Dr Lena M Blott

I would also recommend Gorilla! Really intuitive to use.

These look fab! Do this everywhere, please?

Researchers: are you reassured by Prolific's efforts to tamp down bot activity? Have you seen changes in data quality? We've gone from rejecting about 10% of the sample for headphone checks to more like 40%. Not confident the rest are real.

Another vote for minutes!

Just as a comment: you're right that F1 and F2 ANOVA used to be standard in psycholinguistics, but these have been replaced by GLMMs and LMMs in the past decade. Haven't seen an ANOVA in a psycholinguistics paper in a good few years!

In my ongoing attempts to give you a break from *gestures widely* THE HORRORS:

Let's talk about the utterly bizarre science of cats' purring.

Here's a weird starting point:
There are NO cats (Felidae) that can BOTH roar AND purr.

Cats that roar = genus Panthera
Cats that purr = genus Felis

YouTube video by Angela Collier this is what 2 years of chatgpt does to your brain

A very funny and satisfying rant by @acollierastro.bsky.social about the impact of "AI" on education and specifically the professor who outsourced his work to ChatGPT, lost it, then told the world about it (25 min.) www.youtube.com/watch?v=7pqF...

So wholesome, thank you for sharing!

Honestly, these are so good. Thank you!!!!

Thank you so much for all the work you've put into producing these, and for generously sharing them. I'll be assigning them in my class this term. Not just as an explainer of how LLMs work but to hopefully get students to reflect on their usage of them.

I've been looking for an easy-to-follow explainer on how LLMs work for a class I'm teaching this term, but this series is so much more than that. Worth sharing with students (and everyone else)!

I'm looking to interview, read, and share disaster narratives. If you have a story to tell, data to share, and want to talk, I'm taking trains and bikes across the EU and want to talk to you. #academicsky #disasters #openscience

www.andrewrusso.me/dnp

So... as I write syllabi for next semester, I have to ask a question that makes me feel incompetent:

Does anyone have solid resources on AI policies in class? This could be how to organize assessment to minimize AI use, or any of the other Faustian bargains.

#sociology #academicsky #highered

It’s the same with academic writing. The difficulty doesn’t lie in formulating grammatical sentences, but in deciding what it is you actually want to say. Writing is part of the process by which you figure that out, and isn’t something that can be outsourced to a machine.

You may be surprised to learn that authors of scientific articles don't get paid for said articles ( which typically already contain an abstract)

We present our new preprint titled "Large Language Model Hacking: Quantifying the Hidden Risks of Using LLMs for Text Annotation". We quantify LLM hacking risk through systematic replication of 37 diverse computational social science annotation tasks. For these tasks, we use a combined set of 2,361 realistic hypotheses that researchers might test using these annotations. Then, we collect 13 million LLM annotations across plausible LLM configurations. These annotations feed into 1.4 million regressions testing the hypotheses. For a hypothesis with no true effect (ground truth $p > 0.05$), different LLM configurations yield conflicting conclusions. Checkmarks indicate correct statistical conclusions matching ground truth; crosses indicate LLM hacking -- incorrect conclusions due to annotation errors. Across all experiments, LLM hacking occurs in 31-50\% of cases even with highly capable models. Since minor configuration changes can flip scientific conclusions, from correct to incorrect, LLM hacking can be exploited to present anything as statistically significant.

🚨 New paper alert 🚨 Using LLMs as data annotators, you can produce any scientific result you want. We call this **LLM Hacking**.

Paper: arxiv.org/pdf/2509.08825

Text reads: About synthetic panels Recruiting the right participants for a study can be difficult. You may not get the exact demographics you need, and the shorter the deadline, the less sure you can be that everyone will answer on time. One possible solution can be to use synthetic panels. Synthetic panels are powered by a first party proprietary AI model developed here at Qualtrics. Our synthetic panel is trained on thousands of responses from a variety of demographic backgrounds in order to more accurately predict how certain populations would respond to a survey. Our synthetic panel is based on the United States General Population, and is only available in English. This panel comes with ready-made quotas and target breakouts in order to represent your chosen population and make it easy to launch your survey right away.

Text reads: Question-writing best practices To get the most reliable and actionable results from synthetic audiences, consider these question-writing best practices: Ask forward-looking and attitudinal questions. Synthetic panels perform best with perceptions, preferences, and intent-based questions. For example, “How likely are you to try…?” Synthetic panels are less applicable for studies on past behaviors, detailed recall, brand recall, or awareness questions. For example, “When did you last visit…?”

Text reads: Discussion The current study aimed to conduct a meta-analysis of the TPB when applied to health behaviours which addressed the limitations of previous reviews by including only prospective tests of behaviour, applying RE meta-analytic procedures, correcting correlations for sampling and measurement error, and hierarchically analysing the effect of behaviour type and sample and methodological moderators. Some 237 tests were identified which examined relations amongst model components. Overall the analysis indicated that the TPB could explain 19.3% of the variance in behaviour and 44.3% of the variance in intention across studies. This level of prediction of behaviour is slightly lower than that of previous meta-analytic reviews which have found between 27% (Armitage & Conner, 2001; Hagger et al., 2002) and 36% (Trafimow et al., 2002) of the variance in behaviour to be explained by intention and PBC.

Did you know that from tomorrow, Qualtrics is offering synthetic panels (AI-generated participants)?

Follow me down a rabbit hole I'm calling "doing science is tough and I'm so busy, can't we just make up participants?"

I'm trying to understand why such garden paths don't trigger that "ungrammaticality" intuition but sentences like "I want going to the cinema" do (even though there too, you could ignore the difficulty in the grammar and find a possible interpretation)?

In that literature, it has often been found that ppl are surprisingly resistant to fully resolving a difficult parse (e.g. they retain the interpretation that 'Anna bathed the baby' when having read 'While Anna bathed the baby started crying').

Super interesting! I've not had time to read the paper, so forgive me if this is something you're covering, but I'm wondering whether you have thought about how to bring the interpretation of garden path sentences into the picture?

Merry Christmas @jennirodd.bsky.social

Commiserations, my hair is going through the same right now 😭 I'm hoping i can avoid the full on 80s mullet i was sporting after my first pregnancy this time around though

[Scene is a kitchen - a middle aged woman called JANET is boiling peas at the stove. A younger more colourfully dressed woman named LIZ approached her.] JANET: Ugh... LIZ: What's up? JANET: I am so bored of cooking peas! LIZ: Have you tried... AI peas? JANET: AI peas? LIZ: They're peas with AI! [Liz holds up to us a packet of peas labelled: Pea-i AI - Peas with AI]. LIZ: Al-powered peas harness the potential of your peas JANET: What LIZ [Now a voiceover as we cut to a whizzy technology diagram of peas all connected by meaningless dotted lines] Why not take your peas to the next level with Al Peas' new Al tools to power your peas? [Show a techno diagram of a pea with a label reading 'AI' pointing to a random zone in it] LIZ: Each pea has Al in a way we haven't quite worked out yet but it's fine [Show Janet and Liz now in a Matrix-style world of peas] LIZ: With Al peas you can supercharge productivity and make AI work for your peas! JANET: What LIZ: Shut up LIZ: Our game-changing Pea-Al gives you the freedom to unlock the potential of the power of the future of your peas workflow From opening the bag of peas to boiling the peas to eating the peas To spending millions on adding Al to the peas and then having to work out what that even means. JANET: Is it really necessary to- LIZ [Grabbing Janet by the collar]: THE PEAS HAVE GOT AI, JANET [Cut to an advert ending screen, with the bag of peas and the slogan: AI PEAS: Just 'Peas' for god's sake buy the AI peas. [Ends]

Every ad now

Very cool!

New favourite ambiguity example just dropped @jennirodd.bsky.social

The advice I give young people on choice of degree programme. Choose something you love. You’ve got the rest of your life to be frustrated and miserable.

You are right: journals don't pay authors, and sometimes authors even have to pay to publish (it's a silly system!). Any study author would most likely be flattered if you emailed them saying you're interested in their work and probably also more than happy to send you a copy. Definitely ask 🙂

Isn't that essentially what French does in the regular "qu'est-ce que c'est?" ?

I'm a linguist. Can I just say: prescriptivism is yuck.

Beautiful thread