• EvilBit@lemmy.world
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    Believe it or not, the threshold for a reliable confidence measure can be a lot lower than you think. And the difference between a valid sample size for 1,000,000 people versus 350,000,000 people is almost certainly zero. Once you are confident in a result concerning many thousands, that usually extends to effectively everyone.

    If you want 95% confidence with an interval of +/- 5% for every human being on the planet, your sample size only needs to be 385 participants. That gets way larger really quickly where you need to consider cohorts like gender, geography, age groups, income, and so on, but for a generalized read you need surprisingly few people. A higher confidence and smaller interval also increase the number significantly.