Ian Hussey

Ah yes, my "favorite"...

When you don't like the result from your RCT, just focus on within-arm change in the active arm, even if it's exactly the same as that seen in the placebo arm. Scientists and journalists let people get away with this all the time.

As we explain in our recent paper, psychology lacks the coordination that is present in other fields. We do studies we can do, instead of getting together to do the studies we should do.

Paper: www.nature.com/articles/s41...

Preprint: osf.io/preprints/ps...

This is a carry over from poor measurement practices. Eg faced with the reality that the measure is unreliable and has poor convergence, we tend to fall back to the construct being unstable and not the measure being poor. We’re baking in unfalsifiability.

a young man with glasses is screaming in front of a sign that says tight ALT: a young man with glasses is screaming in front of a sign that says tight

But Julia, have you considered that perhaps we have created god? I haven’t checked, but wouldn’t that be tight?

I really hope some ethnologist is observing how psychologists conduct research on LLMs because that stuff can get wild and probably tells us a lot about psychologists.
(Less so about LLMs)

Our (@malte.the100.ci, @ianhussey.mmmdata.io) commentary "cognitive dissonance in large language models is neither cognitive nor dissonant" has just been accepted for publication at PNAS!

Check out the thread below for a summary of our arguments.

osf.io/preprints/ps...

PSA future employers of my students: Asking me in your reference form whether my student is "loyal and discreet" (not applying for the MoD here) and "prone to continual days off sick" is not a good look!

🤯🤯I AM SO MAD RIGHT NOW! 🤯🤯

I would say the opposite, we must start from the assumption that nothing measures anything until proven otherwise. That’s what we (should and somewhat usually) do with substantive hypotheses.

Why assume that they measure the same thing, or anything at all, if reliability convergence is that low? I thing we have made it axiomatic that things measure what we assume they do, and that’s deeply problematic.

Model-based exploration is measurable across tasks but not linked to personality and psychiatric assessments - Scientific Reports Scientific Reports - Model-based exploration is measurable across tasks but not linked to personality and psychiatric assessments

Today in schmeasurement:

Commonly used multi-armed-bandit behavioural tasks nominally measuring exploration behaviours have terrible reliability (ICC .1 to .5), minimal correlations with one another, and therefore poor replicability.

Does peer review work? Is it worth it? What are alternative quality management devices in science?

Hadn’t read this one!

The COVID-19 pandemic transformed this scientist into a research-integrity sleuth Lonni Besançon has faced online abuse, threats and legal challenges. But he remains proud of his service to science and society.

Lonni Besançon faced online abuse, threats and legal challenges in response to his research integrity sleuthing during the COVID-19 pandemic

go.nature.com/4m5DN0X

“6% of sleuths agreed that current reporting channels for misconduct are effective, compared to 77% of research-integrity officers”

I think this is readily explainable by the differential access to information. RIOs do not see what doesn’t even reach them, eg when editors refuse to act.

I also added a simple Shiny app I wrote to visualise Cohen's d when its calculated from Likert data. This can be useful to sanity check a given effect size for a given scale's constraints and norms.

Examples welcome!

(1/6) New pre-print alert! 🤩🚨

We provide new evidence for contra-hedonic functions of NSSI & ED behaviors in adolescents.

See 🧵 for a brief summary 👇
osf.io/preprints/ps...

Check out this new exciting work by @francesghart.bsky.social! We find that contra-hedonic functions of NSSI & eating disorder behaviors are not just present but *prevalent*.

In other words: sometimes people engage in self-destructive behaviors not to escape negative emotions, but to amplify them.

This post also gave me an opportunity to include this figure, which I love.

If researchers find Cohen’s d = 8, no they didn’t.

(probably.)

Thoughts on the distribution of Cohen's d values in psychology and why we should think about their plausibility when reading articles.

I will today. I’ve contacted Gloster before about impossible results in other articles and he’s never replied.

pubpeer.com/publications...

Definitely that but also other issues, boring to explain here. This is just correcting the Standard Error error for that one study.

Cohen's d = 8 in treatment resistant patients? 👀

ERROR @error.reviews was awarded a Commendation from the Society for the Improvement of Psychological Science!

a person in a green scrub is walking down an empty city street . ALT: a person in a green scrub is walking down an empty city street .

The American mind cannot comprehend how empty European faculty offices are during summer. I think there’s one other person in the building right now.

Surprised Pikachu GIF ALT: Surprised Pikachu GIF

MFW it turns out that floating point precision errors do indeed cause all sorts of odd behavior, are a pain in the ass, and I can no longer ignore them.

John Cleese and John Goodman looking like scientists

Never-before-seen footage of me and @jamesheathers.bsky.social working on GRIM, unearthed by @samwestwood.bsky.social

The other fascinating/terrifying analysis is the collective cost of grant application (total hours of work for applicants and reviewers) vs the amount actually distributed by the grant. Many smaller grants cost the taxpayer more to apply for than they distribute.

Even @robcaliff.bsky.social says "real world" data is only at the "promise" stage. The only thing it seems to be "missing" is accuracy, completeness, and the ability to account for bias. Their words not mine! 😜 www.nejm.org/doi/full/10....

ht @emilymoin.com