Marko Bachl

Pic of villa engler

Very nice location for our 4th DiMES workshop today with @audreyalejandro.bsky.social. Thanks to @jojukao.bsky.social for organizing and @dennmis.bsky.social for setting us up in Villa Engler @freieuniversitaet.bsky.social

www.polsoz.fu-berlin.de/en/soziologi...

✊

🤖 In einer Zeit, in der KI-basierte Systeme und Plattformen tief in unser tägliches Leben eingreifen – von Informationsverbreitung bis hin zur #Meinungsbildung –, stellt sich die dringende Frage:

So all students in the control group would get a 0 on the “minutes listened to the podcast” measure? And then Assignment -> Minutes listened -> learning outcome?
Or should the use a general measure like “time spent on course materials” instead?

Yes, I can genuinely understand the general intuition why one wants to omit non-compliers from the treatment group, because how would the treatment work if they did not take it?

But not to see the obvious selection bias in favor of the interested students and having a very clear COI, idk...

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: The audio-learning materials used in this study were developed by the authors in collaboration with Anywyse B.V. The senior author is partly employed by Anywyse B.V (1 day/week). However, employment played no role in the validity of the results. The study was pre-registered and code has been made publicly available. Data is available upon request.

I did not see the "declaration of competing interest" before. Now this seems to be problematic when you decide to use a data cleaning procedure that would likely bias your results in a predictable way, even if it is preregistered...

Screenshot from the paper: "4.4. Data analysis plan As pre-registered, we excluded participants in the audio-learning group who listened to less than 76.8 min (20%) of the total 384 min of audio materials. "

5.1. Data preprocessing From the 257 participants in the audio-learning condition, 137 participants were excluded from the analyses as they listened to less than 20% of the audio-learning materials. All analyses were performed on the remaining 273 participants.

This paper seems relevant, claiming (per title) "AI-assisted audio-learning improves academic achievement through motivation and reading engagement" (doi.org/10.1016/j.ca...). Yet, excluding half of the treatment for not listening to the podcast seems like a problem, no?

Great experiment on AI podcasts and learning by Nanda Jafarian & Anne-Wil Kramer. AI-generated podcasts led to better student achievement by enhancing motivation and engagement. They were especially helpful for students with diverse learning needs. The authors generated podcast scripts from the textbook based on ChatGPT and then used Anywyse to generate podcasts. What was key, in my opinion, was the pedagogy behind AI. The scripts used effective pedagogical approaches : - Activate what students already know - Give a simple overview of key points - Use relatable examples - Add questions to check understanding Great use of AI, grounded in solid pedagogy... Link in the first comment.

And the hype is sure to follow on the hype network

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: The audio-learning materials used in this study were developed by the authors in collaboration with Anywyse B.V. The senior author is partly employed by Anywyse B.V (1 day/week). However, employment played no role in the validity of the results. The study was pre-registered and code has been made publicly available. Data is available upon request.

I did not see the "declaration of competing interest" before. Now this seems to be problematic when you decide to use a data cleaning procedure that would likely bias your results in a predictable way, even if it is preregistered...

Finally, as pre-registered, participants in the experimental group who listened to less than 20% of the audio-learning materials were excluded from the analysis. This decision was made to evaluate the audio-learning modules’ efficacy among those who actively engaged, as including non-engaged participants would conflate the effects of the intervention with a lack of exposure. Additionally, including non-engaged participants would create an unfair comparison with the control group, as their lack of engagement would align them more closely with controls than active users. However, it is worth noting that individuals may have different learning preferences, and audio-based learning is not effective or appealing for everyone. The exclusion of approximately half of the participants in the experimental group suggests that the audio-learning modules may not have been sufficiently engaging for all students. Future studies should explore strategies to improve engagement, while also considering individual learning preferences and identifying the factors that influence why some students choose to engage and others do not.

I get the logic of the limitations statement ---how can someone be affected if they don't comply with the treatment---, but this seems like a typical example of how "Conditioning on Posttreatment Variables Can Ruin Your Experiment" (doi.org/10.1111/ajps...)

Screenshot from the paper: "4.4. Data analysis plan As pre-registered, we excluded participants in the audio-learning group who listened to less than 76.8 min (20%) of the total 384 min of audio materials. "

5.1. Data preprocessing From the 257 participants in the audio-learning condition, 137 participants were excluded from the analyses as they listened to less than 20% of the audio-learning materials. All analyses were performed on the remaining 273 participants.

This paper seems relevant, claiming (per title) "AI-assisted audio-learning improves academic achievement through motivation and reading engagement" (doi.org/10.1016/j.ca...). Yet, excluding half of the treatment for not listening to the podcast seems like a problem, no?

🙈

YouTube video by BHZ PABLO FÜR IMMER - KONZERT

<3

Short recommendation: Don't do it.

Longer: Don't do it unless the theory is not only plausible but supported by causal evidence from other studies. "Plausible" is too low a bar.

How willing are scientists to act as experts in the news media? In the context of the #COVID-19pandemic, @birteleonhardt.bsky.social, Daniel Nölleke & Folker Hanusch interviewed 24 Austrian #ScientificExperts about their perception of news media logics. www.doi.org/10.5771/2192...

Why do people search for individual politicians? Because they (don't) like them. In a new paper we find sympathy/antipathy to be important predictors of political information seeking. We also find searches for female politicians more often concerned with appearance & family. doi.org/10.1177/1461...

Hey, @malarija.bsky.social and I published a research. Click below to find out more.

EPSA is a for-profit???

UZH: LECTURER RESEARCH IN DIGITAL HEALTH COMMUNICATION (PERMANENT POSTDOCTORAL POSITION) AT IKMZ – DEPT. FOR COMMUNICATION & MEDIA RESEARCH The lecturer position is a permanent postdoctoral position (100%, paid according to cantonal salary scheme; 123K CHF p.a.) with optional promotion to senior lecturer Workplace is Zurich

Please spread the word about this fantastic job-opportunity at the UZH department for communication and media research @ikmz.bsky.social

✔️ permanent contract ♾️
✔️ 100% paid (123K CHF p.a.) 💰
✔️ best city 🏞️
✔️ fantastic colleagues 🤩
✔️ research with a purpose ⚕️

jobs.uzh.ch/offene-stell...

On a personal note: This is a project that started during the pandemic when we were both at home with kids, teaching a remote seminar together @hmtmh.bsky.social sky.social, and otherwise just tried to hold it together.

Thank you @malarija.bsky.social for always staying on top on this!

Hey, @malarija.bsky.social and I published a research. Click below to find out more.

YouTube video by BHZ PABLO FÜR IMMER - KONZERT

<3

🎉 Wir feiern 75 Jahre Hans-Bredow-Institut #HBI75!
... jetzt auch auf BlueSky! #firstpost
🎂 75 Jahre Medienforschung – das bedeutet 75 Jahre Forschung über medienvermittelte öffentliche Kommunikation, interdisziplinär & international vernetzt.
🙏 Wir sagen Danke und freuen uns auf alles, was kommt!

IfPuK and WI team group pic

Look at all these smart people from @ifpuk-berlin.bsky.social (@freieuniversitaet.bsky.social) and @weizenbauminstitut.bsky.social. #ica25

Yes, this is not easy to do while moderating a live discussion. I tested the q&a submission out of curiosity (with a sensible q), but the q never showed up. Only got a message that the q will be reviewed. Fine for me, but potentially frustrating for remote attendees who want to ask q.

I thought I was disappointed with the #ica25 @bsky.app experience. But then I took a quick look at LinkedIn and now I think it‘s not too bad here.

Does that scale measure how familiar a respondent is with the online privacy literature? Might be a relevant addition to the literature ;)

And while you are there to enjoy the light, also check out our (w/ @jojukao.bsky.social and Dalreen Ramos) presentation on media bias in tennis press conferences. Panel on „Perceptions of Gender and Sexuality Bias in Sports Media“ starts at 1:30 p.m. #ica25

And while you are there to enjoy the light, also check out our (w/ @jojukao.bsky.social and Dalreen Ramos) presentation on media bias in tennis press conferences. Panel on „Perceptions of Gender and Sexuality Bias in Sports Media“ starts at 1:30 p.m. #ica25

Pic of conference room with window

Amazing upgrade of the conference experience: Rooms with windows :) come to room agate b+C to see the light (and for presentations on sport communication) #ica25