Henrik Singmann

Henrik Singmann

@singmann.bsky.social

Associate Professor at UCL Experimental Psychology; math psych & cognitive psychology; statistical and cognitive modelling in R; German migrant worker in UK

1.2K followers1.2K following227 posts

Pinned

Henrik Singmann
@singmann.bsky.social

Finally out in Psychological Review (psycnet.apa.org/doi/10.1037/...), our update to Signal Detection Theory. We show that contrary to the prevailing Gaussian assumption, evidence distributions in recognition memory are likely minimum extreme Gumbel!

Henrik Singmann@singmann.bsky.social

Honey, we fixed Signal Detection Theory (SDT)! In this preprint, Constantin Meyer-Grant, David Kellen, Sam Harding, and I critically evaluate the (unequal-variance) Gaussian SDT model in recognition memory and pursue the Gumbel-min model as a principled alternative: doi.org/10.31234/osf... 🧵

doi.orgExtreme-Value Signal Detection Theory for RecognitionMemory: The Parametric Road Not TakenSignal Detection Theory has long served as a cornerstone of psychological research, particularly in recognition memory. Yet its conventional application hinges almost exclusively on the Gaussian…

Henrik Singmann reposted

Dirk Eddelbuettel
@eddelbuettel.com

New CRAN Policy revision just dropped (per my cronjob posting only to @CRANPolicyWatch@fosstodon.org) and now I want to of course know which #Rstats package at CRAN was the offender 🤯

Henrik Singmann reposted

Lucas Castillo
@lcastillo.bsky.social

🎉 New paper at Behavior Research Methods!!! We present the samplr package for R, which helps use sampling algorithms and cognitive models for human data - show theoretical background + examples. Thanks to @yunxiao-li.bsky.social and @asanborn.bsky.social. Paper link.springer.com/article/10.3...

link.springer.comThe samplr package: A tool for modeling human cognition with sampling algorithms - Behavior Research MethodsMany models of cognition have sampling as a central component. Researchers have used them to explain human performance across a wide range of tasks involving perception, probability reasoning, and cho...

Henrik Singmann reposted

Dora Matzke
@doramatzke.bsky.social

I’m hiring a postdoc in my lab at the University of Amsterdam! Interested in mathematical psychology, evidence-accumulation models & Bayesian hierarchical modeling? Join us to develop new models of cognitive fluctuations across multiple timescales. Please share! werkenbij.uva.nl/en/vacancies...

werkenbij.uva.nlVacancy — Postdoctoral Position in Modeling Cognitive FluctuationsCurious about why cognitive abilities fluctuate over time? Interested in using cognitive models and Bayesian hierarchical approaches to tackle complex, real-world data? Join this ERC-funded project to...

Henrik Singmann reposted

London Judgement and Decision Making seminar series
@ljdm.bsky.social

One day to go!⏳ Our first seminar of the term is TOMORROW (14 Oct), with Professor Samantha Kleinberg @healthailab.org on the gap between trolley problem and everyday choices 🧩 📍313, 26 Bedford Way, or on Zoom 💻(full details: tinyurl.com/ucl-ljdm)

London Judgement and Decision Making seminar series@ljdm.bsky.social

We are excited to have our FIRST seminar next Wednesday (14 Oct, Leverhulme Lecture🎓)! 🤩 Professor Samantha Kleinberg (@healthailab.org) from Stevens Institute of Technology makes the case for studying real-world decisions🌍 📢All are welcome! (full details: tinyurl.com/ucl-ljdm)

Henrik Singmann
@singmann.bsky.social

New preprint: When should you use linear mixed models (LMMs), and when is repeated-measures (RM) ANOVA enough? LMMs offer greater flexibility. But does that make them a better default? My paper argues for statistical parsimony: use RM-ANOVA when both methods are appropriate. osf.io/preprints/ps...

osf.ioWhen to Use Mixed Models and When RM-ANOVA - An Argument for Statistical ParsimonyWhen should researchers use linear mixed-effects models (LMMs) versus repeated-measures analysis of variance (RM-ANOVA) for analysing repeated-measures data? I first outline the conceptual and statist...
View full thread
Henrik Singmann
@singmann.bsky.social

Parsimony also needs consideration within LMMs. Simply dropping random slopes can inflate Type I error rates. In the simulations, random-intercept-only models performed poorly, and simplifying models step by step did not always provide adequate protection.

Henrik Singmann
@singmann.bsky.social

New preprint: When should you use linear mixed models (LMMs), and when is repeated-measures (RM) ANOVA enough? LMMs offer greater flexibility. But does that make them a better default? My paper argues for statistical parsimony: use RM-ANOVA when both methods are appropriate. osf.io/preprints/ps...

osf.ioWhen to Use Mixed Models and When RM-ANOVA - An Argument for Statistical ParsimonyWhen should researchers use linear mixed-effects models (LMMs) versus repeated-measures analysis of variance (RM-ANOVA) for analysing repeated-measures data? I first outline the conceptual and statist...
View full thread
Henrik Singmann
@singmann.bsky.social

When do LMMs earn their keep? I explain key cases: missing participant-by-condition observations, partially within-participants factors, continuous within-participants covariates, dependencies involving both participants and stimuli, and research questions about variability itself.

Henrik Singmann
@singmann.bsky.social

The paper then presents two simulation studies. Where both methods were applicable, RM-ANOVA – with the Greenhouse–Geisser correction where needed – generally controlled Type I error rates better, with equal or greater power. Benefits were particularly apparent with sparse data.

Henrik Singmann
@singmann.bsky.social

New preprint: When should you use linear mixed models (LMMs), and when is repeated-measures (RM) ANOVA enough? LMMs offer greater flexibility. But does that make them a better default? My paper argues for statistical parsimony: use RM-ANOVA when both methods are appropriate. osf.io/preprints/ps...

osf.ioWhen to Use Mixed Models and When RM-ANOVA - An Argument for Statistical ParsimonyWhen should researchers use linear mixed-effects models (LMMs) versus repeated-measures analysis of variance (RM-ANOVA) for analysing repeated-measures data? I first outline the conceptual and statist...
Henrik Singmann
@singmann.bsky.social

The paper begins with a conceptual primer on how RM-ANOVA and LMMs handle repeated measures. RM-ANOVA works with aggregated difference scores. This simplifies estimation but discards information. LMMs use the original response values, but estimation requires numerical methods that can fail.

Henrik Singmann
@singmann.bsky.social

Just opened: UBEL PhD studentships (eg at UCL): ubel-dtp.ac.uk/esrc-student... Reach out if you are interested in working with me on computational modelling in decision-making, memory, etc. or quantitative psych. 1st deadline: 30 October Unfortunately, only few places for international students.

ubel-dtp.ac.ukESRC Studentships Main Competition - UBEL DTPESRC Studentships Main Competition The UBEL DTP’s fully funded ESRC studentships cover full payment of tuition fees per year and an annual maintenance stipend. What’s on offer UBEL DTP studentships co...

Henrik Singmann reposted

Cédric Scherer
@cedricscherer.com

An exciting extension to the #ggplot2 universe 🍾 I love how Alex has tied the new features to best #dataviz practices, and he even added a coaching option. I’m especially excited about the easy way to place the y-axis title at the top, plus even more shapes that are usable right out of the box 🤘

Minnesota Aquatic Invasive Species Research Center@maisrc.bsky.social

MAISRC’s Quantitative Ecologist Alex Bajcz has created ggplotplus, an accessibility component for ggplot2, to produce graphs for posters, presentations, papers, and exploration. Especially helpful for ggplot2 users with limited time, energy, or experience. github.com/MAISRC/ggplo...

Henrik Singmann
@singmann.bsky.social

Incredibly sad news. The OSF is synonymous with sharing of data, code, and materials in psychology. If a paper does not have an OSF link, the chance of it having data available is very slim. I fear without OSF file sharing we might go back to the times when data was generally not available.

Brian Nosek@briannosek.bsky.social

Over the next several months, we will phase out some popular OSF features. These changes do not affect current public content, which remains safe and accessible. We have prepared support material to help users adapt. I am very sorry for the disruption to your work. Please read this post for details.

cos.ioOSF Changes | Center for Open ScienceWe are preparing substantial changes to OSF that will reduce its functionality, focus OSF on its unique strengths, and move toward an integrated model with complementary services. As part of that shif...

Henrik Singmann reposted

Gavin Buckingham
@drgbuckingham.bsky.social

The Cognition Group in the Psychology dept at Exeter are at risk of redundancy because "basic cognitive psychology research does not align with the current priorities of funding bodies" If you think this is overtly wrong and misguided, please sign this open letter docs.google.com/forms/d/e/1F...

docs.google.comOpen Letter in Support of the Cognition Group, University of ExeterTo Executive members of the University of Exeter, including Vice Chancellor Prof. Lisa Roberts, Senior Vice-President and Provost Prof. Dan Charman, Pro Vice Chancellor and Executive Dean Prof. Sallie...

Henrik Singmann reposted

Ethan
@ethanrosenthal.com

It’s been 10 years since this post and nothing’s changed 😭 www.ethanrosenthal.com/2016/07/20/l...

ethanrosenthal.comI'm all about ML, but let's talk about OR | Ethan RosenthalYou’ve studied machine learning, you’re a dataframe master for massaging data, and you can easily pipe that data through a bunch of machine learning libraries. You go for a job interview at a SAAS com...
Eugene Vinitsky 🍒@eugenevinitsky.bsky.social

Tools from operations research solve so many problems and the fact they’re not super widely deployed really tells you that many bottlenecks are not technical

Henrik Singmann
@singmann.bsky.social

You love teaching psychology and London? Then we have the perfect position for you: Associate Professor (Teaching), Experimental Psychology, @uclpals.bsky.social. Your role: Develop a flexible portfolio of psychology undergraduate modules for Non-Psychology students www.jobs.ac.uk/job/DSB720/a...

jobs.ac.ukAssociate Professor (Teaching) at UCLApply for the Associate Professor (Teaching) role on jobs.ac.uk, the top job board for academic positions in higher education. View details and apply now.

Henrik Singmann reposted

Charles Margossian
@charlesm993.bsky.social

📘 With the release of our textbook "Bayesian Workflow" (avehtari.github.io/Bayesian-Wor...), I figured I'd also share the content of my graduate course on the topic at UBC. 🌎 charlesm93.github.io/stat547/ The course contains overlapping and complementary material, homeworks and reading.

avehtari.github.ioBayesian Workflow book: Website – Bayesian Workflow bookWebsite for the Bayesian Workflow book by Gelman, Vehtari, McElreath, et al. — case studies, code, and exercises in R and Stan.

Henrik Singmann reposted

Dr Michele V
@micheleveldsman.bsky.social

“I have done all of the things that one is supposed to do to earn a tenure-track position. And I have done approximately 85% of them by typing prompts into a large language model and then moderately editing the output.” 🤣🤣🤣 brilliant 👌🏽 open.substack.com/pub/inprepar...

open.substack.comOpinion: I Was Not Allowed To Type Prompts Into ChatGPT During My Chalk Talk And This Is DiscriminationBy Dr. Rachel Simmons, Postdoctoral Fellow, Stanford University

Henrik Singmann reposted

Frederik Aust
@frederikaust.com

Have you used Bayes factor to test point-null hypotheses in Hierarchical models and wondered: 1. Bridge sampling (BS) or Savage-Dickey density ratio (SDDR)? 2. If SDDR, which density estimation method? 3. How many MCMC samples? 4. What's the cost-benefit trade-off? This may be of interest to you.

Zurich Cognition Lab@cognition-zurich.bsky.social

How to best estimate Bayes factors for standard hypothesis tests? We compared bridge sampling and the Savage-Dickey density ratio. The latter does a good job in most constellations typical for experiments in cognitive psychology.

link.springer.comReliability, bias, and computational cost of estimating the Bayes factor using bridge sampling and the Savage–Dickey density ratio - Behavior Research MethodsBayes factors often require numerical estimation because closed-form solutions are unavailable. In six simulation studies, we explored the reliability, bias, and computational cost of two easy-to-use ...
Henrik Singmann
@singmann.bsky.social

Super exciting *3-year* postdoc opportunity at Queen Mary University in London. The goal is to use state of the art results from cognitive psychology to develop learning interventions that help individuals with ADHD. If you love basic research and want it to have a real impact, this is for you.

damnmemory@edamizrak.bsky.social

📣 Postdoctoral Researcher Opportunity – ADHD, Working Memory & Learning I am recruiting a postdoc to join my research group at Queen Mary, London on a newly funded MRC New Investigator Grant. ⏱️Deadline: 2 July 2026 🗓️ Start date: 1 October 2026 Apply here: qmul-jobs.tal.net/vx/mobile-0/...

qmul-jobs.tal.netQuick Check Needed

Henrik Singmann reposted

David Shiffman, Ph.D. 🦈
@whysharksmatter.bsky.social

Scientific journal editor here: If you're invited to peer review an article and you can't do it, please just click "no." That tells me "this person can't do it, I should invite someone else." Otherwise the system waits 3 weeks to tell me that you didn't respond and I should invite someone else.

Henrik Singmann
@singmann.bsky.social

Finally out in Psychological Review (psycnet.apa.org/doi/10.1037/...), our update to Signal Detection Theory. We show that contrary to the prevailing Gaussian assumption, evidence distributions in recognition memory are likely minimum extreme Gumbel!

Henrik Singmann@singmann.bsky.social

Honey, we fixed Signal Detection Theory (SDT)! In this preprint, Constantin Meyer-Grant, David Kellen, Sam Harding, and I critically evaluate the (unequal-variance) Gaussian SDT model in recognition memory and pursue the Gumbel-min model as a principled alternative: doi.org/10.31234/osf... 🧵

doi.orgExtreme-Value Signal Detection Theory for RecognitionMemory: The Parametric Road Not TakenSignal Detection Theory has long served as a cornerstone of psychological research, particularly in recognition memory. Yet its conventional application hinges almost exclusively on the Gaussian…
View full thread
Henrik Singmann
@singmann.bsky.social

There are two versions of the 2M-task. In the 2M-Max task participants have to select one item they are most certain is old (studied). In the 2M-Min task participants have to select one item they are most certain is new.

Henrik Singmann
@singmann.bsky.social

The Gumbel-min predicts that accuracy in the 2M-Min task is not affected by M. That is, having more new items to pick from does not make it easier to select a new item. This predictions is beautifully confirmed. Performance in the 2M-max task increases with M, an expected pattern from all SDT models

Henrik Singmann
@singmann.bsky.social

Finally out in Psychological Review (psycnet.apa.org/doi/10.1037/...), our update to Signal Detection Theory. We show that contrary to the prevailing Gaussian assumption, evidence distributions in recognition memory are likely minimum extreme Gumbel!

Henrik Singmann@singmann.bsky.social

Honey, we fixed Signal Detection Theory (SDT)! In this preprint, Constantin Meyer-Grant, David Kellen, Sam Harding, and I critically evaluate the (unequal-variance) Gaussian SDT model in recognition memory and pursue the Gumbel-min model as a principled alternative: doi.org/10.31234/osf... 🧵

doi.orgExtreme-Value Signal Detection Theory for RecognitionMemory: The Parametric Road Not TakenSignal Detection Theory has long served as a cornerstone of psychological research, particularly in recognition memory. Yet its conventional application hinges almost exclusively on the Gaussian…
Henrik Singmann
@singmann.bsky.social

The main evidence for the Gumbel-min model does not require any model fitting. Instead, the Gumbel-min model makes a unique prediction for the 2M-task, where participants always see M studied and M non-studied items (e.g., 1 studied and 1 non-studied; 2 studied and 2 non-studied, etc).