Appreciate the link. I looked for it but couldn't find it.
I didn't even realize they were able to establish a control group. That makes the "hurr durr everyone is fat and sick in the study" argument even stupider. Their control has the same characteristics as the test population. There is probably still a skew towards the unwell on the test group side. But we're not flying blind. They were able to control for ICU admits vs. non ICU admits, low-risk vs. high-risk, etc. You're talking a 720-830% increase in risk.
The weaknesses they mentioned are disclaimed in just about every study like this and at least one of their acknowledged issues would mean the study underestimated the impact.
"I looked through the data tables and there are several problems. I’m not going into them here."
Weak, dude. "This statistical study is totally wrong but I'm not going to tell you why but trust me it totally sucks". I don't see anything wrong with what they've presented. The first table just shows the test population vs. the general population. The rest are their results comparing the test population to a mirrored control group.
The only issue with this study is that it's not peer-reviewed and I assume that's coming. But the only thing peer-review would catch is gross error (IE someone fat-fingered something in Excel) or straight up fraud. Fraud is unlikely given that it's the government of the UK. I also want to see the detail behind how they built the control group. The way they explain it makes perfect sense, but I'd like to see the actual methodology behind building it (whether code or manual process).
All-in, it's a well done study and barring egregious incompetence or fraud, it should hold up to peer review.
Cliff's notes:
- Your link makes the "they're all fat and sick" argument even dumber
- It looks like a pretty good statistical analysis to me. I'd love to hear your alleged objections to it.