#modelbased
LightNews LightNews
easystats.github.io
🎉 Great news for #rstats users! If you love the native R graphics feel of #tinyplot AND you're a fan of the powerful #easystats #modelbased package, this is for you!

Thanks to @gmcd.bsky.social, we significantly enhanced the tinyplot integration.

🔗 Read more: easystats.github.io/modelbased/a...
Plotting estimated marginal means with tinyplot
easystats.github.io
December 12, 2025 at 7:22 AM
ketanjoshi.co
Deeply absurd. This Google PDF published on a blog (arxiv, not peer reviewed) claims an LLM is "PhD level" but in most cases the MAJORITY of reference URLs were invalid or inaccessible.

A PhD sitting down and just fabricating >50% of sources = career ending

arxiv.org/abs/2511.11597
CLINB: A Climate Intelligence Benchmark for Foundational Models Michelle Chen Huebscher1 , Katharine Mach2 , Aleksandar Stanić1 , Markus Leippold1,3, Ben Gaiarin1 , Zeke Hausfather4 , Elisa Rawat , Erich Fischer5 , Massimiliano Ciaramita1 , Joeri Rogelj6 , Christian Buck1 , Lierni Sestorain Saralegui1 and Reto Knutti5 1Google DeepMind, 2University of Miami, 3University of Zurich, 4Stripe, 5ETH Zurich, 6 Imperial College London Evaluating how Large Language Models (LLMs) handle complex, specialized knowledge remains a critical challenge. We address this through the lens of climate change by introducing CLINB, a benchmark that assesses models on open-ended, grounded, multimodal question answering tasks with clear requirements for knowledge quality and evidential support. CLINB relies on a dataset of real users’ questions and evaluation rubrics curated by leading climate scientists. We implement and validate a model-based evaluation process and evaluate several frontier models. Our findings reveal a critical dichotomy. Frontier models demonstrate remarkable knowledge synthesis capabilities, often exhibiting PhD-level understanding and presentation quality. They outperform “hybrid" answers curated by domain experts assisted by weaker models. However, this performance is countered by failures in grounding. The quality of evidence varies, with substantial hallucination rates for references and images. We argue that bridging this gap between knowledge synthesis and verifiable attribution is essential for the deployment of AI in scientific workflows and that reliable, interpretable benchmarks like CLINB are needed to progress towards building trustworthy AI systems. Total Reference URLs Generated claude-opus-4-1 claude-sonnet-4 gpt-5 hybrid gemini-2.5-pro gemini-2.5-flash o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Reference URL Status hybrid gemini-2.5-pro claude-opus-4-1 o3 gemini-2.5-flash claude-sonnet-4 0 200 400 600 800 1000 Count of URLs Total Image URLs Generated hybrid claude-opus-4-1 gemini-2.5-flash gemini-2.5-pro claude-sonnet-4 o3 0.0 0.2 0.4 0.6 0.8 1.0 Proportion Image URL Status Status OK INACCESSIBLE_CONTENT INVALID_URL ERROR Figure 3 | Number of reference (top), and image (bottom), URLs and their status. Ablations We perform several ablation studies with the autorater (Table 4). Notably, removing the question-specific rubrics from the prompt changes the results only in the bottom half, with the Hybrid answers overtaken by Gemini 2.5 Flash and Claude Sonnet 4. This suggests that the additional resolution provided by the rubrics applies primarily to the kind of responses used to develop the rubrics. Or, in other words, that rubrics are far from complete. Hence, it is important that rubrics adapt to new data as better models become availab A New Expert-Grounded Benchmark for Scientific AI We introduce CLINB, a benchmark for modelbased evaluation of frontier models on complex, multimodal scientific communication. Its core is a new dataset of real-world climate questions paired with data-driven, question-specific evaluation rubrics, curated and validated by leading climate scientists through a novel three-phase, human-in-the-loop process.2 PhD-Level Synthesis vs. Attribution Failures Frontier models demonstrate remarkable knowledge synthesis, often exhibiting a PhD-level understanding. However, this performance masks a critical inadequacy in grounding. We report substantial hallucination rates for references (10% to 25%) and even more failures for images (50% to 80% in certain settings), exposing a major gap between synthesis and verifiable attribution. Insights into Human-AI Collaboration Dynamics Autonomous frontier models surpass ’hybrid’ answers (curated by experts using weaker AI assistance), revealing the assisting model’s capability—not human oversight—as the primary bottleneck. Counter-intuitively, highly motivated non-specialists (our ’Advocates’) who deeply engage with AI tools can produce higher-quality answers than domain experts who engage less with AI during answer curation. A Validated Methodology for Scalable Oversight We validate a rigorous, rubric-based autorater. Ablation studies demonstrate that structured prompts and automated evidence-checking are essential for mitigating inherent LLM judge biases. This process is hampered by inaccessible sources (up to 50%). Furthermore, we identify evaluation challenges, including model familiarity bias in human raters and the limitations of rubrics to generalize across models.
November 24, 2025 at 7:36 PM
cranupdates.bsky.social
Updates on CRAN: aifeducation (1.0.2), archive (1.1.11), boiwsa (1.1.3), ClassComparison (3.3.5), correctR (0.3.1), diceR (3.0.0), EpiNow2 (1.7.0), grates (1.4.1), HVT (25.2.2), jskm (0.5.10), modelbased (0.9.0)
February 5, 2025 at 1:24 PM
easystats.github.io
Yes, see argument `estimate`. Using this argument, you should be able to easily reproduce results from emmeans and marginaleffects::avg_predictions. It's more a matter of naming things/wording, where the modelbased approach differs from emmeans or marginaleffects.
May 31, 2025 at 11:46 AM
herowearexo.bsky.social
https://www.cambridge.org/core/journals/wearable-technologies/article/comparing-the-risk-of-lowback-injury-using-modelbased-optimization-improved-technique-versus-exoskeleton-assistance/6381C5C9F9C605D181DCCA12734DCC11
Comparing the risk of low-back injury using model-based o...
Comparing the risk of low-back injury using model-based o...
www.cambridge.org
November 23, 2024 at 9:57 AM
simonsmart.bsky.social
And I think the solution in your case when the binary isn't yet available is to use install.packages("modelbased", type = "source")
May 2, 2025 at 5:10 PM
ereinbergs.bsky.social
Not a critical issue but... I can't seem to update from modelbased .10 to modelbased .11 in the easystats package? Any one have any ideas? Seems like maybe version .10 is still the CRAN version even tho easystats.github.io/modelbased/ says otherwise? #rstats @easystats.bsky.social
May 2, 2025 at 1:28 PM
cranupdates.bsky.social
Updates on CRAN: archetypes (2.2-0.2), BinMat (0.1.6), EpiReport (1.0.4), equateMultiple (1.1.2), growthPheno (3.1.12), metasens (1.5-3), modelbased (0.11.0), rextendr (0.4.0), RSQLite (2.3.10), text (1.5), tidytlg (0.1.6)
May 2, 2025 at 1:33 PM
strengejacke.de
(and by post-estimation toolbox/framework I mean all the stuff you can do with #rstats packages like marginaleffects, modelbased or emmeans)
September 23, 2025 at 7:29 AM
cranberriesfeed.bsky.social
CRAN updates: ANSM5 ebvcube frscore modelbased paradox YatchewTest #rstats
June 11, 2024 at 1:02 PM
notsofunny.bsky.social
It kicks ass. I recently learned about this package that pulls from both the marginal effects and emmeans packages: easystats.github.io/modelbased/
Estimation of Model-Based Predictions, Contrasts and Means
Implements a general interface for model-based estimations for a wide variety of models, used in the computation of marginal means, contrast analysis and predictions. For a list of supported models, s...
easystats.github.io
May 27, 2025 at 9:38 PM
easystats.github.io
We're sorry that we missed to post a function last week, but we're currently in the process of revising our {modelbased}📦 significantly! Look forward to great new features, leveraging the power of {marginaleffects} and {emmeans}, providing an absolutely simple and inuitive user interface! More soon!
easystats.github.io easystats @easystats.github.io · Jan 27
One function per week, this week with `data_seek()` from the {datawizard} package. This function helps you finding variables by their names, variable or value labels in data sets Labelled data is also supported. #easystats #rstats easystats.github.io/datawizard/r...
January 27, 2025 at 11:27 PM
cranupdates.bsky.social
Updates on CRAN: cophescan (1.4.1), insight (0.20.1), KnockoffHybrid (1.0.1), lares (5.2.8), leontief (0.3), modelbased (0.8.8), SchoolDataIT (0.1.2)
June 11, 2024 at 5:15 PM
easystats.github.io
A new version of {modelbased} just hit CRAN, including bug fixes and many new features. modelbased let's you easily compute marginal means, contrasts and pairwise comparisons, and marginal effects (slopes). Find a lot of examples and vignettes online at: easystats.github.io/modelbased/
Estimation of Model-Based Predictions, Contrasts and Means
Implements a general interface for model-based estimations for a wide variety of models, used in the computation of marginal means, contrast analysis and predictions. For a list of supported models, s...
easystats.github.io
March 10, 2025 at 8:35 PM
mzloteanu.bsky.social
#statstab #458 Causal inference for observational data using {modelbased}

Thoughts: IPW, g-computation, and more. Learning OS and ways to compute ATE for (more accurate, but still not great) inference.

#gcomputation #ipw #iptw #observational #inference

easystats.github.io/modelbased/a...
Case Study: Causal inference for observational data using modelbased
easystats.github.io
November 12, 2025 at 8:23 PM
strengejacke.de
Maybe this post is also useful: easystats.github.io/modelbased/a...
Uses the modelbased package, but this is more or less a convenient wrapper around marginaleffects and emmeans.
Mixed effects models
easystats.github.io
August 14, 2025 at 11:01 AM
razzcus.bsky.social
🚨 #TR#TRENDINGPOSTtyle-Based totally Checking out: The Hidden Accelerator of Protected and Scalable IoT Methods
https://razzc.sbs/style-based-totally-checking-out-the-hidden-accelerator-of-protected-and-scalable-iot-methods/
Accelerator Hidden IoT ModelBased Scalable Secure Systems testing #Vi#Viral
app-development-testing.jpg?ssl=1
December 2, 2025 at 12:12 PM
cranberriesfeed.bsky.social
CRAN updates: modelbased purrr sampling #rstats
July 10, 2025 at 6:02 PM
cranupdates.bsky.social
Updates on CRAN: bayesplot (1.11.1), brms.mmrm (0.1.0), fbst (2.2), geoR (1.9-4), modelbased (0.8.7)
February 15, 2024 at 9:20 AM
joss-openjournals.bsky.social
Just published in JOSS: 'modelbased: An R package to make the most out of your statistical models through marginal means, marginal effects, and model predictions' https://doi.org/10.21105/joss.07969
May 30, 2025 at 12:21 PM
strengejacke.de
(code in ALT text)
# variance components parameters::model_parameters(m, effects = "random") # BLUPs parameters::model_parameters(m, effects = "grouplevel") # fixef+ranef parameters::model_parameters(m, effects = "random_total") # same as modelbased::estimate_grouplevel(m) # dot plot of BLUPs parameters::model_parameters(m, effects = "grouplevel") |> plot() # dot plot of BLUPs modelbased::estimate_grouplevel(m) |> plot()
August 6, 2025 at 8:32 PM
easystats.github.io
Got a thing for social and health inequalities?
easystats.github.io/modelbased/a...

Or maybe you're into the nitty-gritty of intersectional analysis?
easystats.github.io/modelbased/a...
Case Study: Measuring and comparing absolute and relative inequalities in R
easystats.github.io
August 31, 2025 at 8:27 AM
cranberriesfeed.bsky.social
CRAN updates: archetypes EpiReport modelbased rextendr RSQLite #rstats
May 2, 2025 at 10:02 AM
cranupdates.bsky.social
Updates on CRAN: alqrfe (1.3), backbone (3.0.3), broman (0.92), collinear (3.0.0), fairGATE (0.1.1), ggdiceplot (1.0.1), ipw (1.2.2), mlt.docreg (1.1-12), mlts (2.0.0), modelbased (0.13.1), moonboot (2.0.1), mrap (1.0.1), MSCMT (1.4.1), pqrfe (1.3), SimEUCartelLaw (1.0.4), xoi (0.74)
December 8, 2025 at 5:22 PM
easystats.github.io
A new version of the #rstats #easystats {modelbased}📦just hit CRAN! {modelbased} helps you to easily compute marginal means, contrast analysis and model predictions. It's a wrapper around the great {marginaleffects}📦, providing an easy and intuitive syntax! easystats.github.io/modelbased/
Estimation of Model-Based Predictions, Contrasts and Means
Implements a general interface for model-based estimations for a wide variety of models, used in the computation of marginal means, contrast analysis and predictions. For a list of supported models, s...
easystats.github.io
February 6, 2025 at 12:01 AM