• forrgott@lemm.ee
    link
    fedilink
    arrow-up
    5
    arrow-down
    7
    ·
    edit-2
    5 hours ago

    Uh. 500 sample size. To estimate 19 million people’s thoughts?

    Ok. Sure. What’s a “representative sample” then???

    Edit: go read the source. The choice of wording alone gives away the fact that this is NOT a properly done analysis.

    • WoodScientist@sh.itjust.works
      link
      fedilink
      English
      arrow-up
      5
      arrow-down
      1
      ·
      5 hours ago

      Read a damn stats book. Jesus Wept! This could literally be a question on a sophomore-level undergrad stats class, and you would fail that question.

        • Zaktor@sopuli.xyz
          link
          fedilink
          English
          arrow-up
          2
          ·
          3 hours ago

          That was for the “Biden voters in swing states” poll, so a much smaller population than most polls. Most nationwide political polls of the general public were in the 1,000-1,500 range for a 2-3 point margin of error. Polls with larger sizes were likely to also get useful crosstabs.

          MOE for this poll was +/-4.9%, which is high, but not “this is meaningless trash, what even is statistics?!?”, especially when the headline numbers are just general sentiment rather than a head to head. In the worst case if “Gaza” was 24 and “the economy” was 29, it’s not a very large difference in the finding.