Lightnews — Scholar-powered news

Reposted by Anne Scheel

Rich Lucas @richlucas.bsky.social · 6d

We already know that lagged effects in CLPMs are likely to be upwardly biased, but just how easy is it to find significant effects? Way too easy. I tested CLPMS in 100 randomly selected pairs of correlated variables and found significant effects in 98 of them. New preprint: osf.io/preprints/ps...

OSF

osf.io

4 16 33

Anne Scheel @annemscheel.bsky.social · 8d

Congratulations Felix, that’s great news!!

1

Anne Scheel @annemscheel.bsky.social · 8d

Brilliant choice!

Felix Thoemmes @felixthoemmes.bsky.social · 9d

Excited to share that I’ll be the incoming Editor of AMPPS. My first priority is building a diverse team of Associate Editors and Editorial Board members. If you’re interested, DM me or add your name via this super simple survey.
cornell.ca1.qualtrics.com/jfe/form/SV_...
Please share!

Qualtrics Survey | Qualtrics Experience Management

The most powerful, simple and trusted way to gather experience data. Start your journey to experience management and try a free account today.

cornell.ca1.qualtrics.com

Reposted by Anne Scheel

Peder M Isager @isager.bsky.social · 12d

Extremely honored to recieve Oslo New University College's science award for 2025. ONH has been a fantastic base to conduct my research at for the past 4 years, and I have an amazing team of colleagues around me to thank for that. From the bottom of my heart, thank you all!

2 2 19

Reposted by Anne Scheel

Candice Morey @candicemorey.bsky.social · 12d

Call for collaborators! 🧵

The TL;DR: we seek collaborators on a #ManyLabs #RegisteredReport about what causes rapid forgetting.

In-principle accepted Stage 1: osf.io/ahjn5

Expressions of interest: cardiffunipsych.eu.qualtrics.com/jfe/form/SV_...

Further details in the 🧵:

1 16 7

Reposted by Anne Scheel

Jamie Cummins @jamiecummins.bsky.social · 20d

Can large language models stand in for human participants?
Many social scientists seem to think so, and are already using "silicon samples" in research.

One problem: depending on the analytic decisions made, you can basically get these samples to show any effect you want.

THREAD 🧵

The threat of analytic flexibility in using large language models to simulate human data: A call to attention

Social scientists are now using large language models to create "silicon samples" - synthetic datasets intended to stand in for human respondents, aimed at revolutionising human subjects research. How...

arxiv.org

12 150 330

Reposted by Anne Scheel

Jamie Cummins @jamiecummins.bsky.social · 21d

Waiting for my preprint to be accepted, so in the meantime a teaser: here's what happens when you try to estimate a between-scale correlation based on LLM-generated datasets of participants, while varying 4 different analytic decisions (blue is the true correlation from human data):

1 8 39

Reposted by Anne Scheel

Julia M. Rohrer @dingdingpeng.the100.ci · 21d

How is that not sci-fi

Prime editing is part of the CRISPR toolkit, but it adds new features that make it more precise and versatile.

Standard CRISPR-Cas9 works like molecular scissors: it cuts through both strands of DNA at a target site, and the cell’s own repair machinery fills in the gap. The process can be unpredictable.

Prime editing, by contrast, makes only a single-strand cut and brings along its own repair kit (which includes a modified Cas9 protein to find the target, a guide RNA to position it, and an attached reverse transcriptase enzyme that writes in the new, correct DNA sequence directly from a built-in template). This lets scientists change DNA bases exactly where needed, without relying on the cell to guess how to fix the break.

2 14 68

Anne Scheel @annemscheel.bsky.social · 23d

Probably not what you’re looking for, but Schotter & Trevino (2014) begins with a general intro that I’ve found useful in the past www.annualreviews.org/doi/10.1146/...

Belief Elicitation in the Laboratory | Annual Reviews

One constraint we face as economists is not being able to observe all the relevant variables required to test our theories or make policy prescriptions. Laboratory techniques allow us to convert many ...

www.annualreviews.org

1 2

Reposted by Anne Scheel

Mark Fabian @markfabian.bsky.social · 26d

My impression is that being constitutionally impervious to criticism, always assuming you're the first person to have a thought, and never meaningfully engaging with people who aren't your friends all increase the likelihood that you'll succeed as an academic. That explains some of this phenomenon.

1 4 9

Reposted by Anne Scheel

Saloni @scientificdiscovery.dev · 26d

Designers!

We're looking to hire a graphic designer and typesetter based in London to oversee and do the production of our upcoming print edition at Works in Progress magazine.

If you like this work below, you might be the person for us!

More detail: www.worksinprogress.news/p/were-hirin...

Cover 'On the necessity of gardening' Botany and Cultivation.

Book pages with floral drawings in the middle of two columns of text

Abstract illustration of horse with detail

3 12 31

Reposted by Anne Scheel

Celestial Mediation @celestialsem.bsky.social · 27d

1 3 14

Reposted by Anne Scheel

ODISSEI @odissei.bsky.social · 27d

Interested in using @cbsstatistiek.bsky.social microdata for your research?

💻 The ODISSEI Microdata Access Grant funds 6 projects per round and offers step-by-step support.

Join the Q&A on 23 Sept (15:00–16:00 CET) to learn more & ask questions: odissei-data.nl/event/microd...

Microdata Access Grant 2025-2 Q&A - ODISSEI – Open Data Infrastructure for Social Science and Economic Innovations

Together with CBS, ODISSEI hosts an Information Event and Q&A Session to clarify the questions regarding the programme.

odissei-data.nl

2 1

Reposted by Anne Scheel

Julia M. Rohrer @dingdingpeng.the100.ci · 27d

New paper out with @boryslaw.bsky.social 🥳 In which we sketch out how to rethink measurement invariance causally for applied researchers. And provide a causal definition of measurement invariance!

www.sciencedirect.com/science/arti...

Rethinking measurement invariance causally

Highlights:
It is preferable to work with a causal definition of measurement invariance
A violation of measurement invariance is a potentially substantively interesting observation
Standard tests for measurement invariance rely on strong assumptions
Group differences can be thought of as descriptive results

Conceptual graph illustration the central points of the manuscript. A group variable is potentiall connected to a construct of interest which affects items. Measurement invariance is violated if the group variable directly affects the items, for example by modifying the loadings from the construct to the items, or by directly affecting an item

To make this less abstract, consider a scenario where students take an exam, R, meant to capture some ability, T, and then are admitted to a program, V, depending on their exam results: R → V. This is sufficient to result in a violation of the statistical definition of measurement invariance. Exam results and admission are not independent given ability because exam results have a direct effect on admission. Even if we know somebody’s ability (e.g., we know it’s very high), learning about their admission status (e.g., they were not admitted) can tell us something about their exam result (e.g., it may have been worse than expected). According to the causal definition, this in itself does not constitute measurement bias, which seems a sensible conclusion here. After all, the scenario does not involve any reason to believe that the measurement process varied systematically by admission status. Admission happens after the exams took place, it cannot retroactively influence the measurement process (and, for example, lead to unfair treatment depending on admission status).

3 36 110

Anne Scheel @annemscheel.bsky.social · 28d

I think Noah van Dongen has taught this and so does/did Fred Hasselman. AFAIK neither of them is on here but I’d just send them an email (anyone teaching theory construction in psych will probably be delighted to hear that others are interested too)

2

Reposted by Anne Scheel

Jamie Cummins @jamiecummins.bsky.social · Sep 1

I was thrilled to be invited to contribute to a forthcoming special issue for The Psychologist magazine called "Psychology needs a ... revolution".

I wrote about AI and its use in psychology: particularly how we can learn lessons from the past to avoid repeating old mistakes.
@psychmag.bsky.social

Psychology needs… an AI revolution | BPS

Psychology is in the midst of an AI revolution. But it’s not the one it needs, argues Jamie Cummins.

www.bps.org.uk

8 18

Reposted by Anne Scheel

Saloni @scientificdiscovery.dev · Sep 3

NEW EPISODE of HARD DRUGS!

“5 hours is too long!” some of you said after our first episode. So our second episode is 20 minutes 🤭

@jacobtref.bsky.social and I explore the world of proteins: how proteins fold into complex shapes, why complexity matters, how crowded and dynamic a cell really is.

Proteins: Weird blobs that do important things

open.spotify.com

5 15 48

Anne Scheel @annemscheel.bsky.social · Aug 28

Maybe LLMs make the same mistake as psychology students and think they can best understand human behaviour through the psych literature

9

Reposted by Anne Scheel

Julia M. Rohrer @dingdingpeng.the100.ci · Aug 25

Ever stared at a table of regression coefficients & wondered what you're doing with your life?

Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social)
Preprint: doi.org/10.31234/osf...
Website: j-rohrer.github.io/marginal-psy...

Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities

Abstract
Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models).

Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve.

A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals).

Illustrated are
1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals
2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and
3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.

49 280 940

Reposted by Anne Scheel

Daniel Lakens @lakens.bsky.social · Aug 25

If you are preparing your bachelor statistics course and would like to add optional material for students to better understand statistics on a conceptual level (see topics in the screenshot) my free textbook provides a state of the art overview. lakens.github.io/statistical_...

4 68 210

Reposted by Anne Scheel

Dorothy Bishop @deevybee.bsky.social · Aug 24

Huge variability documented in how publishers respond when informed about a problematic body of work by a research group. www.jclinepi.com/article/S089...
#publishers #retractions

Bar chart showing % of articles retracted, with expression of concern, or no action from different publishers. 100% no action from Elsevier and Wolters Kluwer; 100% retraction from Taylor and Francis.

2 50 110

Reposted by Anne Scheel

Thom Volker @thomvolker.bsky.social · Aug 21

Been working on a tutorial on synthetic data for open science for @lmu-osc.bsky.social
A draft version is now up: lmu-osc.github.io/synthetic-da...

It covers model building, evaluating synthetic data utility with density ratio estimation, and disclosure risk.

Feedback is very welcome!

3 13 38

Reposted by Anne Scheel

Julia M. Rohrer @dingdingpeng.the100.ci · Aug 20

Conditional on somebody being a more senior scholar active in replication initiatives, they are less likely to be known for their own highly visible scientific achievements.

In other news, conditioning on the outcome reliably makes it very hard to understand how the world works.

Daniel Lakens @lakens.bsky.social · Aug 14

It can be so frustrating to read STS research. Here are Bartscherer and Reinhart osf.io/rbyt6_v1/ Look at their flawed logic: Replication is *used as a career strategy*. The evidence? People were not known for earlier research. The clear confound? ECR’s drove the replication movement! 1/x

2 4 33

Reposted by Anne Scheel

Daniel Lakens @lakens.bsky.social · Aug 20

The full program for the PMGS Meta Research Symposium 2025 is online: docs.google.com/document/d/1... If you are interested in causal inference, systematic review, hypothesis testing, and preregistration, join is October 17th in Eindhoven! Attendance is free!

Meta Research Symposium 2025 PMGS

PMGS Meta Research Symposium 2025 16-17 October 2025, TU/e Eindhoven Conference website: https://paulmeehlschool.github.io/workshops/ Program Day 1 - Pre-Symposium Mini-Workshop Time Activity…

docs.google.com

13 21

Reposted by Anne Scheel

Ian Hussey @ianhussey.mmmdata.io · Aug 18

{truffle} is an R package for teaching users to process data.

Semi-realistic psychological datasets with predetermined effects (via `truffles_` functions) are then hidden in common data processing headaches (via `dirt_` functions) for students to clean and analyze.

mmmdata.io/posts/2025/0...

3 34 120