How to evaluate the statistical error of social. survey within the same city?

Question

Mathematics

There is a poll on social media. networks in one of the city's publics. The essence of the survey is to determine the version of the android device. The results obtained must be evaluated and the error calculated, but it is not clear to me how to do this and by what criterion to calculate the error. With a live poll, everything is clear: the sample is taken in different parts of the city and people dance from here. But here the sample was made in the public of a city with a population of 200,000 people with a public size of 4,000 and a sample of 200 people.

Answer 1

The error does not depend on the city size, it is determined solely by the sample size (1 sigma does not exceed 1/(2*sqrt(n)) ). So for 200 people write "error (2*sigma) does not exceed 7%".
But the problem will be with who enters the public and with what probability. After all, as you know, Internet surveys show that 100% of the population uses the Internet ... And if the survey is voluntary, then how likely is a representative of one option or another to want to participate in the survey: it is possible that the owners of the oldest and newest versions will brag about more willingly than the "middle peasants".

Answer 2

Stick to three percent. It's kind of like the gold standard. Ceteris paribus.