Comment on Harvard study links GLP-1 weight-loss drugs to a 27-point jump in women's employment — bigger than a college degree gap

EntheoNaut@lemmy.ml ⁨1⁩ ⁨day⁩ ago

• reallykindasorta@slrpnk.net ⁨1⁩ ⁨day⁩ ago

  I don’t read this one as a total plant, but I don’t think the methodology great. Using a ucla cohort study that’s been in progress since 2024 (quarterly questionnaires) the author compares two groups.

  1. People who received treatment
  2. People who didn’t receive treatment but indicated that they would like to in future (control group)

  Clearly not an excellent control group for the kind of claims she wants to make.

  She puts effort into statistically ‘matching’ a bunch of parameters to try to make it a better control group. She also discusses this as a possible issue in the text and tries to give some statistical evidence but I don’t find it very convincing. Doubt it will ultimately get published.

  original

  • Clent@lemmy.dbzer0.com ⁨1⁩ ⁨day⁩ ago

    Clearly not an excellent control group for the kind of claims she wants to make.

    You’re claiming p-hacking without any proof or qualifications stating why this is so clear to you.

    The control group setup is to remove people who have no need for the drug.

    original

    • reallykindasorta@slrpnk.net ⁨1⁩ ⁨day⁩ ago

      I meant to imply the author was in good faith and not an industry plant like other white papers in the journal have been. I actually read the paper and don’t feel the need to prove my identity to you in a forum in order to have an opinion. I think it’s obvious and most people in adjacent fields would think so. The author acknowledges it as well which I mentioned. Also said nothing about p-hacking, I definitely don’t think that.

      This is an unreviewed white paper, they’re putting it out looking for feedback.

      original