The methodology is actually pretty good, though I don’t understand why, for a computerized survey, they only chose to survey 0.0015% of the population.
I feel like they could’ve easily surveyed more people.
It’s actually a relatively large sample size, enough to get an accurate survey, and increasing it further leads to diminishing returns in terms of improved accuracy of the result. You don’t need that many people to get an accurate survey. If you play with a calculator like this one you can see that this sample size for a US population of 340 million people gets you into the region of 99% confidence with a 1.8% margin of error.
You can check the methodology of the survey to see whether it has any issues:
https://www.pewresearch.org/politics/2025/02/07/views-of-trump-methodology/
Thank you!!
The methodology is actually pretty good, though I don’t understand why, for a computerized survey, they only chose to survey 0.0015% of the population.
I feel like they could’ve easily surveyed more people.
It’s actually a relatively large sample size, enough to get an accurate survey, and increasing it further leads to diminishing returns in terms of improved accuracy of the result. You don’t need that many people to get an accurate survey. If you play with a calculator like this one you can see that this sample size for a US population of 340 million people gets you into the region of 99% confidence with a 1.8% margin of error.
But is that sanple diverse enough to reflect the overall demographics of that total 340 million? Like accounting for geography, race, etc?
Hard to say, but their description of methodology indicates that they are trying and taking active steps to achieve that.
Fair enough! It’s been a very long time since I took stats haha