Nick Holmes

youtu.be/S-sqsvcZJ2w?...

Your recovery is impressive, and such beautiful data too. Best wishes 🙏

Has anyone tried teaching *only* Bayesian inferential #stats for undergrad psych (or other UG sosci programs)?
Like completely foregoing NHST, confidence whatevers, p values, CLT...

I'd be very interested in seeing a syllabus of such a course/ program!

Wow, no - the blowback I'd get from all other staff in my dept would be career-ending!

(also I'm not a Bayesian, yet)

Tickell is now running the uni I work for. I hope lessons have been learned 🙏.

In Sussex? Sorry to hear that.

Confirms my biases, so reposting 👇

It's not just 'bots' that are problematic for online research, but also fake humans:

doi.org/10.2196/58432

doi.org/10.1016/s221...

Just what we need in the Hand Lab - a wipe-clean chair for our #TMS participants 👍

If you think 'AI' is better at doing research than you, then I think you are not very good at doing research.

Unless by "doing research" you mean producing text & images sufficient to convince non-experts that you're doing research.

We have enough of the latter kind of research already, thank you.

(which, to be clear, is not my view!)

Says a lot about "most professors"

Appreciate the aim, but why put outlier removal as a preprocessing step? 'Outliers' can only be determined after the full model is fit, visualised and everything understood. Appreciate this is common practice but it drives me nuts that people just throw away eg 5% of the data before looking at it 😵‍💫!

One of my favourite sides of politics/media.

Oh, interesting, thank you - did not know that was an option!

The FTSE 100 stock price, 2016 to 2026. Going up wildly from 2025. Bubble?

I'm sure this is totally fine and doesn't suggest a bubble at all. #FTSE

Ahhhh, now i'm beginning to understand what folks mean by "shiny apps" etc, neat that it uses R.

I'm clearly old skool, so i will start with html, php, and mysql. I'll have to collaborate with someone much cooler than myself to do anything in shiny/R/python!

* this is the main aim really - to make a living meta-analysis on the web, so i can code it all in php/html (until someone shows me a better way)

Ah, good to know! I've implemented everything apart from recalculating the weighted mean on each iteration, so i'll do that bit tomorrow.

After that, i'll do the other pub. bias tests, then maybe MLE, then put it onto a website*

(I've not found much matlab code for meta-analysis, wonder why not?)

It was a good one. I just noticed that trim & fill is not in my JASP anymore - i guess it's been abandoned?

To understand. Then to do something better 🙏

Spending some time coding the "trim and fill" algorithm for #metaanalysis in matlab, working from some existing R code (which i don't understand) and the original paper (which i also don't understand). Plan to test it against results in JASP (which i don't understand). What could go wrong? #stats

[cue a horde of slop-jockeys coming here to tell me how wrong i am and how i should really just try it already]

It's good to know that my skepticism of LLMs remains well-founded (yes, even Claude, about which I've heard multiple folks say: "But have you tried Claude yet?"). I want to be an *expert*, to *understand* & produce high quality work. I'm not interested in sloppy short cuts.

Thank you Richard 🙏

Javascript code for a noncentral t cdf produced by Claude. It is simply a call to the normal CDF, which is not correct (though will be a decent approximation with large N).

I've been testing Claude to see how well it can "vibe out" a stat. power app that I've already coded completely myself - so I know what I want. It mostly gets things right with animations (those are easily verifiable) but looking into the backend stats code is nightmare inducing (see pic).

Just FYI: amphibians migrate around this time each year in the UK. A lot become too flat to sustain life, following sudden unexpected vehicular-tarmacadam compression syndrome (SUVTCS). You can help with just some rubber gloves, high vis, and a bucket.

🐸🦎

20 toads emptied out of a bucket. I mean, what more did you expect here?

If you've ever wondered what 20 toads emptied out of a bucket looks like, it's this 👇

I lived in Fallowfield & Withington for three years in the 'mad fer it' 90s :-)

Yup. 6-8 weeks. MSc are 1 year in total. In my current dept. they are mostly physiotherapy students, so working 9-5 on placements in 6 week blocks all year.

Some programmes have longer for the project, but my experience of UK in general is that it has been squeezed from all sides. Not ideal!

✅️ Our MSc students typically have 6-8 weeks for their diss. There's no way designing, ethics approval, running, analysing & writing a study can happen in that time.

I dont use a list, but say: here's the methods we can use, and some questions i think are interesting, then we co-design the details.