Dr Mircea Zloteanu 🍁 🍃
LightNews LightNews
banner
mzloteanu.bsky.social
Dr Mircea Zloteanu 🍁 🍃
@mzloteanu.bsky.social
3.1K followers 680 following 1.1K posts
Lecturer Psych & Crim @kingscollegelondon.bsky.social | Deception Detection; Emotions; JDM | Open Science; R; Bayes | @ukrepro ReproTea & StatsTea | #statstab | 🇷🇴 🇬🇧🌍

Stats blog: https://mzloteanu.substack.com/
Posts Replies Media Videos
mzloteanu.bsky.social
#statstab #456 Shall we count the living or the dead?

Thoughts: survival ratio -> if the intervention increases risk of the outcome
risk ratio -> if the intervention reduces risk of the outcome

#causalinference #riskratios #survivalanalysis #estimand

arxiv.org/abs/2106.063...
Shall we count the living or the dead?
In the 1958 paper "Shall we count the living or the dead?", Mindel C. Sheps proposed a principled solution to the familiar problem of asymmetry of the relative risk. We provide causal models to clarif...
arxiv.org
November 10, 2025 at 7:01 PM
mzloteanu.bsky.social
#statstab #455 {plotROC} Generate ROC Curve Charts for Print and Interactive Use

Thoughts: I dislike ROC curves, just check out the way you show uncertainty on them.

#roc #auc #roccurve #specificity #sensitivity #signaldetection #sdt
#ggplot #r

sachsmc.github.io/plotROC/arti...
Generate ROC Curve Charts for Print and Interactive Use
sachsmc.github.io
November 7, 2025 at 8:43 PM
mzloteanu.bsky.social
Maybe if statisticians started every sentence with "Yes, you are right. Let me do that for you"
statsepi.bsky.social Darren Dahly @statsepi.bsky.social · 3d
The advances we've made in statistics, experimental study design, and causal inference over the past century are remarkably useful for understanding our world. But there is never been a push to make people use them like we are seeing with generative AI. Perhaps take a moment to consider why.
November 7, 2025 at 9:36 AM
mzloteanu.bsky.social
#statstab #454 {Projoint} The One-Stop Conjoint Shop

Thoughts: Conjoint analyses will be the future for many (once they figure out what they are)

#conjointanalysis #rstats #r #groupanalysis

yhoriuchi.github.io/projoint/ind...
Conjoint Analysis with Reliability Correction and Visualization
Provides tools for analyzing data generated from conjoint survey experiments, a method widely used in the social sciences for studying multidimensional preferences. The package implements estimation o...
yhoriuchi.github.io
November 6, 2025 at 7:36 PM
mzloteanu.bsky.social
Can you spot a traitor? Unlikely. But can someone's demeanour affect our perception of them?

I contributed to this @BBC piece relating to the finally of #CelebrityTraitors

#TheTraitors #bbc #liedetection

www.bbc.co.uk/news/resourc...
The science of a trustworthy face - and how to spot a liar (or Traitor) - BBC News
Scientists suggest that trustworthiness is influenced by facial expressions - and levels of attractiveness.
www.bbc.co.uk
November 6, 2025 at 12:06 PM
mzloteanu.bsky.social
#statstab #453 {Bayes Power}
A General Application of Power and Sample Size Calculation for the Bayes Factors

Thoughts: Blending frequentist notions of power with bayes hypothesis testing.

#poweranalysis #bayesian #bayesfactor #errorrate #rstats #nhbt

doi.org/10.31234/osf...
OSF
doi.org
November 5, 2025 at 7:08 PM
mzloteanu.bsky.social
#statstab #452 Assumption Checking Prevents Assumption Hacking in Multiverse Meta-Analyses

Thoughts: Heterogeneity, publication bias, multiverse analysis, z-curve and more.

#zcurve #multiverse #heterogeneity #metaanalysis #bias #publicationbias

replicationindex.com/2025/06/13/a...
Assumption Checking Prevents Assumption Hacking in Multiverse Meta-Analyses - Replicability-Index
In a nutshell Statistical methods for meta-analyses make different assumptions. For this reason, different methods can produce different results with the same data. Meta-analysts often struggle to mak...
replicationindex.com
November 4, 2025 at 6:16 PM
mzloteanu.bsky.social
📄 New paper out: Behind Bars and Lies

Are good liars also good lie detectors? Not really, but it depends; for prisoners there was little evidence for a “deception-general” ability. For the public, we find moderate evidence for one.

See full paper:
onlinelibrary.wiley.com/doi/10.1002/...
Behind Bars and Lies: Dark Personality Traits, Lying, and Detecting Deception in the Prison Population Versus the General Community
We investigated the existence of a “deception-general” ability (association between the ability to produce and detect lies) in 140 individuals (80 from prison and 60 from the general community) using...
onlinelibrary.wiley.com
November 4, 2025 at 9:22 AM
mzloteanu.bsky.social
#statstab #451 Weighted Brier Score

Thoughts: Prediction is important for some research questions (e.g. clinical). But the standard one has some limitations.

#prediction #brierscore #auc #clinical #risk

arxiv.org/html/2408.01...
Weighted Brier Score - an Overall Summary Measure for Risk Prediction Models with Clinical Utility Consideration
arxiv.org
November 3, 2025 at 8:42 PM
mzloteanu.bsky.social
#statstab #450 Fitting GAMs with brms

Thoughts: Assuming linearity of your continuous predictors is not needed when you can add wiggles!

#gam #glmm #linearmodel #modelling #brms #rstats #bayes #tutorial #splines #r

fromthebottomoftheheap.net/2018/04/21/f...
Fitting GAMs with brms: part 1
Regular readers will know that I have a somewhat unhealthy relationship with GAMs and the mgcv package. I use these models all the time in my research but recently we’ve been hitting the limits of the...
fromthebottomoftheheap.net
October 31, 2025 at 5:20 PM
mzloteanu.bsky.social
#statstab #449 Significance tests, p-values, and falsificationism

Thoughts: A statistician and a philosopher debate p-values (not the setup to a joke). Good thread.

#pvalues #significance #nhst #fisher #greenland #epistemology #statistics

discourse.datamethods.org/t/significan...
Significance tests, p-values, and falsificationism
This thread takes its inspiration from the recent discussions in social science and statistics about significance tests, what they’re good for, whether p-values should be banned, and what all of that ...
discourse.datamethods.org
October 30, 2025 at 5:40 PM
mzloteanu.bsky.social
#statstab #448 {metaforest} Small sample meta-analysis

Thoughts: "a machine-learning based, exploratory approach to identify relevant moderators in meta-analysis"

#ML #MachineLearning #metaanalysis #smallsample #samplesize #heterogeneity #moderator

cjvanlissa.github.io/metaforest/a...
Introduction to metaforest
cjvanlissa.github.io
October 29, 2025 at 6:24 PM
Reposted by Dr Mircea Zloteanu 🍁 🍃
matti.vuorre.com
I am hiring PhD candidates to study the psychology of attention & technology use at @tilburg-university.bsky.social.

We're looking for motivated & curious scholars with expertise in cognitive psychology and statistics, and offer a friendly work environment with great terms & benefits.

tiu.nu/22989
An aerial photograph of the Tilburg University campus.
October 23, 2025 at 3:04 PM
Reposted by Dr Mircea Zloteanu 🍁 🍃
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
October 28, 2025 at 2:02 PM
mzloteanu.bsky.social
#statstab #447 Why do we use a one-tailed test F-test in analysis of variance (ANOVA)?

Thoughts: Directional tests are not the same things as one-tailed tests! Confusing at first.

#onetailed #directional #Ftest #ANOVA #chisquared

stats.stackexchange.com/questions/67...
Why do we use a one-tailed test F-test in analysis of variance (ANOVA)?
Can you give the reason for using a one tailed test in the analysis of variance test? Why do we use a one-tail test - the F-test - in ANOVA?
stats.stackexchange.com
October 28, 2025 at 7:05 PM
mzloteanu.bsky.social
#statstab #446 {causaldata} Packages of Example Data for The Effect

Thoughts: On your journey to learning Causal Inference you can use some nice datasets to figure out how horrible it can all go.

#causalinference #observational #python #r #DAG #OS

github.com/NickCH-K/cau...
GitHub - NickCH-K/causaldata: Packages of Example Data for The Effect
Packages of Example Data for The Effect. Contribute to NickCH-K/causaldata development by creating an account on GitHub.
github.com
October 27, 2025 at 5:33 PM
mzloteanu.bsky.social
After years of seeing people apply t-test to every data type, design, and question under the sun in psychology, yeah I do have a visceral reaction to it!
economeager.bsky.social Rafe Meager (they/them) @economeager.bsky.social · 17d
people are so reflexively mad about "parametric distributional assumptions" used for shrinkage and prediction im gonna start calling it "Gauss-adjusted squared loss shrinkage and prediction" instead
October 25, 2025 at 8:04 AM
mzloteanu.bsky.social
#statstab #445 What are credible priors and what are skeptical priors?

Thoughts: An excellent thread on prior elicitation by some of the big names in the field (frequentist and bayesian).

#priors #bayesian #bayes #likelihood #medicine #clinical #debate

discourse.datamethods.org/t/what-are-c...
What are credible priors and what are skeptical priors?
A few weeks ago Dan Scharfstein asked a group of colleagues about how to report an odds ratio of 1.70 with 95% confidence limits of 0.96 and 3.02. Back-calculating from these statistics gives a two-si...
discourse.datamethods.org
October 24, 2025 at 6:55 PM
Reposted by Dr Mircea Zloteanu 🍁 🍃
annemscheel.bsky.social
It also reminded me of the “stats maven” concept coined by @sanjaysrivastava.com some very long time ago, I believe here? thehardestscience.com/2014/12/04/s...
October 22, 2025 at 4:18 PM
mzloteanu.bsky.social
#samplesize and #ethics question: You plan for a study needing n=100 (50 per cell). Power analysis is all set up and pre-reg. But, because you do research in a uni, you are told you need to allow for more participants (students) as there is a set number of credits all need to reach. What do you do?
October 24, 2025 at 2:51 PM
mzloteanu.bsky.social
A: The study is OBS, so we must be cautious in our inference
B: But wait, we have thousands of data points!
A: Why didn't you say that from the start!? Then it's all good! We can conclude whatever!
B: INDEED, everyone knows when N is big randomization isn't needed!
* everyone cheers *
koenfucius.bsky.social koenfucius @koenfucius.bsky.social · 18d
The Big Data fallacy 😱

Research by Vosgerau et al suggests decision-makers tend to interpret correlational relationships as causal when sample sizes are large (vs small)—even when given experimental evidence showing no or opposite causal effect:

buff.ly/5ZjWJ8E
October 24, 2025 at 9:24 AM
mzloteanu.bsky.social
#statstab #444 {popower}: Power and Sample Size for Ordinal Response

Thoughts: Not the most intuitive but useful if you know the DGP will use ordinal data.

#ordinal #poweranalysis #power #effectsize #rstats #stats

rdrr.io/cran/Hmisc/m...
popower: Power and Sample Size for Ordinal Response in Hmisc: Harrell Miscellaneous
rdrr.io
October 23, 2025 at 7:34 PM
mzloteanu.bsky.social
🎉🍻🥳 Thank you all for subscribing. More posts to come this year. Subscribe now:
mzloteanu.substack.com

#substack #statstab #statistics #FiguringStuffOut
October 23, 2025 at 4:11 PM
mzloteanu.bsky.social
#statstab #443 Dienes Bayes factor calculator

Thoughts: Dienes presents a different way to compute BFs using the sample data. But, this can be seen as an acceptable double-dipping.

#bayesian #bayesfactor #evidence #likelihood #guide

bencepalfi.shinyapps.io/Dienes_BF_ca...
Dienes Bayes factor calculatorDienes Bayes factor calculator
bencepalfi.shinyapps.io
October 22, 2025 at 6:45 PM
Reposted by Dr Mircea Zloteanu 🍁 🍃
damiepak.bsky.social
Can you predict when Lou Bega's greatest hit, Mambo No. 5, was released based on the names of all the women mentioned in the song?

Weirdly yes.

yawpr.substack.com/p/project-a-...

#databs #rstats
October 21, 2025 at 8:24 PM