doesn't work like that, but it's what the aggregates try to hack their way through, because the media wants a magic number to sell to it's audience.

in theory, a poll with closer to 5000 respondents that uses good sampling could get a result with a 1-2% margin of error. if the spread is more than 4 points, that poll would be seen as predictive.

but the elections nowadays are so close (often 1-2% spreads) that the kind of polling that is actually done across the board has little to no predictive value and if it happens to get the right outcome landing in the centre of the projected range of outcomes, it's really just via dumb luck. you'd need to be getting closer to 10,000 respondents to get predictive polling results in these very close races.

conversely, if you do bad polling with small sample sizes, you can get results that suggest a race is more competitive than it actually is, and the media might do that on purpose to generate headlines for clickbait and/or manipulate outcomes. i suspect that's the case in kansas and maybe in texas.

that doesn't make the polling completely useless, though. if media outlets want to keep doing these low quality surveys on the cheap, they need to start to be more honest about the results and state the truth of it upfront - the results of the polling data are not sufficient to make a projection. the results of the election are consequently unclear. that is a valid survey result - we did the polling and we don't know what's going to happen.

that's where we are in both the house and the senate, despite what the media is trying to project.