I just noticed ... exactly two years ago (give or take a few hours, perhaps) I passed 10K reputation.

Hi guys, I have a question that I think has been answered many times, but I can't find clear info about it. I have a poll, let's say a poll of the Swedish election. There are 8 parties to vote for. I conduct a survey, and I'd like to report some confidence intervals for this. How do I approach it? I'm thinking individual ranges for each party. Is this the way to go? And can I treat the eight different percentages as separate polls and calculate the CIs from that?

Don't know if using the same, say 1000, persons for calculating all the individual ranges make sense, or if I'd essentially would have to conduct eight polls, one yes/no poll for each party.

@amoeba thanks for the comment.

@victor sounds like a question for the main site.

@VictorSand, there is a question here: stats.stackexchange.com/q/74767/7290, but it doesn't have an answer.

Thanks @Glen_b and @gung