zshahn.bsky.social
Causal inference, assistant professor at CUNY SPH. Hoping this will be an unhealthy addiction I can feel good about. All likes are endorsements, but maybe by my 3 year old who stole my phone.
16 posts
57 followers
213 following
Conversation Starter
comment in response to
post
(That article is also good ammunition to use against reviewers and editors)
comment in response to
post
Here's a more forcefully worded argument for this point: pmc.ncbi.nlm.nih.gov/articles/PMC...
comment in response to
post
Yes, bookmarks are a crutch for the weak
comment in response to
post
jackson and vanderweele: pmc.ncbi.nlm.nih.gov/articles/PMC...
comment in response to
post
There’s one more “kernel of truth”. Because both looks and smarts can lead to success, they’re somewhat negatively correlated specifically among prominent people due to collider bias
comment in response to
post
Trialists do seem to have a (non-)inferiority complex about this
comment in response to
post
Well I don’t want to put words in his mouth and get him trouble with the fda but that’s what I would do if an important experiment were “broken” based on my understanding of the arguments in what if. I think we all agree you should adjust for prognostic covariates
comment in response to
post
Maybe call it a secondary analysis but argue its the one people should really pay attention to?
comment in response to
post
I don’t think he’s implying you shouldn’t prespecify. But I guess he’s implying you should break your prespecification and adjust for large imbalances that might matter but you didn’t plan for. At least if your research question is more important than harm you might do to norms around prespec
comment in response to
post
I was about to write this but you put it better than I would have!
comment in response to
post
Yeah, Miguel uses the word “confounding” to describe chance imbalance in an individual study sometimes, which I also find odd and would prefer to reserve for systematic bias in expectation over repeated samples. But there’s not misunderstanding at the root of the word choice
comment in response to
post
I don't think he's saying there's a problem with randomization... The moral of the story is to both randomize and adjust to hopefully make large random imbalances less frequent/severe
comment in response to
post
This allows for “worst case” interference. Wonder if it’s possible to incorporate some type of assumptions on limits to how “bad” interference is to get some precision back?
comment in response to
post
g-estimation