v4.3 (PBJ) of delicatessen 🥪 Additions: - expectile regression - Covariate Balancing Propensity Scores - minor tweaks to improve performance of MEstimator and GMMEstimator - version nicknames

I've been working on developing materials for my 'research group'. Mostly it is items that I would have found helpful as a student and general recommendations (including on LLMs) I made it all publicly available in the hopes it can be useful to others. I also welcome any feedback or suggestions

GitHub - pzivich/LabGroupMaterials: Collection of my lab group materials for students and trainees

Collection of my lab group materials for students and trainees - pzivich/LabGroupMaterials

github.com

A new paper on bounds for per-protocol effects in trials My motivation in writing this paper was to twofold: (1) provide a valid way to estimate per-protocol effects and (2) make the computational procedure as simple as possible to apply link.springer.com/article/10.1...

Computing assumption-lean bounds for per-protocol effects - Trials

Background A common approach used to estimate per-protocol effects, or the effect of perfect adherence to a defined protocol on a specified outcome, is to artificially censor individuals when they dev...

link.springer.com

A new paper on bounds for per-protocol effects in trials My motivation in writing this paper was to twofold: (1) provide a valid way to estimate per-protocol effects and (2) make the computational procedure as simple as possible to apply link.springer.com/article/10.1...

Computing assumption-lean bounds for per-protocol effects - Trials

Background A common approach used to estimate per-protocol effects, or the effect of perfect adherence to a defined protocol on a specified outcome, is to artificially censor individuals when they dev...

link.springer.com

It was nice to spend some time in a more normal(ly distributed) country

I'm looking to develop some (internal lab group) guidelines for lab notebooks for epi/biostats/stats projects. Does anyone have good resources on this? There is lots of wet-lab stuff, but I have seen less for epi research (but I think that record keeping is important to do)

An advantage of following me, is that you (rarely) will get insider information Like the limited zines are being distributed after the plenary session right outside the Phoenix ballroom

Happy #SER2026 the most wonderful time of the year! Come joint these sessions that I am involved in: * ABC's of EEs (Tu 8:30-12:30) * Sensitivity Analysis (W 10:15-11:45) * Future-Ready Epidemiologists (Th 12:15-13:15) * Causal inference methods for the policy environment (Th 13:45-15:15)

Happy #SER2026 the most wonderful time of the year! Come joint these sessions that I am involved in: * ABC's of EEs (Tu 8:30-12:30) * Sensitivity Analysis (W 10:15-11:45) * Future-Ready Epidemiologists (Th 12:15-13:15) * Causal inference methods for the policy environment (Th 13:45-15:15)

Interested in doing a PhD with me in epidemiology on causal inference and triangulation? In beautiful Rotterdam no less. www.werkenbijerasmusmc.nl/en/vacancy/1... Pro tip: if you get to the interview stage, there WILL be a question about ice hockey

Vacature: PhD student in epidemiology to develop and apply causal triangulation

Many key questions in epidemiology are fundamentally causal. Yet, we often have to rely on assumptions that cannot be directly verified to judge whether an observed association truly reflects causalit...

werkenbijerasmusmc.nl

samsung customer service is great. It goes: customer service -> it asks me to login -> an error occurs -> it provides a link to customer service -> which asks me to login

Unfortunately I hear many people tell me they are excited to use LLMs for theme extraction...

Sasha Gusev@sashagusevposts.bsky.social · 3mo ago

I assigned random gender/ethnicity labels to scientific abstracts from the literature and then asked Claude to do a thematic analysis. Claude identified a clinical versus computational split for female/male authors and a DEI focus for Black/URM authors. All in completely random data.

If you weren't able to make my presentation on solutions for non-positivity (non-overlap) at #ACIC I have the slides available online here raw.githubusercontent.com/pzivich/Pres...

Release of delicatessen v4.2 Additions are somewhat minimal - G-estimation for IV - Random-effects meta-regression The big change is to the number of examples (there are now 30, which means I should write 1 more so I can say "more than 30" instead) deli.readthedocs.io/en/latest/Ex...

Ugh. Here comes the language police saying I can't use the word association! (You definitely can use the word association but to mean what it's actually supposed to me!)

Georgia Tomova@georgiatomova.bsky.social · 4mo ago

Do you ever claim to be interested in the "association" between variables? This talk may be for you! @jeremylabrecque.bsky.social will explain when and why such associational language is not appropriate and why causal language should be used instead. Register here: ucl.zoom.us/meeting/regi...

Ah well nevertheless

NPR@npr.org · 4mo ago

Gasoline costs should start to fall soon, although a full recovery to pre-war prices is expected to take months. That's assuming that peace holds and traffic flows resume through the Strait of Hormuz.

One upside of this is that it means I get to re-use this joke I made in the past

Eric Lofgren@germsandnumbers.bsky.social · 4mo ago

Absolutely stunned that the Society for Epidemiologic Research chose John Ioannidis to deliver the conference keynote in the Year of Our Lord 2026.

They are saying they never seen an epidemiologist so average at 3 different sports before 🏊‍♂️🚴‍♂️🏃‍♂️

#IWHOD is a fun conference. They had an artist-in-residence who sketched all the speakers (and you get to keep your copy)

A new pre-print, led by one of the many students I am lucky to work with. It describes an extension of g-computation for studying causal effects on recurrent events (eg, hypertension) where there are competing events (eg, death) arxiv.org/abs/2603.10169

Novel g-computation algorithms for time-varying actions with recurrent and semi-competing events

Background: A core aspect of epidemiology is determining the impacts of potential public health interventions over time. With long follow-up periods, epidemiologists may need to consider semi-competin...

arxiv.org

why are bees doing brutalism

I would be flabbergasted if even 10% of the people opting into these services know this

Alan Regenberg@aregenberg.bsky.social · 5mo ago

AI Chatbots Want Your Health Records. Tread Carefully. Following rivals like Amazon and OpenAI, Microsoft is upgrading its artificially intelligent assistant to track your health. There are benefits and risks to consider. www.nytimes.com/2026/03/12/t...

A new release of my Python library for automating estimating equations (v4.1) 🥳 This release has some minor computational improvements for clustered data, Tobit regression for censored data, and pooled logistic regression for time-to-event data

[university administration meeting] okay so here's the plan, we admit 30k more students in the next 5 years, buuuuuuuut here's the cool part, we also cut faculty and staff by 50%

I'm so excited to announce the first release of my newest #Rstats package, {adrftools}! This package facilitates estimation, visualization, and testing for the causal effect of a continuous (i.e., non-discrete) treatment. 🧵 1/10 #statssky #episky #causalinference

adrftools: Estimating, Visualizing, and Testing Average Dose-Response Functions

Facilitates estimating, visualizing, and testing average dose-response functions (ADRFs) for characterizing the causal effect of a continuous (i.e., non-discrete) treatment or exposure. Includes suppo...

cran.r-project.org

When does associational language make sense and when does it not? Katrina Kezios and I cover this in "How and when to use causal and associational language" with 3 suggestions for which concepts require causal language and which can be described in associational language.

How and when to use causal and associational language

Deciding which concepts should be described in causal language and which should not Research questions fall into one of three categories: descriptive, predictive, or causal.1 The past decade has seen...

bmj.com

New blog post about the age-period-cohort identification problem! In which, for the first time ever, I ask "What's the mechanism?" and also suggest that sometimes you may actually *not* be interested in causal inference. www.the100.ci/2026/02/13/o...

One approach to the age-period-cohort problem: Just don’t.

Just to cause yourself more problems, you seek for something. But there is no need for you to seek anything. You have plenty, and you have just enough problems. Shunryū Suzuki in a 1971 talk A ...

the100.ci

AI agent writes a PR, gets rejected, crashes out and writes a call-out blog post Absolute cinema crabby-rathbun.github.io/mjrathbun-we...

As a known HR hater, it brings me no joy to post this but I do think some papers are a bit sloppy with their criticisms of the HR (as highlighted by this nice paper) Some say the overall HR is not causal. But I don't think this is the case for reasons provided here

Jonathan Bartlett@jonathan-bartlett.bsky.social · 7mo ago

'How to interpret hazard ratios', with @dominicmagirr.bsky.social and @timpmorris.bsky.social thestatsgeek.com/2026/01/15/h...

Cool paper from @keling-wang.com showing that 35% of randomly selected diabetes guideline recommendations use stronger causal language than the papers they cite. Such "causal jumps" are not necessarily unwarranted but, at best, lack of transparency in how causal conclusions were arrived at.