At what point in your career do people stop telling you that your definition of a p-value is wrong?

Really enjoyed this clear and concise presentation of the problem with studies that throw 10 different covariates into a multiple regression and interpret them causally. So obvious in hindsight that this is problematic, but I have definetly been guilty of this. bmjmedicine.bmj.com/content/4/1/...

Sunk costs and spite are basically the vodka red bull of the examined life

An LLM-controlled robot experiences a mental breakdown during a test to pass the butter from the kitchen. arxiv.org/abs/2510.21860 Some snapshots of its inner monologue from Appendix D:

📝 New Research Methods & Reporting article: Covariate adjustment in cluster randomised trials 📝 A practical guide for when and how to adjust for covariates. Read more ⬇️ www.bmj.com/content/391/...

Idea for a Christmas BMJ paper: submit loads of Christmas BMJ papers, then do a survival analysis of time to rejection. Then submit the survival analysis as a Christmas BMJ paper the following year

Also, sometimes people just either don't care, or more commonly are so in love with a hypothesis that they ignore other possibilities. Personally I don't think methods resolve this, it just needs hard case by case work. academic.oup.com/ije/article/...

It's possible to prove that under certain conditions, shrinkage plus data perturbation will lead to asymptotically zero false positive rate - but, it makes interpretation of the regression coefficients even harder.

We wrote an article explaining why you shouldn't put several variables into a regression model and report which are statistically significant - even as exploratory research. bmjmedicine.bmj.com/content/4/1/.... How did we do?

Sports stats

A really good talk on the stepped wedge design! Including some fascinating history of the first ever stepped wedge trial

Are cloud based data analysis platforms a massive practical joke

Annual PSA: if the interaction term in your regression model is significant, this suggests a deviation from additive or multiplicative effects depending on if you choose an additive or multiplicative model. Interactions can actually reverse depending on this decision!

When a corporate department gives you a template to complete, why is it always in Word 97-2003 format with floating text boxes and tiny margins

There are parallels with the "industry playbook" of citing insufficient evidence to support public health interventions

Thank you!!

The more I use R, the more I realise they did a pretty good job of deciding what you probably need in the basic package #RStats

Top tip for journal editors and funding bodies who review evaluations of public health interventions: rather than send it to peer reviewers, just ask "did you realise your intervention is clustered?"

Genuinely odd comparison!

jumpscare on instagram @dandekadt.bsky.social

If you never set par(xpd = NA) when making plots in R, you haven't lived

'He might be telling those people to go home. As our country's health and social care secretary, I want to say directly to those people... they are home' Wes Streeting criticises Reform's plans to deport people from the UK if they fail to meet stricter rules👏

"While many people would demand a significant amount of money to stop using Instagram and Tik Tok, they would also be willing to pay to eliminate Instagram and Tik Tok from their community ... people often spend time or money on goods whose existence they deplore." papers.ssrn.com/sol3/papers....

What are the trajectories of a thing? And following that, what factors are associated with those trajectories: a multinomial regression analysis with stepwise variable selection and null hypothesis tests that you didn't know exist

When research papers start out with "little is known about X", you must wonder if people actually need to know about X

My most dreaded situation: being in a small group discussion at a workshop and being nominated to feed back to the whole group

Seeing students getting lost with methods and chatGPT. It's the first point of call, rather than textbooks or foundational methods articles. chatGPT is taking them off on the wrong path and they don't realise it's not always a great source, or consider reading original sources to be labour intensive

Anyone here into unlocking potential or harnessing the power of stuff?

Prof John Wright, Director of Born in Bradford @bihr.bsky.social, giving us a warm welcome to Bradford

Personally I just can't get on board with anything that is not a panacea. What's the point? ❌Small effect sizes ❌ Anything multifaceted ❌The Rose Hypothesis ❌"Partial" achievement of learning outcomes in your educational portfolio ❌Results that might be useful as part of a meta analysis No thanks!

The madly insane efforts of reviewers to reject my research are somehow indications of its power

If it's not a panacea, I'm not interested

Let's stop studying things that are not silver bullets. Silver bullets seem much more effective and I think we should focus on those

See if you can figure out this note to an NHS press release, explaining what it means when there is no statistical evidence of a difference between two numbers - www.england.nhs.uk/2023/11/one-...

Ranked list of "things" most feared by Catholic high school boys in Maryland, 1939. Rank (prevalence). www.tandfonline.com/doi/abs/10.1...

Given that trails recruit participants with different baseline risk, eg. because participants in one trail are older than those in another, does it make sense to pool relative effect sizes in meta analysis?

Me before revising the manuscript: WTF. I hate doing this. Me after revising the manuscript: Well, that was pleasantly easy. Nice work. Actually better than before.

Among people who use heroin, tobacco smoking and illegal drugs cause a similar number of premature deaths buff.ly/tIC0Fqp @danlewer.bsky.social @tattanbirch.bsky.social @sharoncox.bsky.social

New episode of Money Power Health with the wonderful Dr Nicola Dennis and Dr Chantal Edge on prisons, justice and health in the UK: open.spotify.com/episode/48Cg... Ad free on Apple, spotify. @uoe-sps.bsky.social @lhgp.bsky.social @adphuk.bsky.social @ukhsa.bsky.social @healthfoundation.bsky.social

The appropriate technical term is bullshit

The truth is that what leads you astray is not the model, it's staring at tables of model coefficients to figure out what your research question was.

Tobacco smoking & illegal drugs each appear to account for ~ 1/4 of premature deaths among people who use heroin. Smoking also exacerbates their poverty which makes quitting heroin even harder. Also reduces treatment outcomes onlinelibrary.wiley.com/doi/10.1111/...

We did some life table modelling among people who use heroin. Tobacco smoking causes as many premature deaths in this group as illegal drugs. Yet public health policies overlook smoking: onlinelibrary.wiley.com/doi/full/10.... New research with @sharoncox.bsky.social @tattanbirch.bsky.social

when this shows up at like 70min I had to pause the video and go show my wife that bayesians are mainstream now

Omg I'm in my own meme

What is common knowledge in your field, but shocks outsiders? On the whole, screening people for asymptomatic diseases is harmful

"Dear Academic. Would you like to spend a year putting together a special issue for our journal? You will provide us with all your contacts and edit the issue for us for free. We will then sell access to the work. In return for your efforts, you will be allowed to read the issue, for no extra cost".

We've just published a scoping review of projects that aim to improve opioid agonist therapy (eg. methadone) in acute hospital settings: journals.plos.org/mentalhealth.... This is part of our excellent iHOST project (www.lshtm.ac.uk/research/cen...). A couple of findings ...

I beg you please stop calling p-hacking exploratory analysis. Call it data dredging, fishing expedition, cherry picking, significance chasing. Labeling your non-pre-reg'd analyses clearly as exploratory isn't a license to p-hack. Stop bastardizing scientific exploration and stop hiding behind it.