Regarding the JAMA study, I'm having a bit of trouble understanding their appendix especially table 1 which they claim is their alternative subgroup analysis. Are they anywhere reporting what happens when they don't weight by all these potential confounders?
And do you take the remark on page 1 about the confounders being fixed to mean they precommitted to the full analysis beforehand? Because that really matters in a study that is using so many different confounders in a seemingly complex model with many choices especially when the difference in outcomes is 81% vs 88%. If they did precommit then the first question doesn't really matter much.
Regarding the JAMA study, I'm having a bit of trouble understanding their appendix especially table 1 which they claim is their alternative subgroup analysis. Are they anywhere reporting what happens when they don't weight by all these potential confounders?
And do you take the remark on page 1 about the confounders being fixed to mean they precommitted to the full analysis beforehand? Because that really matters in a study that is using so many different confounders in a seemingly complex model with many choices especially when the difference in outcomes is 81% vs 88%. If they did precommit then the first question doesn't really matter much.