DhRS

DhRS

@dhrsprof.bsky.social

Health outcomes researcher

32 followers240 following80 posts

DhRS
@dhrsprof.bsky.social

A question for #StatsSky. I use marginaleffects in R for both average marginal effect (and counterfactual pred) of a binary variable after logit glm. I am also curious what adjusted predictions are for two groups. I get large AME but adj pred (avg_predictions) are similar. 1

DhRS
@dhrsprof.bsky.social

I know how both are calculated (done manually too) and what they estimate. My question is: in what cases, would there be such a big discrepancy? For other vars, AME/counterfactual pred is opposite of adjusted pred. 2

DhRS
@dhrsprof.bsky.social

A question for #statsky, if I want to compare difference in a stat measure between two sub-groups (say, males vs females), do I bootstrap them separately (maintaining n in each group) or together (and separate them before the stat)? Different LLMs provide conflicting info

DhRS
@dhrsprof.bsky.social

Holy hell, this is so good. The comments highlighted some stuff I had overlooked and were very relevant too, unlike the generic ones one would get from ChatGPT/Claude. Surely, there is a human in the loop?

Ben Golub@bengolub.bsky.social

I've been working on a new tool, Refine, to make scholars more productive. If you're interested in being among the very first to try the beta, please read on. Refine leverages the best current AI models to draw your attention to potential errors and clarity issues in research paper drafts. 1/

DhRS
@dhrsprof.bsky.social

The transfer of one manuscript from one journal after rejection is increasingly becoming a clown show. It's not really a transfer if I have to reformat the manuscript, revise for new word count, and re-upload all documents.

DhRS
@dhrsprof.bsky.social

Conduct and submit to a journal a descriptive study in an area where not much is known. Clearly mention the descriptive study design in title/objective. Reviewer: This is a descriptive study and lacks rigor. What about [a bunch of confounders]? Editor: Several methodological concerns. Reject.

DhRS
@dhrsprof.bsky.social

Percent vs percentage points stikes again!

Aaron Rupar@atrupar.com

Trump makes a tariff announcement in Pennsylvania: "We are going to be imposing a 25% increase -- we're gonna bring it from 25% to 50%, the tariffs on steel into the US."

DhRS
@dhrsprof.bsky.social

As someone who may be affected by cuts in NIH indirects, I don't know how to feel about quotes like "we are halting development of the next game-changing treatment". Research is messy, time-consuming and with high noise-to-signal ratio. I wish people would see research as a normative good 1

DhRS
@dhrsprof.bsky.social

Sure, some would ultimately lead to new meds, optimization of existing meds. But most will "fail" too. Knowing something does not work is equally cool too. It's still expanding human knowledge. As someone who finds fulfillment in bittersweet process of research, it's a lose-lose situation to be in 2

DhRS
@dhrsprof.bsky.social

A question on ps matching for #StatsSky, #EpiSky folks. If I have a treatment (X) group where X is given at an outpatient visit. I would like a control group who had the similar visit but without X. But a patient can have multiple visits, hence multiple index time for being a control. In the past, 1

DhRS
@dhrsprof.bsky.social

I've randomly picked one visit, used covariates based on that date and done matching/wt. Fear it's inefficient. Or, could use all visits of control group as index time, and do matching where a control's multiple index times are used. But could lead to matching with replacement. Which is better? 2

DhRS
@dhrsprof.bsky.social

How come a paper titled "An R tutorial for application of XYZ method" has no R codes in it (not in the appendix; no links to github or any external repo)!!! Beggars belief

DhRS
@dhrsprof.bsky.social

Question for #econsky: so much methodological work has been done of late for DiD strategy. But, I don't see much work on comparative interrupted time series, which even allows for modeling underlying trend and differences in pre-trends. Doing some simulations and don't get why CITS doesn't get love

DhRS
@dhrsprof.bsky.social

Hi #AcademicSky, I have been asked to review SBIR grants. I have no experience reviewing such grants, so a question: as a reviewer, how differently does one approach SBIR grants compared to R01/R21?

You are up to date.