This blog post is a master class in (1) communicating a message so few articles (and analysts 😬) seem to fully understand/engage with and (2) the power of good data viz. Such a good and worthwhile read!

Kristoffer Magnusson@rpsychologist.com · last mo.

New interactive blog! "Why Adjusted Regression Coefficients Are Less Descriptive Than They Look" rpsychologist.com/descriptive-...

New interactive blog! "Why Adjusted Regression Coefficients Are Less Descriptive Than They Look" rpsychologist.com/descriptive-...

I've made an R package for Bayesian Rasch #psychometrics with brms models, easyRaschBayes (on CRAN), implementing simple functions to create figures and tables with model fit metrics, etc. Attaching figures from conditional item infit, item-restscore with GK gamma, and the log-likelihood criterion.

No one:

 Absolutely no one: Billion-dollar publisher in 2026: “We published the wrong version so our only option is to issue a retraction under your name”

We are not ready for the next generation of research misconduct. "Our proprietary AI robots independently recreate any open source project from scratch. The result? Legally distinct code with corporate-friendly licensing. No attribution. No copyleft." malus.sh

If only...

Methodologists' theoretical papers: "you MUST do X or your estimates are meaningless." Methodologists' empirical papers: [does not do X]

Love methods papers where the authors cover an estimator’s properties in painstaking detail, then add a footnote: “We assume X, Y, Z, which are unrealistic, but relaxing them is too complex” Thanks. I suppose it’s up to us applied researchers to deal with those gaps then?

Opened Bluesky for some USA vs Canada hockey banter. Everyone’s discussing R pipes. 😑

Yes, archive OCI (Open Container Initiative) images for each project. Then we can simply docker / podman pull & run, without installing and fixing broken system or R package deps. This is much more future-proof and convenient than any R-only solution

Matti Vuorre@matti.vuorre.com · 6mo ago

Don't get me started on targets/drake. Make it easy for others to use your code please.

Just added to PowerLMM.js v0.3: Interactive power contour plots! Visualize how statistical power changes across parameter combinations. powerlmmjs.rpsychologist.com?view=contour

Just added to PowerLMM.js v0.3: Interactive power contour plots! Visualize how statistical power changes across parameter combinations. powerlmmjs.rpsychologist.com?view=contour

Wasn't that happy with how performance crashed as time points grew, so I dusted off some linear algebra to optimize the calculations. Benchmark showed a ~1Mx speedup using new implementation (with 100 time points) 😅

Kristoffer Magnusson@rpsychologist.com · 9mo ago

New release of PowerLMM.js! Browser-based power analysis for longitudinal models with dropout. Now includes: - Power analysis summary report - Reproducible & shareable configs (URL/JSON) - Calculations validated against R - Hypothesis region visualization powerlmmjs.rpsychologist.com

🎉 @rpsychologist.com 's PowerLMM.js is the online statistics application of the year 2025 🎉 powerlmmjs.rpsychologist.com - Calculate power (etc) for multilevel models - Examine effects of dropout and other important parameters - Fast! (Instant results)

Who’s gonna pay @rpsychologist.com to make a mega tool converting all g*power analysis types to an interpretable web tool like this?

Kristoffer Magnusson@rpsychologist.com · 9mo ago

New release of PowerLMM.js! Browser-based power analysis for longitudinal models with dropout. Now includes: - Power analysis summary report - Reproducible & shareable configs (URL/JSON) - Calculations validated against R - Hypothesis region visualization powerlmmjs.rpsychologist.com

New release of PowerLMM.js! Browser-based power analysis for longitudinal models with dropout. Now includes: - Power analysis summary report - Reproducible & shareable configs (URL/JSON) - Calculations validated against R - Hypothesis region visualization powerlmmjs.rpsychologist.com

Turns out proportions aren't collapsible after all. Take that, OR haters. "Globally, approximately 3.3% of the population reports using AAS, with prevalence rates of 6.4% among men [2] and 4% among women [3]."

I watched this so you don’t have to. A couple of highlights: ”I sometimes say that 'screen use', when putting that into statistical models, it’s sort of like looking at 'car use' when trying to find out why injuries are happening from car accidents” 🇳🇴 www.youtube.com/live/XKo6oUY...

Digital media and its impact on children and adolescents – livestreamed Nordic and EU expert meeting

YouTube video by Folkhälsomyndigheten

youtube.com

Great talk by @jeremylabrecque.bsky.social "Before and after causal inference: recognizing causal questions and answering them when assumptions are violated" www.youtube.com/watch?v=9B-X...

2025 CAUSALab Methods Series with Jeremy Labrecque

YouTube video by CAUSALab at Harvard T.H. Chan

youtube.com

Not sure where to start with this one. Internet gaming addiction… is a risk factor for internet gaming disorder

Regarding the expert consensus on smartphone use: let’s not forget that the related “behavioral addiction” gaming disorder, is essentially a copy-paste job. Below is a visual comparison of the ICD-11 criteria for gaming and gambling disorder—essentially a substitution of terms.

A common sentiment is “I can’t share my code because it’s messy and there might be errors” But if it’s that untrustworthy, why are you publishing the results it generates?

Julia M. Rohrer@dingdingpeng.the100.ci · last yr.

Folks, always share your code. It doesn’t have to be perfect to be helpful. And if you feel that it’s still too messy or not sufficiently clean to be shared, you shouldn’t submit yet. After all, there could be mistakes in your mess.

I see this a lot in longitudinal analysis—people assume missingness can depend on random effects (like slopes). But those aren’t conditioned on, so that kind of missingness would be non-ignorable (MNAR).

Kristian Bernt Karlson@kbkarlson.bsky.social · last yr.

Amazed how some scholars still think and argue that a multilevel (or random effects) model 'control for' higher-level characteristics. I often wonder whether multilevel models did more harm than good 👀 #sociology

This new package for mediation is great! The user interface is *so intuitive* but the unsung hero of this software is the documentation:

Kristoffer Magnusson@rpsychologist.com · last yr.

🚨New R package! {easymediation}🚨 The *Simplest* and *Most Correct* Way to Do Causal Mediation Analysis Are you tired of explaining mediation analysis to your colleagues? Just send them this package. github.com/rpsychologis...

Despite the well-polished workflow of this package, I tried to analyze the “useless data set” that comes with it using {mediation}. Treatment and mediator were (strongly) confounded, so we need to pay close attention to the sensitivity analysis!

Kristoffer Magnusson@rpsychologist.com · last yr.

🚨New R package! {easymediation}🚨 The *Simplest* and *Most Correct* Way to Do Causal Mediation Analysis Are you tired of explaining mediation analysis to your colleagues? Just send them this package. github.com/rpsychologis...

To think that some scientists working in medicine will even brag about not understanding the methods used in the papers they "lead". Like it's beneath them. What a world.

This is great. It even handles binary and count outcomes appropriately!

Kristoffer Magnusson@rpsychologist.com · last yr.

🚨New R package! {easymediation}🚨 The *Simplest* and *Most Correct* Way to Do Causal Mediation Analysis Are you tired of explaining mediation analysis to your colleagues? Just send them this package. github.com/rpsychologis...

🚨New R package! {easymediation}🚨 The *Simplest* and *Most Correct* Way to Do Causal Mediation Analysis Are you tired of explaining mediation analysis to your colleagues? Just send them this package. github.com/rpsychologis...

Step 1: Say data are MNAR Step 2: Toss in random values Step 3: Call it conservative I’ll definitely adopt this model-free way of handling MNAR, what could go wrong?