Lightnews — Scholar-powered news

Danny Maupin

Danny Maupin

@dmaupin.bsky.social

120 followers 1K following 140 posts

🔬Research Fellow in Health Science University of Surrey 🩺Specialist Vestibular Physiotherapist

Posts Media Videos Starter Packs

Danny Maupin @dmaupin.bsky.social · 1d

Discovering Statistics by Andy Field - he does a great job introducing beginner stats concepts in a way that's not too heavy. Great entry for those with non-stats/math backgrounds

Danny Maupin @dmaupin.bsky.social · 3d

Employment benefits to underemployed workers to benefit their mental health. Now time to find a home for this piece!

#academicsky #opensci #episky

Danny Maupin @dmaupin.bsky.social · 3d

New preprint on MedRxiv!

In this we looked at how unemployment benefits impact the mental health of underemployed workers. We had relatively small sample sizes so couldn't use causal methods and had to settle for meta-regression, but this piece provides some initial evidence for expanding /1

The Impact of Unemployment Benefits on the Mental Health of Underemployed Workers: Findings from the Understanding America Study

Background: Underemployment, working fewer hours than desired, has a negative impact on mental health. Research suggests that unemployment benefits can support the mental health of unemployed individu...

www.medrxiv.org

Danny Maupin @dmaupin.bsky.social · 9d

Impactful publications but that is hard to measure

Danny Maupin @dmaupin.bsky.social · 9d

Particularly if all that is being churned out is junk or fast churn science. It's easy to use AI to develop about x predicts y and another paper about z predicts why or redundant publications. This has been a problem before AI but is accelerated now. Looking forward to it publishing more robust /

Danny Maupin @dmaupin.bsky.social · 10d

No I don't think it would be, just write things vague enough

Danny Maupin @dmaupin.bsky.social · 10d

You think he will bother pre-registering? I'm doubtful

Danny Maupin @dmaupin.bsky.social · 10d

I may be confusing the proof-evidence distinction so please educate me if I am, but I don't think we make proof

Danny Maupin @dmaupin.bsky.social · 10d

The hypothesis remark was specific to a reply about looking for a hypothesized effect.

I don't think you make proof if you're doing research without a hypothesis though? You are discovering proof that exists. I think people take issue with the word make particularly with someone openly biased

Danny Maupin @dmaupin.bsky.social · 10d

The same page about the idea being ridiculous

Danny Maupin @dmaupin.bsky.social · 10d

So I still take issue with the make proof aspect, we could test the hypothesis, collect data all that but make proof has a different connotation especially when said by someone openly biased.

I think this conversation is in good fun as it's interesting debating science philosophy and glad we are /

Danny Maupin @dmaupin.bsky.social · 10d

But we shouldn't look for hypothesized effects. We make a hypothesis and then we test it. That's different. And even the species example the more I think about it isn't making proof, we are looking for evidence yes but we don't make the proof. The proof has existed previously we just didn't see it /

Danny Maupin @dmaupin.bsky.social · 10d

using science as a monolith when there are examples that prove otherwise so they probably should've been more specific

Danny Maupin @dmaupin.bsky.social · 10d

I'd say that's a little disingenuous because we aren't looking at discovering new species, this is specifically health research, and not health research that is looking for a new disease, but looking at causality of a condition with multiple confounders. The other poster is being too broad /

Danny Maupin @dmaupin.bsky.social · 10d

Not that it always is as researchers often have biases, but this feels like poor wording from individuals that don't seem concerned with gold standard science despite them saying they are

Danny Maupin @dmaupin.bsky.social · 10d

I think people take issue with the wording due to the belief that they are going to interfere with any data or conclusions to back up their point, thus making proof.

If I understand correctly you are saying science is always looking to make proof, but ideally this will be proof for or against /

Danny Maupin @dmaupin.bsky.social · 10d

Congratulations Niall! Looking forward to working alongside you

Danny Maupin @dmaupin.bsky.social · 10d

Excited to be a part of this cohort and can't wait to work with fellow colleagues to explore this area further

#academicsky #metascience

UK Research and Innovation @ukri.org · 11d

We're pleased to be funding early career researchers to explore how AI is transforming science.

The UK Metascience Unit has announced a cohort of 29 researchers that will receive funding through the AI Metascience Fellowship Programme.

Read more: www.ukri.org/news/interna...

International fellowships to explore AI’s impact on science

New £4 million programme funds early career researchers in the UK, US and Canada to investigate how artificial intelligence (AI) is transforming science.

Danny Maupin @dmaupin.bsky.social · 11d

Really good blog, enjoyed how you didn't completely dismiss the appeal of using LLMs. I do worry about ChatGPT being done for stats, particularly as less stats disciplines use it. What happens if they say o just a do t test on this data cause that's what I know even if it's not ideal for the ?

Danny Maupin @dmaupin.bsky.social · 12d

Awesome thank you!

Danny Maupin @dmaupin.bsky.social · 13d

Was this recorded?

Danny Maupin @dmaupin.bsky.social · 13d

On impactful outputs.

Danny Maupin @dmaupin.bsky.social · 13d

That makes sense. I guess my concern (mentioned in that thread discussion) is that single papers often become such hot topic despite being done poorly (to be fair different issue) or not being replicated but the idea sticks in the public. This may be more of a fault of science as a whole with focus

Danny Maupin @dmaupin.bsky.social · 13d

Looking at this. Am I right in assuming that you are not worried about replication crises because that's what science should do? Continue to iterate and drop ideas that consistently don't work even if there are odd results that do?

I get too that this doesn't mean fraud stat magic etc.

Danny Maupin @dmaupin.bsky.social · 13d

Interesting thread and thank you for sharing! My first thought when it comes to replication is being able to produce the same result with the same data as described in methods though I know this is not always what is assessed. Your thread has communicated well the idea the variation in studies 1/