Got it - thanks to all who replied!

#canihazpdf Löchner, J., Santangelo, P.S., Ansell, E.B. et al. Ambulatory assessment in mental health: expert consensus and recommendations. Nat. Mental Health 4, 926–935 (2026). https://doi.org/10.1038/s44220-026-00658-w

The standard random-effects model in meta-analysis assumes that the amount of heterogeneity (i.e., the variance in the true effects) is homoscedastic. This assumption may not be true. I derived and examined a variety of tests for heteroscedastic heterogeneity. Among these, a bootstrapped score test performs quite well. A preprint describing these methods can be found here: https://osf.io/cavhy

#MetaAnalysis #RStats

As another alternative to the standard forest plot for visualizing many effect size estimates in a meta-analysis, one can use 'orchard plots' (Nakagawa et al., 2021). Example code to create such figures can be found here: https://www.metafor-project.org/doku.php/plots:forest_plot_orchard_style

#MetaAnalysis #RStats

In large meta-analyses, standard forest plot can become excessively large. One solution is to group the estimates together within studies. I just added an example to the metafor website illustrating this possibility:

https://www.metafor-project.org/doku.php/plots:forest_plot_with_grouped_estimates

#MetaAnalysis #RStats

The updated data and code for my chapters can be found here: https://www.wvbauer.com/doku.php/publications#book_chapters (they will also be updated on the book website soon).

The third edition of the Open Handbook of Experience Sampling Methodology is out! The book includes 14 chapters that cover all aspects of designing, running, and analyzing the data from an ESM/EMA study. The book is freely available here: https://www.kuleuven.be/samenwerking/real/real-book

I do not know who the authors are. However, the journal does. In my opinion, such behavior should lead to a lifetime ban from ever submitting an article to the journal again. This should even apply to all journals from the same publisher. The publisher should also approach the university of the authors and demand disciplinary consequences. If these are not made, then the entire university should be banned. 3/3

When I realized what was going on, I was livid. I certainly see the usefulness of LLMs for certain tasks, but this is abhorrent. It wastes the time of editors and reviewers, who are already under severe time pressure. Basically, this breaks the entire system. 2/3

I recently reviewed a paper for a top journals in statistics that was written front to back with a LLM. To make things worse, the conclusions drawn from the simulation study did not match what the results actually showed and instead just conformed to what the authors intended to show (presumably based on their prompt). So the authors were either too lazy or incapable of reading the generated text to detect this inconsistency. 1/3

He Fedi, I'm entertaining the idea of taking on some freelance data analytics work to keep me active between permanent jobs, but I haven't done very much freelance work, and that which I have done was found via friends-of-friends.

Does anyone have a reliable source of need-to-know information about navigating the online freelance market these days? What I've managed to find via search has been... kind of giving off scam vibes.

#FediHelp #FediHire #Jobs #DataScience

An update on the CRAN Task View on MetaAnalysis https://cran.r-project.org/view=MetaAnalysis which is still slowly growing. It now has more than 200 packages covering a wide range of situations.

#metaanalysis #Rstats

I took the German or Autistic diagnostic and scored Neither. I am apparently the control group (and somewhat disappointed). https://german.millermanschool.com

@ianhussey Of course this can raise false positives (maybe in this example). To not annoy users, I would only issue this warning once per session. Also, the cutoff used can be adjusted via an argument. This still leaves the question what the default cutoff should be. There are already some papers around this idea: https://www.medrxiv.org/content/10.1101/2025.11.12.25340062v1 and tools based on this: https://tools.metapsy.org/reference/flageffectsizes Any comments on this general idea and the cutoff value to use? 2/2

Ian Hussey (@ianhussey) contacted me about the recurring problem of unusually large SMD values appearing in published meta-analyses. Very large SMDs can result from using SEs in place of SDs in their calculation, which is a mistake I have also seen in practice. The question is whether software like metafor can help to mitigate such problems.

We are discussing whether a check could be put into metafor to warn users when SMDs are unusually large. For example, like shown in the screenshot. 1/2

A new version (1.6-0) of the metadat package (https://wviechtb.github.io/metadat/) has been released on CRAN (https://cran.r-project.org/package=metadat). The package provides over 100 meta-analytic datasets, which are useful for teaching and testing purposes.

#RStats #MetaAnalysis

A new version (5.0-1) of the metafor package has been released on CRAN. It includes some smaller updates, including more ways to visualize prediction intervals / distributions. Further details here: https://www.metafor-project.org/doku.php/news:news #Rstats #MetaAnalysis

The best of both worlds:

`<=` <- .Primitive("<-")
x <= c(2,5,6)
mean(x)

#Rstats

Why has nobody ever told me that a snake fight is part of a PhD defense? https://www.mcsweeneys.net/articles/faq-the-snake-fight-portion-of-your-thesis-defense