30-40 per year, and probably somewhere around 5 hours on each.

Ah, interesting! It works for a 2x2 design, but seems to go wrong again if you add factors. My concern is that researchers often use default settings, which give strange values - and mispowering studies as a result. I’ve lost count of the number of times I’ve seen this in manuscripts.

And the errors it makes are so blatant. Eg if I use “within-between interaction” for a 2x2 design, GPower says I have much greater power than for the same sample size for an independent t-test, even tho they are doing same thing (interaction is just a t-test of the within-subs difference scores).

Right, there may be a way to fiddle things to get the correct answer in GPower, but it’s not obvious how.

I often see papers using GPower for power analysis in repeated measures factorial ANOVA designs. I don’t know what it’s doing, but it’s definitely giving wrong answers: substantially underestimates required sample size. MorePower seems like a much better alternative github.com/LewisPeacock...

Wow! That’s pretty off. As you say, the way it’s written gives the strong impression that you were interviewed. Shoddy work.

A lovely paper David! Muy bien.

Habits Need Time: Evidence of Habit-like Behavior After Short Training Likely Reflects Failures in Habit Testing: https://osf.io/zh39t

Agreed!

New paper in Cognition and Emotion with Daniel Pearson and @mikelepelley.bsky.social! We focus on best practice for measuring attentional disengagement - we hope it will be a useful reference for all attention researchers, and those interested in the role of attentional processes in psychopathology.

📢 Department of Psychological Sciences, School of Health Sciences, Swinburne University of Technology is hiring! 📢

There is an ongoing, full-time, level B position available in the domain of Clinical Psychology. Applications close 23/02/2025.

Learn more here: swinjobs.nga.net.au?jati=14B144C...

Agreed - thanks for organising, @tombeesley.bsky.social ! A really fantastic symposium.

New paper in QJEP with Dave George, Dom Dwyer and @markhaselgrove.bsky.social. We discuss a recent claim to have shown complex statistical reasoning in crows, and show the simplest S-R learning would give the same behaviour. A fun paper to write with old friends.
journals.sagepub.com/doi/10.1177/...

I’ve spoken to students about this - they tell me they prefer to watch recordings of live lectures than pre-recorded versions (even though I’m guessing the quality of pre-recorded is often higher). They prefer the interactive feel of live lectures, even though they’re not actually there. Very odd.

Keywords for this opinion? And who would you suggest to review it? Did you preregister your opinion?

Scandalous! I thought Bluesky was meant to be a haven from hate speech. (And not guilty in this case - I’m not sure I agree with the replicate/reproduce distinction anyway. Surely you just need to be careful to specify what you are replicating: method or results?)