Florian Keusch | Lightnews

Light up
your news

About Privacy Terms Help

Florian Keusch

Florian Keusch

@floriankeusch.bsky.social

520 followers 230 following 58 posts

Austrian in Mannheim; Prof. of Social Data Science & Methodology at U of Mannheim; survey methodologist; (mobile) web surveys; passive mobile data collection

Sociology 26%

Political science 26%

more

Posts Replies Media Videos

Florian Keusch @floriankeusch.bsky.social · 20h

If you work with individual-level data from Very Large Online Platforms or Search Engines (VLOPSEs), then you probably want to read - and hopefully sign - this letter!

DSA40 Data Access Collaboratory @dsa40collaboratory.bsky.social · 21h

We've published an Open Letter urging EU policymakers to reject the Digital Omnibus' proposed Amendment to Article 12(5) GDPR, which would undermine citizens' data access rights & the researchers who depend on them for #DataDonation research.
Read & sign: shorturl.at/e1mr4 - your signature matters.

the background depicts the proposed amendment, the foreground reads "Open Letter: The Amendment to Article 12, paragraph 5 GDPR in the Omnibus Proposal Undermines Evidence-Based Policymaking"

Reposted by Florian Keusch

DSA40 Data Access Collaboratory @dsa40collaboratory.bsky.social · 21h

We've published an Open Letter urging EU policymakers to reject the Digital Omnibus' proposed Amendment to Article 12(5) GDPR, which would undermine citizens' data access rights & the researchers who depend on them for #DataDonation research.
Read & sign: shorturl.at/e1mr4 - your signature matters.

the background depicts the proposed amendment, the foreground reads "Open Letter: The Amendment to Article 12, paragraph 5 GDPR in the Omnibus Proposal Undermines Evidence-Based Policymaking"

Reposted by Florian Keusch

Brady West @bradytwest.bsky.social · 4d

Heads up! If you are interested in pursuing graduate studies in survey and data science (read: all things quantitative social science) at the University of Michigan, please check out these upcoming info sessions!

Reposted by Florian Keusch

Joshua Claassen @jclaass.bsky.social · 5d

🚨 Last chance to sign up for the next #CS3Meeting 🚨

Zaza Zindel talks about Social media as a tool for survey recruitment on November 25, 03:15 to 04:00 PM (CET).

🗓️ Sign up until Monday: ww3.unipark.de/uc/cs3_meeti...

Reposted by Florian Keusch

Nils Weidmann @nilsweidmann.bsky.social · 5d

www.pnas.org/doi/10.1073/...

Florian Keusch @floriankeusch.bsky.social · 7d

Yes. Some of the cheap vendors are already struggling because they cannot compete with the new wave of even cheaper data (‘silicon samples’).

Florian Keusch @floriankeusch.bsky.social · 7d

These panels already exist:
amerispeak.norc.org (@norc.org)
uasdata.usc.edu/index.php (@dornsife.usc.edu)
ssrs.com/ssrs-solutio... (@ssrsresearch.bsky.social)
www.lissdata.nl (NL)
www.gesis.org/en/gesis-pan... (D @gesis.org)
and many more...

AmeriSpeak | NORC at the University of Chicago

amerispeak.norc.org

Reposted by Alexander Wuttke

Florian Keusch @floriankeusch.bsky.social · 7d

I really see a potential threat for studies that recruite samples via social media platforms (but these samples were wacky even before LLMs) or some river sampling. Not for well managed (nonprobability) online panels and certainly not for high-quality probability panels.

Reposted by Alexander Wuttke

Florian Keusch @floriankeusch.bsky.social · 7d

But how would this work in reality? Say I build multiple bots, then I would need to register for each of that bot in a panel. If a panel follows the ESOMAR guidelines then they do double-opt-in verification and might even require a bank account or at least a unique email.

Florian Keusch @floriankeusch.bsky.social · 7d

But there are a number of high quality probability based online panels where I think the influence of LLMs and AI-respondents will be very limited. These are not cheap for a reason..

Florian Keusch @floriankeusch.bsky.social · 7d

True for nonprobability online access panels that allow self-selection and river sampling-like approaches. Many of them had major quality issues even before LLMs.

Reposted by Florian Keusch

Graduate School of Economic and Social Sciences - GESS @gessunimannheim.bsky.social · 9d

📣 Apply Now: Be part of the GESS 📣

🎓 #ApplyNow to our fully funded doctoral programs beginning Fall 2026!

📊 #Business
📊 #Economics
📊 #SocialSciences

Application Deadlines:
📬 Early: 15 Jan 2026
📬 Final: 31 March 2026

🔗 Don’t miss your chance to apply via our Online Application Portal:

Online Application: Graduate School of Economic and Social Sciences

gess-application.uni-mannheim.de

Reposted by Justin H. Kirkland, Josh Pasek, Florian Keusch

Joy Wilke @joywilke.bsky.social · 12d

come work with us! @bluelabs.bsky.social is hiring a survey scientist. we're looking for someone who loves working with survey data and wants to solve tricky problems facing the polling industry. please share widely!

job-boards.greenhouse.io/bluelabsanal...

Survey Scientist

Remote or Washington D.C.

job-boards.greenhouse.io

Reposted by Florian Keusch

Joshua Claassen @jclaass.bsky.social · 13d

📢 New insights on #GenAI interviewing agents asking sensitive open questions compared to a text-based web survey.

Answers to male agent include more topics, but no evidence of social desirability.

👉 New #OpenAccess paper with @jkhoehne.bsky.social #cneuert in #IJMR.

🌐 doi.org/10.1177/1470...

Reposted by Luca Rossi, Paul Nightingale, Florian Keusch , and 2 more Luca Rossi, Paul Nightingale, Florian Keusch, Tom Louwerse, Roman Senninger

Johannes Breuer @johannesbreuer.com · 17d

"dplyr but make it bussin fr fr no cap"
hadley.github.io/genzplyr/

dplyr but make it bussin fr fr no cap

`genzplyr` is an alternative syntax for `dplyr` that replaces boring old function names with GenZ slang. Your data wrangling is about to hit different.

hadley.github.io

Reposted by Florian Keusch

Public Opinion Quarterly - POQ @poqjournal.bsky.social · 20d

Got a great paper on how AI is reshaping public opinion research? Submit it to POQ's special issue on Artificial Intelligence and the Future of Survey Research!

Papers are being reviewed and accepted on a rolling basis starting now – full details here: s3-us-west-2.amazonaws.com/clarivate-sc...

Reposted by Florian Keusch

Jan Karem Höhne @jkhoehne.bsky.social · 20d

Attention please‼️ MASS workshop meets up with WEB DATA OPP in Barcelona in 2026😎😎 Join @floriankeusch.bsky.social @peterlugtig.bsky.social #BellaStruminskaya #MelanieRevilla and me. Check out the call⬇️

massworkshop.org

Reposted by Florian Keusch

Graduate School of Economic and Social Sciences - GESS @gessunimannheim.bsky.social · 21d

📣 New Social Data Science doctoral track launching in 2026 at our Center for Doctoral Studies in Social and Behavioral Sciences (CDSS)!
Applications open mid-November 2025.
ℹ️ www.uni-mannheim.de/gess/apply/admission-requirements/for-cdss/
#PhD #DoctoralStudies #DataScience #SocialStudies

GESS expands doctoral programme: New track in social data science to start in 2026 |

www.uni-mannheim.de

Reposted by Florian Keusch

Nicole Schwitter @nschwitter.bsky.social · 21d

Can AI-generated images work as visual vignettes in survey experiments? I argue yes - and discuss how . in my new @sscratsage.bsky.social paper: doi.org/10.1177/0894...

Two AI generated images showing 1) a boy walking through a disorderly neighbourhood with graffiti and trash and 2) a boy walking through a very similar but cleaned up neighbourhood without trash ad graffiti.

Reposted by Florian Keusch, Shelley Boulianne

Deana A. Rohlinger @deanarohlinger.bsky.social · 24d

@sscratsage.bsky.social received it 1,000th submission today - November 1st. Last year, the journal broke its submission record at 1,020. I wonder what we will do this year?

Reposted by Alexander Wuttke, Johannes Breuer, Sebastian Stier , and 1 more Alexander Wuttke, Johannes Breuer, Sebastian Stier, Florian Keusch

Valerie Hase @valeriehase.bsky.social · 27d

Interested in understanding #bias in #CSS?

✨Here's our Special Issue editorial in Communication Methods and Measures:

"Critical, but constructive: defining, detecting, and addressing bias in Computational Social Science"

👉 doi.org/10.1080/1931... (with @bachl.bsky.social & Nathan TeBlunthuis)

Screenshot of Abstract describing the Special Issue on Bias in Computational Social Science in the Journal Communication Methods and Measures

Reposted by Brady T. West, Jill A. Jacobson, Florian Keusch

Joshua Claassen @jclaass.bsky.social · 28d

Ever wondered how to get better survey feedback from respondents? 🧐

👉 Check out our new @jssam.bsky.social paper together with @jkhoehne.bsky.social @jessicakuhlm.bsky.social testing different (1) visual designs and (2) answer formats of #FinalCommentQuestions.

🌐 academic.oup.com/jssam/advanc...

Asking for Feedback: Innovating Final Comment Questions in Self-Administered Web Surveys

Abstract. Web surveys frequently include so-called “final comment questions” (FCQs) to provide respondents the opportunity to express their experiences wit

academic.oup.com

Reposted by Florian Keusch

Doug Parry @dougaparry.bsky.social · Oct 26

Any experiences with data donation of #Spotify listening history?

I requested to download my full listening history and it said that it would take 30 days to produce the download.

2 hours later I received the link to the zip file with full JSON records.

Any other experiences with this?

Reposted by Florian Keusch

Kristen Olson @olsonkm.bsky.social · Oct 24

New article! About a year ago, I asked ChatGPT 3.5 to calculate the reading level of survey questions. It was totally wrong. And this paper was born! Trent Buskirk and I examine multiple ChatGPT and Claude models for evaluating question reading levels & inputs.

journals.sagepub.com/doi/10.1177/...

“ChatBot” is a Two Syllable Word...Or Is It?: Using Generative AI for Survey Question Readability Assessments - Kristen Olson, Trent D. Buskirk, 2025

Market and survey researchers aim to write survey questions so that the target population can understand them. A common recommendation for general population st...

journals.sagepub.com

Reposted by Florian Keusch

Raphael Nishimura @rnishimura.bsky.social · Oct 22

"Representative sample" is a flawed, ambiguous concept and it does not have a precise definition in the technical literature of survey sampling, so you should avoid using it at all costs.
/TheEnd

Reposted by Florian Keusch

Raphael Nishimura @rnishimura.bsky.social · Oct 22

I'm a sampling statistician and a lot of people ask me about "representative samples". Since there is quite some confusion about it, I'm putting here an entire thread with everything you need to know about "representative samples": 🧵 [1/n]

Reposted by Johannes Breuer, Florian Keusch

Joshua Claassen @jclaass.bsky.social · Oct 21

🚨 Upcoming #CS3Meeting 🚨

@wanlo.bsky.social talks about analytic flexibility in silicon samples on October 29, 3:15 to 4:00 PM CET).

Great opportunity to gain novel insights into how survey responses can be generated with #LLMs.

Sign up now: ww3.unipark.de/uc/cs3_meeti...

Reposted by Jacob Montgomery, Andrew K Przybylski, Florian Keusch

Ian Hussey @ianhussey.mmmdata.io · Oct 20

New hobby:

Remaking article abstracts as movie trailers to expose hype and fearmongering.

Reposted by Jochen Müller, Kai Arzheimer, Simon Otjes , and 3 more Jochen Müller, Kai Arzheimer, Simon Otjes, Sebastian Stier, Florian Keusch, Tom Louwerse

Marius Sältzer @msaeltzer.bsky.social · Oct 17

Job Alert! We are hiring two post-docs (full time, 4+ years) in our project SCEPTIC - Social, Computational and Ethical Premises of Trust and Informational Cohesion with @annanosthoff.bsky.social @guzoch.bsky.social and Prof. Andreas Peters (uol.de/informatik/s...)

Reposted by Florian Keusch

Jakob-Moritz Eberl @jamoeberl.bsky.social · Oct 15

Selbstselektion & Mehrfachteilnahme unterminieren Grundlagen seriöser Stichprobenlogik.
Eine nachträgliche Gewichtung kann diese systematischen Verzerrungen nicht kompensieren.
Dass der ORF eine solche Umfrage als inhaltl. Aufhänger nutzt, ist methodisch & journalistisch fragwürdig & enttäuschend.

Screenshot: https://www.orffragt.at/index-ende.php

Zur Umfrage

Die Online-Umfrage „ORF fragt“ fand von 17. September bis 8. Oktober 2025 statt. Als Teil der ORF-Dialog-Offensive waren alle Menschen in Österreich eingeladen, dabei mitzumachen und ihre Meinungen zu alltagsrelevanten Themen abzugeben. Insgesamt beteiligten sich rund 80.000 Personen – Basis für die Auswertung waren dann 77.684 Interviews mit Personen, die laut eigener Angabe in Österreich wohnen und mindestens 14 Jahre alt sind. Inhaltlich begleitet wurde die Umfrage vom Markt- und Meinungsforschungsinstitut INTEGRAL, das auch die Auswertung durchgeführt hat. Aufgrund der Anlage der Umfrage – jeder, jede Interessierte konnte sich online daran beteiligen – wurde keine Stichprobe nach strengen wissenschaftlichen Kriterien gezogen, weshalb die Ergebnisse nicht repräsentativ für die gesamte österreichische Bevölkerung sind. Die Ergebnisse wurden aber nach soziodemografischen Merkmalen (z. B. Alter, Region) gewichtet, um sie an die tatsächliche Verteilung in der österreichischen Bevölkerung anzupassen.