Andrew A. McAleavey

Can you use off the shelf generative LLMs (eg ChatGPT, Claude) to assess personality from brief bits of unconstrained, open-ended text? If so, how well do they do?

Turns out you can, and they do amazingly well. I couldn't believe how well.

Here's our pre-print on it and a 🧵

Alternate angle of the mustache mafia:

🧵 Psychology practice & training in the U.S. is at an inflection point. The APA is developing guidelines on the scope of practice for master’s-level practitioners—and what training for these roles should look like. #Psychology #MentalHealth 1

it‘s been a first — @sprofficial.bsky.social! a blast of diverse, topical inputs; inspiring people, #starstruck. already looking forward to the next years! #sprkrakow2025

Depending on my mood: youtu.be/-KACt6YhOyY?...

White line graffiti on a yellow-brownish wall, depicting a stern or grumpy looking man with a moustache and a high, wild hairdo

Well, until I saw this picture of you, @andrewathan.bsky.social , I thought it was only my strange mind. But now I am convinced, that I found a portrait of you when I was hunting for #Graffiti in #Krakow

Admittedly, you usually don't look as grumpy 😅

#SPRKrakow2025

Found a cheerful but slightly hungover @herzog.bsky.social on the plane home from #SPRKrakow2025 See you next year, @sprofficial.bsky.social

What an amazing time at #SPRKrakow2025 💫 I feel so honored to have received the Collaborative Grant to work with @nilisolomonov.bsky.social on investigating late-life suicidality using EMA. Grateful for the inspiring talks, the deeply supportive community, and so many meaningful conversations

I'm most of the remaining 1% and proud of it

Totally agree.

Looking forward to it! To be clear, I totally agree that better analytic decisions should generally be used esp when CIs have been misused. Despite how hard it is to write concisely for me, I think we are mostly just three dudes agreeing

Yeah I’m not saying we shouldn’t model correctly, just that it’s a tool for people to use, so the best test is whether it helps improve decisions, in this case about treatment. I’d definitely hypothesize that a more accurate model would help make better decisions, but that’s an empirical question

Sounds good. In a way this kind of clinician-facing statistic is really still unvalidated because we have so few trials of reasonable alternates, and the ultimate validity question is how much they improve important outcomes. Beter to be wrong and helpful than the opposite, would be my preference

Looking even more forward to discussing the contributions in this panel as earlier this week this excellent paper by Jana Uher was published which refers to the work of multiple presenters:
www.frontiersin.org/journals/psy...

#SPRKrakow2025 #MixedMethods #QualitativeMethods #PhilSky #Psychometrics

Say more? Robust against what?

193. Data-informed treatment: New Methodologies, Mechanisms of Action, and Large Scale Implementation [Panel [In Person]] Saturday | 8:30 am-9:30 am | Aula Duża A Live-streamed Organizer: Kim de Jong, Leiden University, Netherlands Moderator: Kim de Jong, Leiden University, Netherlands Therapist expectations as a mechanism of progress feedback during psychological therapy Kim de Jong, Leiden University, Netherlands; and Melanie Simmonds-Buckley, University of Sheffield, UK show abstract Data-informed feedback research using clinical support tools enhanced by Large Language Models (LLMs) Wolfgang Lutz, University of Trier, Germany; Steffen Eberhardt, University of Trier; Rebekka Wassmann, University of Trier; and Antonia Vehlen, University of Trier show abstract Implementation of feedback-informed treatment in the UK National Health Service Jaime Delgadillo, Kings College London, UK show abstract Discussant: Andrew Page, University of Western Australia, Australia;

Very good panel with @kimdejong.bsky.social presenting on a new #ROM #CST study,

@wlutzpsyres.bsky.social briefly touching on the Trier Treatment Navigator (*1) and the Transtheoretical Treatment and Training Model (4TM, *2)

Jaime Delgadillo reports on extensions of their...

#SPRKrakow2025

I missed @herzog.bsky.social talk but luckily it was recorded - schema therapy and DBT were equally good at reducing costs from illness from severe borderline PD. #SPRKrakow2025

Her findings suggest that therapists sometimes feel a patient doesn’t fit them, often when patients are dissimilar from the therapist in some way, which lowers the therapists’ expectations

A lot of good things at #SPRKrakow2025 this morning, but here is the excellent doctoral student Janne Låver presenting her recently published paper on how therapists’ outcome expectations are formed!

Does anyone at #SPRKrakow2023 know why the slides are approximately a million miles off the ground?

This is 100% my experience too. This city doesn’t sleep

We have a good crowd at our #SPRKrakow2025 pre-conference workshop on using measurement in psychotherapy. I realize this picture doesn’t prove there’s a crowd, please just believe me

Attn #SPRKrakow2025: there are two awesome but somewhat breakable things in your registration bag. A mug and a dragon candle #dragonsrule

PhD position in psychometrics and self-reported health data (282039) | Western Norway University of Applied Sciences Job title: PhD position in psychometrics and self-reported health data (282039), Employer: Western Norway University of Applied Sciences, Deadline: Monday, August 18, 2025

If you or anyone you know wants a PhD working with self-report data from mental health in Norway, come! www.jobbnorge.no/en/available...

#SPRKrakow2025

bsky.app/profile/did:...

A picture of a person dancing at a nightclub

Me on the dancefloor at #SPRKrakow2025

Ok I’m using #SPRKrakow2025 but very disappointed to miss the chance to use the ó

No I’ll be at #PsychotherapyResearchInKrakowHybridOnlineAnnualInternationalMeet but we have a journal meeting that morning so can’t attend this session

I will do whatever the crowd does, we might need an official adjudication from @sprofficial.bsky.social or @spr-eu-youngs.bsky.social or @tkaiser.science