Donald Szlosek
LightNews LightNews
dszlosek.bsky.social
Donald Szlosek
@dszlosek.bsky.social
1.1K followers 4.6K following 250 posts
Biostatistician @IDEXX formerly at harvardmed, @BIDMChealth, @nasa. Big data, clinical trials, and medical diagnostics. Mainer. Opinions are my own. he/him
Posts Replies Media Videos
Reposted by Donald Szlosek
f2harrell.bsky.social
One thing @statberry.bsky.social does a lot that we should all be doing more is showing investigators the results of simulating their trial several times. When you know the true treatment effect but you get wild estimates with non-large N it usually gets their attention.
December 1, 2025 at 12:54 PM
dszlosek.bsky.social
I didn't realize this until the other day but I usually keep matches and a candle on my coffee table (an old wooden chest) and it is essentially all the ingredients you need to estimate pi and solve Buffons Needle problem. I can't unsee it now. My poor future houseguests. #statssky #statistics #math
November 30, 2025 at 3:33 PM
Reposted by Donald Szlosek
datavisfriendly.bsky.social
#rstats
It is with profound sadness I heard that my long-time friend and colleague, John Fox passed away this week.
He was the author of {car}, {effects}, {Rcmdr}, ... and numerous influential books. I will miss him greatly.
www.john-fox.ca
John Fox: Books and Software
www.john-fox.ca
November 28, 2025 at 3:26 PM
dszlosek.bsky.social
Interesting move of a preprint server to ban papers on AI, apparently due to record numbers of submissions of which low quality content was generated using AI #academicsky #AI #statssky
socarxiv.bsky.social SocArXiv @socarxiv.bsky.social · 4d
1. Pausing new submissions about AI topics for 90 days. That is, papers about AI models, testing AI models, proposing AI models, theories about the future of AI, etc. We will make exceptions for papers that are already accepted for publication (or published) in peer-reviewed scholarly journals
/2
PAUSE sign
November 27, 2025 at 8:37 PM
dszlosek.bsky.social
This whole interaction. Is exactly why I am on this platform. Experts, discussing and debating. Sharing resources. 👏 👏 More of this please
jdwilko.bsky.social Jack Wilkinson @jdwilko.bsky.social · 5d
Why does every target trial emulation study have to include a bit that says "this shows how great target trial emulation is. RCTs are non-generalisable and expensive"? Imagine if every RCT said "this shows how great RCTs are. We did not have to rely on strong, untestable assumptions".
November 27, 2025 at 4:18 PM
dszlosek.bsky.social
Because @laurewynants.bsky.social and @benvancalster.bsky.social do no get on here much #DataScience #MachineLearning #MissingData #Imputation #Rstats #Python #AI journals.plos.org/plosone/arti...
November 23, 2025 at 2:08 PM
Reposted by Donald Szlosek
brendannyhan.bsky.social
One of the most upsetting articles I've read in a long time www.theargumentmag.com/p/when-grade...

UCSD report senate.ucsd.edu/media/740347...

We are failing a generation of kids.
When grades stop meaning anything
The UC San Diego math scandal is a warning
www.theargumentmag.com
November 19, 2025 at 2:05 AM
Reposted by Donald Szlosek
kenkoonwong.bsky.social
@f2harrell.bsky.social you were right. Coverage for TMLE xgboost grid search size 5 appeared better than 20 but still the coverage is disappointing. I’ve updated this examination for my learning. Thanks again for the guidance. www.kenkoonwong.com/blog/tmle/
November 22, 2025 at 11:52 PM
dszlosek.bsky.social
I was thinking about the efficiency of nonparametric tests and remembered this SO post about the Wilcox being 96% as efficient as T test, even in small samples. For those who have a copy, Lehman & Romano did an thorough job in <5 pages detailing situation. #statssky #statistics #rstats #AcademicSky
November 21, 2025 at 1:22 PM
Reposted by Donald Szlosek
hansonmark.bsky.social
We wrote the Strain on scientific publishing to highlight the problems of time & trust. With a fantastic group of co-authors, we present The Drain of Scientific Publishing:

a 🧵 1/n

Drain: arxiv.org/abs/2511.04820
Strain: direct.mit.edu/qss/article/...
Oligopoly: direct.mit.edu/qss/article/...
A table showing profit margins of major publishers. A snippet of text related to this table is below. 1. The four-fold drain 1.1 Money Currently, academic publishing is dominated by profit-oriented, multinational companies for whom scientific knowledge is a commodity to be sold back to the academic community who created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis, which collectively generated over US$7.1 billion in revenue from journal publishing in 2024 alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit margins have always been over 30% in the last five years, and for the largest publisher (Elsevier) always over 37%. Against many comparators, across many sectors, scientific publishing is one of the most consistently profitable industries (Table S1). These financial arrangements make a substantial difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor & Francis revenues were generated in North America, meaning that North American researchers were charged over US$2.27 billion by just two for-profit publishers. The Canadian research councils and the US National Science Foundation were allocated US$9.3 billion in that year. A figure detailing the drain on researcher time. 1. The four-fold drain 1.2 Time The number of papers published each year is growing faster than the scientific workforce, with the number of papers per researcher almost doubling between 1996 and 2022 (Figure 1A). This reflects the fact that publishers’ commercial desire to publish (sell) more material has aligned well with the competitive prestige culture in which publications help secure jobs, grants, promotions, and awards. To the extent that this growth is driven by a pressure for profit, rather than scholarly imperatives, it distorts the way researchers spend their time. The publishing system depends on unpaid reviewer labour, estimated to be over 130 million unpaid hours annually in 2020 alone (9). Researchers have complained about the demands of peer-review for decades, but the scale of the problem is now worse, with editors reporting widespread difficulties recruiting reviewers. The growth in publications involves not only the authors’ time, but that of academic editors and reviewers who are dealing with so many review demands. Even more seriously, the imperative to produce ever more articles reshapes the nature of scientific inquiry. Evidence across multiple fields shows that more papers result in ‘ossification’, not new ideas (10). It may seem paradoxical that more papers can slow progress until one considers how it affects researchers’ time. While rewards remain tied to volume, prestige, and impact of publications, researchers will be nudged away from riskier, local, interdisciplinary, and long-term work. The result is a treadmill of constant activity with limited progress whereas core scholarly practices – such as reading, reflecting and engaging with others’ contributions – is de-prioritized. What looks like productivity often masks intellectual exhaustion built on a demoralizing, narrowing scientific vision. A table of profit margins across industries. The section of text related to this table is below: 1. The four-fold drain 1.1 Money Currently, academic publishing is dominated by profit-oriented, multinational companies for whom scientific knowledge is a commodity to be sold back to the academic community who created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis, which collectively generated over US$7.1 billion in revenue from journal publishing in 2024 alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit margins have always been over 30% in the last five years, and for the largest publisher (Elsevier) always over 37%. Against many comparators, across many sectors, scientific publishing is one of the most consistently profitable industries (Table S1). These financial arrangements make a substantial difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor & Francis revenues were generated in North America, meaning that North American researchers were charged over US$2.27 billion by just two for-profit publishers. The Canadian research councils and the US National Science Foundation were allocated US$9.3 billion in that year. The costs of inaction are plain: wasted public funds, lost researcher time, compromised scientific integrity and eroded public trust. Today, the system rewards commercial publishers first, and science second. Without bold action from the funders we risk continuing to pour resources into a system that prioritizes profit over the advancement of scientific knowledge.
November 11, 2025 at 11:52 AM
dszlosek.bsky.social
I have been a reviewer on 5 papers this year for AI/ML validation and each time they used R or R Squared. Had to request a Bier Score, Emax, even slope and intercept would be more ideal. Would be nice if ICI was more popular too
grimalkina.bsky.social Cat Hicks @grimalkina.bsky.social · 13d
calling this AI is ridiculous
A screenshot from the study reads "we present the R-squared values of all our regression models"
November 18, 2025 at 6:02 PM
Reposted by Donald Szlosek
Survey Statistics: weights and MRP for voters
statmodeling.stat.columbia.edu/2025/11/11/s...
Survey Statistics: weights and MRP for voters | Statistical Modeling, Causal Inference, and Social Science
statmodeling.stat.columbia.edu
November 12, 2025 at 10:05 PM
dszlosek.bsky.social
It’s fascinating how cohorts of statisticians have parallel pockets of hyper-specific knowledge due to textbooks, like the collective familiarity with blue fiddler crab mating patterns, courtesy of Agresti. Palmer penguins, iris data, ect, any others I'm missing? #statssky #statistics #rstats
November 12, 2025 at 5:09 PM
Reposted by Donald Szlosek
srvanderplas.bsky.social
Well, it's official. #UNL Chancellor Bennett submitted his final proposal and it includes eliminating the #statistics department. budgetprocess.unl.edu/final-budget...

Guess it's time to go look for jobs. Anyone looking for a couple of very talented #datavis researchers?
Final Budget Reduction Plan | Budget Process | Nebraska
budgetprocess.unl.edu
November 10, 2025 at 10:09 PM
Reposted by Donald Szlosek
jburnmurdoch.ft.com
Quick thread on the BBC and the political and societal significance of recent developments:

One of the main reasons the UK has historically been so much less polarised than the US, is that Britain has a shared source of information, consumed and trusted by most people regardless of their politics.
November 10, 2025 at 1:43 PM
dszlosek.bsky.social
One thing I have noticed a lot in meta-analysis (thinking of those involving comparison of prognostic models) is that most focus on Embase and Medline, but no mention of Compendium (think IEEE) WoS or Scopus. Granted the sheer volume for data extraction would be astronomical.
November 8, 2025 at 7:21 PM
dszlosek.bsky.social
Does anyone else automictically reach for the PDF symbol when opening a new article on a journal site? It just seems like I can read the them faster. #AcademicSky
November 8, 2025 at 2:31 PM
dszlosek.bsky.social
It's a shame I didn't find this before Oct. 31st. And also a shame the term "Ghost-time bias" never caught on.
November 7, 2025 at 6:16 PM
Reposted by Donald Szlosek
gscollins.bsky.social
NEW PAPER: "Reporting guidelines for studies involving generative artificial intelligence applications: what do I use, and when?"

--> www.nature.com/articles/s41...

#HealthTech #ClinicalAI #MachineLearning #MedAI #AIinMedicine #TransparentAI #HealthInnovation #GenAI
November 7, 2025 at 12:44 PM
dszlosek.bsky.social
Probably the papers I share the most often are @f2harrell.bsky.social 1996 on how to effectively develop multivariate models and @benvancalster.bsky.social 2019 paper on Calibration. Both extremely well written and informative. What are you most shared stats paper? #statistics #statssky #academicsky
November 7, 2025 at 12:12 PM
dszlosek.bsky.social
I recently rederived the poisson pdf distribution from the binomial. Reminds me of @kareemcarr.bsky.social recent blog about memorization vs proof. No computer around and needed it for something. Certainly there are times where knowing something will be time saving then re-deriving it from scratch.
November 6, 2025 at 1:21 PM
Reposted by Donald Szlosek
alxndrmlk.bsky.social
Can Game Theory Help Us with Causal Discovery?

What's Your Guess?

In this week's issue of Causal Python Weekly, expect a solid mix of history, fresh discoveries, and good teaching:

1/

#CausalSky #EconSky #StatSky #MLSky #EpiSky
November 2, 2025 at 10:19 AM
Reposted by Donald Szlosek
danlewer.bsky.social
We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?
A "methods primer" article in the journal "BMJ Medicine", titled "Factors associated with: problems of using exploratory multivariable regression to identify causal risk factors"
October 27, 2025 at 5:39 PM
dszlosek.bsky.social
I have hiked and led hikes up Mt Washington about 30 times now. Fall is definitely one of the most deceptively dangerous times of year to hike it and you need to be prepared for overnight and winter weather!
faineg.bsky.social Faine Greenwood @faineg.bsky.social · Oct 29
People are so fascinatingly stupid about Mt. Washington.

“Oh, it’s not that tall,” they say, little knowing it is haunted by Weather Demons
wolfperson1.bsky.social Hamel 🏳️‍🌈 @wolfperson1.bsky.social · Oct 29
I’m sorry but what were over 20 people thinking:
October 29, 2025 at 2:28 AM
dszlosek.bsky.social
I went into the office recently and I realized that the depth of my answer to a question a generally related to my proximity to a whiteboard. Anyone else? #statistis #statssky
October 27, 2025 at 1:08 PM