Check out our new article looking at who wanted to quit Twitter and who decided to wait and see after ownership of the platform changed hands.

Kudos to the amazing leading author @macaukfmak.bsky.social and wonderful collaborators @prowag.bsky.social & Sijia Yang. One of my favorite projects from Madison. Please check out two summary threads from Macau and Mike: 1) bsky.app/profile/maca... 2) bsky.app/profile/prow... DM for a free copy!

How did people react to Elon Musk’s Twitter takeover? Using panel survey data before/after the change, we (@macaukfmak.bsky.social @prowag.bsky.social Sijia Yang) find asymmetric perceptions and a lagging resistance: users intend to quit, but haven’t reduced use. www.tandfonline.com/eprint/RYKKH...

How did Twitter users react right after Elon Musk bought the social media site? @macaukfmak.bsky.social @zeningduan.bsky.social Sijia Yang and I used a panel survey (wave 1 before Musk took control of Twitter, wave 2 after) to find out. 1/ urldefense.com/v3/__https:/...

Using panel survey data collected before and after Elon Musk’s takeover, we ( @prowag.bsky.social @zeningduan.bsky.social Sijia Yang) find evidence of lagging resistance: users who view the takeover negatively intend to quit Twitter but do not (yet) reduce their actual use. 1/ shorturl.at/x3syg

Many thanks to Christopher Assaf, Bin Chen, @rossdahlke.bsky.social, @zeningduan.bsky.social, @meredithpruden.bsky.social ky.social, & @jiyoun-suk.bsky.social for sharing their materials! Also, @heesoojang.bsky.social has a great Twitter thread last year about the job market: x.com/HeesooJang2/...

💓I am so (so, so) very excited to be joining @ddc-sdu.bsky.social and @d-ias.bsky.social, starting February 2026, where I will continue my work on computational methods/AI, researcher data access, multi-platform language flows, and political extremism. 💓 1/

Brilliant scholar @shugars.bsky.social is our discussant. They highlighted some key trends, noting that many panelists raised similar question, including "when and how is it appropriate to use LLMs?" They also mention inequality challenges with using closed-source and expensive models.

We round the panel out with work by @damiantrilling.net on LLM agents. He and his team conduct 3 studies, some of which are fully simulated and some that are more hybrid. #ica25 @icacm.bsky.social

@zeningduan.bsky.social presenting his work with the Social Media & Democracy research group at @uwsjmc.bsky.social. In the study, they use an LLM to annotate accounts for different roles, which they validate with manual coding. Zero shot was good, few shot was better.

Miyoung Chong presents her work with @smithmarisaa.bsky.social and Om Sai on using LLMs for stance detection (zero and few-shot). They apply this to the case of CRT discourse.

Longhan Wei presents his work with Tian Yang looking at ChatGPT users and their online behavior, showing a possible amplification effect. #ica25 @icacm.bsky.social

Torie Kim (my student!!) presenting our work on using open source LLMs (who talk to each other) to label text data. #ica25 @icacm.bsky.social

Standing room only at @zeningduan.bsky.social and @alvinyxz.bsky.social 's @icacm.bsky.social panel on LLMs #ica25

🔎 How are different genders depicted in science discourses across digital platforms? In our paper at JCMC, we analyzed text and visuals in #TikTok and #YouTube videos, revealing patterns on gender stereotypes and user engagement. #scicomm #datascience 👇 academic.oup.com/jcmc/article...

Feels so good to move to here. HELLO, BlueSky!