cross-posted from: https://lemmy.ml/post/1491937

Using surveys, cognitive tests and brain imaging, researchers have identified a type of depression that affects about a quarter of patients. The goal is to diagnose and treat the condition more precisely.

  • Velonie@beehaw.org
    link
    fedilink
    arrow-up
    8
    ·
    1 year ago

    1000 is not small at all and with a properly designed study it’s more than enough for statistical significance

    • Leafeytea@beehaw.org
      link
      fedilink
      arrow-up
      0
      ·
      1 year ago

      I was thinking of the sample in relation to the population being studied. There are reportedly over 21 million people in the US diagnosed with depression. 1000 people is representative of less than 0.005% of that group.

      • Velonie@beehaw.org
        link
        fedilink
        arrow-up
        8
        ·
        1 year ago

        Yes, I’m aware. Sample sizes to get a 95% confidence interval are significantly smaller in most cases than the average person thinks. See https://en.wikipedia.org/wiki/Sample_size_determination and the section on required sample size for hypotheses tests. There’s even sample size calculators online you can find that’ll spit this number for a 95% CI out for you. Personally I think given that information, it’s pretty unlikely the author made that error in a peer reviewed paper where reviewers are certainly aware as well

        • Umbrias@beehaw.org
          link
          fedilink
          arrow-up
          3
          ·
          1 year ago

          Just wait until people hear about all the medical studies with sample size goals of twelve, considered an adequate sample size for medical procedure studies and medical device studies.

        • Leafeytea@beehaw.org
          link
          fedilink
          arrow-up
          2
          ·
          1 year ago

          Thanks. :) My curiosity about the other factors remains. Perhaps they will add more information about this next time.

          • Velonie@beehaw.org
            link
            fedilink
            arrow-up
            1
            ·
            1 year ago

            Yeah, that’s totally fair to question. I mentioned above, but it depends on sample selection and study design moreso than the actual number 👍