James Steele

Seeing Expedition 33 clean up at The Game Awards fills me with joy 😊

So deserving... I genuinely feel fortunate to have been able to experience it.

Made a site comparing the sizes of living things :)

The great Julius Csotonyi spent 5 months painting over 60 illustrations for the site, no ai used

> neal.fun/size-of-life/

Ah, well tbf I am liking the hardware but have pretty much "google-fied" it and am continuing to use everything as I had it on my pixel.

I couldn't bring myself to use a Mac 😅

Was kinda forced by circumstance into switching to an iPhone after being a bit of an apple hater for almost 20 years... I hate to say it but, I kinda like the iPhone 17.

course schedule as a table. Available at the link in the post.

I'm teaching Statistical Rethinking again starting Jan 2026. This time with live lectures, divided into Beginner and Experienced sections. Will be a lot more work for me, but I hope much better for students.

I will record lectures & all will be found at this link: github.com/rmcelreath/s...

A gorgeous tabby cat (Kiki) laid on a tartan blanket atop a leather armchair. A blue cuddly toy rabbit sits behind her. She is looking at the camera.

🥰

Resting energy expenditure of women with and without polycystic ovary syndrome: a systematic review and meta-analysis Context: Polycystic ovary syndrome (PCOS) is common in reproductive-age women, who often have higher BMI classification. This is assumed to stem from lower resting energy expenditure (REE), influencin...

You can read the preprint here:

www.medrxiv.org/content/10.6...

🧵7/7

As we conclude:

"These findings indicate that REE does not meaningfully differ between women with and without PCOS. Group-level differences in resting energy expenditure are small, insignificant, or not physiologically relevant."

🧵6/7

A multi-panel figure summarizing mean resting energy expenditure (REE) estimates in people with and without polycystic ovary syndrome (PCOS). Left panel: A forest plot shows study-specific conditional estimates for REE. Each study is listed on the y-axis. For each study, blue points and intervals represent the control group and gold points and intervals represent the PCOS group. Most studies show slightly higher REE in PCOS, though with overlapping uncertainty intervals. Top-right panel: Two density plots show global grand mean REE estimates by condition. The control distribution (blue) peaks around 1442 kcal (95% quantile interval ~1334–1553), and the PCOS distribution (gold) peaks around 1472 kcal (95% quantile interval ~1359–1587). Bottom-right panel: A grey density plot shows the overall contrast between conditions (PCOS minus control). The distribution centers around a positive difference of about 30 kcal, with a 95% quantile interval spanning approximately –47 to 113 kcal, indicating compatibility with both small positive and small negative differences. A dashed vertical line marks zero difference. A legend indicates colors for Control (blue) and PCOS (gold). Text notes that point estimates and 95% quantile intervals are reported.

Mean REE differed between groups by 30 kcal/day [95% quantile interval: -47 to 113 kcal/day] and the contrast ratio for between person standard deviations was 0.98 [95% quantile interval: 0.71 to 1.33].

🧵5/7

Women with, and without, PCOS have very similar resting energy expenditure (REE) both on average and in terms of how much variance there is between individuals.

🧵4/7

Abstract text which reads: Context: Polycystic ovary syndrome (PCOS) is common in reproductive-age women, who often have higher BMI classification. This is assumed to stem from lower resting energy expenditure (REE), influencing lifestyle intervention guidelines. However, evidence for reduced REE in women with PCOS compared with those without is inconsistent. Objective: To systematically search and meta-analyse the existing literature to estimate and describe the difference in REE between women with and without PCOS. Data Sources: A systematic search was conducted using PubMed, Medline and Web of Science databases of published research from January 1990 to January 2025. Study Selection: Studies that measured REE in women living with PCOS, both with and without control arms of women without PCOS, were included. Data Extraction: Bibliometric, demographic, and REE data was extracted by one investigator and checked in triplicate. Data Synthesis: Thirteen studies were included in a Bayesian arm-based multiple condition comparison (i.e., network) type meta-analysis model with informative priors to compare both mean REE, and between person variation in REE, between women with and without PCOS. Mean REE differed between groups by 30 kcal/day [95% quantile interval: -47 to 113 kcal/day] and the contrast ratio for between person standard deviations was 0.98 [95% quantile interval: 0.71 to 1.33]. Conclusions: These findings indicate that REE does not meaningfully differ between women with and without PCOS. Group-level differences in resting energy expenditure are small, insignificant, or not physiologically relevant.

Almost a year to the day after pre-registering we've now pre-printed it revealing much the same conclusion albeit with a bit more fancy analyses (Bayesian arm-based multiple condition comparison meta-analysis with informative priors to compare both mean REE, and between person SDs in REE).

🧵3/7

Well, with help from the crack team of Richard Kirwan, Georgia Kohlhoff, Hannah E. Cabre, Alyssa Olenick, Leigh Peele, Greg Nuckols and myself we decided to set about turning this into a proper pre-registered systematic review and meta-analysis.

🧵2/7

No, PCOS Doesn't Lower BMR (Scientific Review) This article examines the roots of the claim that women with PCOS have a lower BMR and reviews the research on the subject.

So, folks might recall the great BMR series of articles from the MacroFactor team last year. One of the articles tackled the challenges women with polycystic ovary syndrome (PCOS) face with respect to managing their weight, and claims that this is in part due to "slower metabolism"

🧵1/7

I've been exposed to enough peer review that I think we can look at two classes of reviewers:

1. Constructive peer review
2. Adversarial peer review

I'm looking at this mostly from a psychology/methodology perspective (but wonder what other fields experience)

🧵 1/

Preserving musculoskeletal health through resistance training in individuals undergoing Glucagon-like Peptide-1 Receptor Agonist Therapy: a controlled interrupted time-series analysis (Stage 1 Registe... Glucagon-like peptide-1 receptor agonists (GLP-1-RAs) are increasingly prescribed for weight loss and cardiometabolic health but have been evidenced to lead to loss of lean soft tissue mass. Resistanc...

Just had our first set of reviews from @pci-regreports.bsky.social for our proposed study on RT during GLP-1-RA treatment.

Probably the most constructive and useful set of reviews I have ever had... and having them before starting the study makes them even better!

11/10 would recommend!

Oh nice, thanks 🙏🏻

Ah amazing, that was it! Thanks so much 🙏😊

I saw someone post a critical take on the consumer reports protein and lead piece a while back but I can't find it.

For some reason I vaguely recall it being @scientificdiscovery.dev but I can't find it on your feed.

Does anyone remember this?

I'd also say they don't spend enough time in psychometrics on fundamental ontological assumptions (outside of a few weirdos on the field) or conceptual analysis of what even the things they are trying to measure are proposed to be

+1 great book

‘A fragmented field: Construct and measure proliferation in psychology.’ (2025)

From ‘Language models accurately infer correlations between psychological items and scales from text alone.’ (2025)

From ‘Not within spitting distance: salivary immunoassays of estradiol have subpar validity for predicting cycle phase.’ (2023)

Work in progress with cycle tracking data from the app Clue

Want to make nice graphs with me, starting next year? I'm hiring for a position at the University of Witten/Herdecke.
uni-wh.softgarden.io/job/61280592...

This is my version of metascience by vibes 😎

Sometimes I think the best indicator of whether your field of research is in so called "crisis" is if, when you look around, you come away thinking "jfc... people are doing shitty work in this field" 😅

Fair point... 😅

Oh yea, nothing new in principle here at all... just the "oh wow, technology might make this easier now" albeit still vs "The System/Incentives" 😜

And we'll just be hosting everything via a completely open quarto site as you guys have with AMORE. We'll publish an initial paper introducing the resource moreso than anything and sign posting to the version controlled platform/website.

For our current project we have opted to have the initial analysis plan "red teamed" in so far as the analysis plans go as we have budget to do so. I suspect we may not keep it living though as we have 1000+ studies to include and I doubt more after that will make much difference 😅

JFC tickets for A Painted Symphony in London sold out in like 3 minutes!

Oh, forgot to ask... have you spoken with the team at @pci-regreports.bsky.social at all about this? I spoke with Zoltan about the idea of one of the living evidence syntheses I am working on being reviewed through PCI-RR but we weren't sure how well it would fit.

We just preprinted a huge meta-meta-analysis examining the effects of exercise on cognition, memory, and executive function

In short
- 2239 effect sizes
- extreme between-study heterogeneity
- extensive publication bias
- some subgroup/exercise-specific effects

More below (doi.org/10.31234/osf...)