Hey @apnews.com, I can fix the headline for you: "US Carries Out Another Extra-Judicial Execution Killing 6 Unknown People, Trump Says"

It was bad enough when politicians were proposing bleach to treat COVID-19. But to actually change the label on acetaminophen?!? “Kennedy said that the F.D.A. would issue a note to doctors about the risk of acetaminophen during pregnancy and begin the process to make a safety change on the label.”

References for the above, since these papers are not well known, but should be: King et al (2017): doi.org/10.1111/ajps... Huling & Mak (2024): doi.org/10.1515/jci-... Cho et al (2013): doi.org/10.1111/stan...

My pain point, the RStan viewer doesn't work 😥😥😥

ChatGPT is far better at being my personal mixologist than writing R stats code. #statsky #episky

"Central Limit Theorem tells us that we only need to use Gaussian Models!" Um, Maybe? I guess if you have infinite data and/or don't care about inference at all. Those of us that want to efficiently detect differences with constrained data need to though... #statsky

Science is NOT self-correcting. The arc of the moral universe does NOT bend toward justice. The second Law of Thermodynamics states that an isolated system will only either increase randomness or stay the same, not become more orderly. It requires us to put in energy to see defined change.

#statsky #episky

Proud to be an American...?

This is quite correct in the most important sense. I'll tell you a story: over ten years ago, I built an automatic p-hacker (in Matlab, I'm old). It took some data, changed more or less every analysis parameter possible, used every outlier management schema, etc. and then subgroups. It could...

Should transgender athletes compete with cisgender athletes in sport? I don't know. There are more Opinions, Position Stands, & Reviews than Original Research. The Original Research out there, is HIGHLY confounded, full of garbage, and invalid statistical analyses.

FEEDBACK REQUESTED! I've been tasked with developing a general flow chart for C-suite stakeholders on the process for planned observational causal inference work. I want the figure to be as correct as possible, fit on a Powerpoint slide, and be written for a layperson. Thoughts #statsky #episky?

Basically true for every field. Unfortunately, some fields are worse about this than others.

Don't categorize continuous variables. #statsky

Weird how Stanford Professor and "world's most cited epidemiologist" John Ionnidis can crank out a paper every 3 days and write bmj blog posts about science Kardashians, but is mute on all of this.

A senior machine learning expert at @stanforduniversity.bsky.social once interrupted a lecture of mine saying that #MachineLearning does not need a minimum sample size. I almost didn't know what to say. Almost.

While I know Gordon Guyatt has a lot of citations, either the dude is using really bad AI or has limited vocabulary and a huge ego about his work. I couldn't help myself. #statsky #episky

Flexible distributed lag models for count data using mgcv @amstatnews.bsky.social www.tandfonline.com/doi/full/10....

Headline: "CONSUMER CONFIDENCE SURGES!" Actual data (see picture). Nice to know we're all embellishing findings and it's not unique to a field.

This is all VERY field specific. I'm seeing Pharma moving away from their traditional SAS and more towards R. There are aspects of Life Sciences using Python but I'd be shocked if R weren't the predominant language in Life Sciences in 15 years.

Just bought a smoker. All future posts will now be about smoked meats, not statistics or sport. Be prepared.

My favorite pizza restaurant now adds a 10% tariff surcharge.

Glam journals publish flashy findings. Subject-matter journals publish the real evidence. Does Jet Lag cause changes in team sport performance? Find out soon... #statsky #episky

Hey @noahgreifer.bsky.social does this paper of yours have a peer-reviewed version or is there one in progress? It's great and I just want to suggest the "final version" to people as appropriate. Thanks! #Statsky arxiv.org/abs/2106.10577

I know it's done for simplicity, but it's annoying most study designs are initially framed as a treated v untreated populations In medicine, whether it's a physical/cognitive intervention or drug there are almost always intervention levels. SUTVA violation if bivariate 🤷 #statsky #episky

Once you start using splines, "look at the coefficients" is a nearly useless heuristic. Same goes for Bayes MCMC. Basically, thank the statistical gods for the emmeans and marginaleffects packages. "In my day, we had to write our own code to get marginal effects and BLUPs!" *Shakes Cane* #stats

Has Pearl responded to the recent paper by Dawid and Senn? I scrolled through hit Xitter feed and didn't see anything but he tweets so often maybe it was buried. Any help #statsky or #episky? arxiv.org/abs/2301.11976

I was in a week-long seminar with Don Rubin and Bernie Black and both continually stressed that ANY causal inference method simply needed to be strong enough to convince others. That said, there are a number of Obs Causal Inference methods I don't find "convincing" and I also think that's OK.

I grow weary of the nonsense in my field...

Dear Reviewers, No, I don't know why this obnoxious "error" occurs when my manuscript renders into a PDF on the submission portal. Please just ignore it, I've tried for 2 hours to fix the issue. Thanks, Me

In statistics we'd call this a bullshit sub-group analysis. But since it agrees with my priors, I'll allow it 😁

There are two qualitatively different kinds of NP approaches: rank-based (e.g. Mann-Whitney/Kruskal-Wallis/Spearman corr), which relax distributional assumptions, and 'modern', which typically assume an arbitrarily expandable/infinite-dimensional family of curves/distributions/etc.

Athletic Trainers, Strength Coaches and Dieticians can now get FREE CEUs for reading how Real-World Evidence can work in sport! www.gssiweb.org/toolbox/cont...

"I don't want a Disney vacation of our history! I don't a whitewashed history, I don't want a homogenized history. Tell me the wretched truth about America, because that speaks to our greatness" -- 20 hours into his speech, Cory Booker is spitting absolute 🔥

"This will go down as one of the darkest days in modern scientific history." Story on the decimation of NIH by @maxkozlov.bsky.social www.nature.com/articles/d41...?

A salvo against Rubin + Pearl's counterfactual-based approach to causal inference, authored by Dawid and Senn: "We argue [Pearl's] approach is dangerously misguided and should not be used in practice." arxiv.org/abs/2301.11976 Don't know that I agree, but it's certainly provocative! #stats

Was thrilled to join the American Association of Sports Physical Therapy Podcast to discuss "Casual Stats" for practitioners! It's mostly discussing how to critically evaluate research without needing advanced statistical knowledge. Have a listen! #statsky www.youtube.com/watch?v=Ugx3...

Trump has cut the NIH budget by more than half in defiance of Congressionally mandated appropriations for the agency. No comparison historically, not even the Great Recession. This is clearly not legal and Congress needs to address it.

How many times does Microsoft Word need me to add 'estimand' to it's dictionary?

Most of the time, prediction models are bullshit. Thanks for coming to my machine learning TED talk.

Man, North Carolinians really like Causal Inference. #statsky #episky

The analytics say that going down by more than 20 points in a game substantially decreases your chances of winning. Go figure. @tarheels.bsky.social

Make it a Class-Action Lawsuit.

Nice summary of our recent paper on SMOTE w/ @alcarriero.bsky.social @benvancalster.bsky.social

DAGs are great for articulating belief systems. Here's my little blog on using them to articulate if "God matters" #statsky #episky www.linkedin.com/pulse/does-g...

I'm so excited to announce that I'll be teaching "Causal Inference in R Using MatchIt and WeightIt" with @stathorizons.bsky.social! In this 4-day workshop (Aug 12 - Aug 15), I'll explain the basics of causal inference, matching, and weighting, using my R packages {MatchIt} and {WeightIt}.