Lightnews — Scholar-powered news

Peter Tennant

@pwgtennant.bsky.social

7.7K followers 1.8K following 2.7K posts

Epidemiologist with an interest in causal inference methods at @universityofleeds.bsky.social. Check out my Intro to Causal Inference Course: https://www.causal.training/ #Epidemiology, #EpiSky, #CausalInference, #CausalSky, #AcademicSky

www.causal.training

Posts Media Videos Starter Packs

Pinned

Peter Tennant @pwgtennant.bsky.social · Jul 24

SAVE THE DATE: The 2026 IEA European Congress of Epidemiology and 70th @socsocmed.bsky.social Annual Conference will take place in London, UK on 8th-11th September 2026!

#EpiSky #EuroEpi2026

1 18 37

Peter Tennant @pwgtennant.bsky.social · 12h

Happy to help however I can - let me know

1 1

Reposted by Peter Tennant

Georgia Tomova @georgiatomova.bsky.social · 12h

Thank you for giving me the skills to succeed: DAGs and cynicism.

2 9

Peter Tennant @pwgtennant.bsky.social · 12h

We learn by doing. People seem to have forgotten that.

Reposted by Peter Tennant

Emily Moin @emilymoin.com · 13h

I am open to the idea that there are people who don't have and don't want to gain the skills to engage directly with their data but every single day that I do I learn the answer to a question you'd never even think to ask unless you were personally staring into the abyss of an uncleaned dataset.

5 7 55

Peter Tennant @pwgtennant.bsky.social · 13h

You have become nearly as cynical as me. My job here is done.

a cartoon says well my work is done here with a train in the background

ALT: a cartoon says well my work is done here with a train in the background

media.tenor.com

1 1

Reposted by Peter Tennant

Georgia Tomova @georgiatomova.bsky.social · 16h

"A UK study involving nearly 20,000 people, found that those who spent at least a total of 120 minutes every week in greenery were significantly more likely to report good health and higher psychological well-being."

is it ✨magic✨ or is it ✨confounding✨

Nature and outdoors can help boost your health - here's how

Spending just 20 minutes in nature can lower blood pressure, heart rate and stress levels.

www.bbc.co.uk

2 3 10

Peter Tennant @pwgtennant.bsky.social · 1d

Hope you have a great time!

Reposted by Peter Tennant

Big Book of R @bigbookofr.com · 1d

An Introduction To Forensic Metascience by James Heathers
#RStats
https://bigbookofr.com/chapters/field%20specific.html#an-introduction-to-forensic-metascience

1 7

Reposted by Peter Tennant

Prof Anna Watts @profannawatts.bsky.social · 1d

Lol the Nobels can't even acknowledge women's contribution to discovery. But sure let's acknowledge The Machines.

Headline from an article in Nature this week that states "Prizes must recognize machine contributions to discovery. The future of science will be written by humans and machines together. Awards should reflect that reality."

85 770 3.4K

Reposted by Peter Tennant

Peter Tennant @pwgtennant.bsky.social · 1d

There was a great post by someone saying that choosing to use an LLM for coding is always like choosing to be the hare in the 'hare and tortoise'.

You set off quickly but inevitably run into problems. Eventually your counterfactual tortoise self - who did by building knowledge - walks past.

1 7 34

Peter Tennant @pwgtennant.bsky.social · 1d

Basically the job training intervention was only given to a very specific people, such as those with a low-income, high school dropouts, those with prior criminal records, women on benefits, and Black men. The relationship with race, in particular, is so strong that you cannot control it away.

Reposted by Peter Tennant

Dan Lewer @danlewer.bsky.social · 1d

To be really cynical about it, I'd argue that the people most enthused by LLMs in quantitative research are those who never really liked stats and methodology

1 1 4

Peter Tennant @pwgtennant.bsky.social · 1d

In fairness, teaching causal inference to an audience that don't know basic Epi principles is extremely tough. Because 90%+ of good causal inference is about understanding good study design.

1 6

Reposted by Peter Tennant

Chelsea Parlett @chelseaparlett.bsky.social · 15d

Data Science programs often put too little emphasis on causal inference, and it’s hurting their graduates on the job market! The econometrics people are coming for your jobs lol

10 11 73

Peter Tennant @pwgtennant.bsky.social · 1d

The best thing about everyone using lalonde to test and explore propensity score approaches is the the treatment does not really satisfy the positivity assumption. Yet I almost never see this being explored or mentioned.

1 1

Reposted by Peter Tennant

Erick Scott @erickscott.bsky.social · 1d

Try adding: 'Don't be a sycophant' to your system prompt.

Gemini is more stubborn...

1 1 1

Peter Tennant @pwgtennant.bsky.social · 1d

More important, I personally think that developing educational materials to explain concepts and offer guidance to active researchers does not fall under what is generally classified as 'teaching' by the research ecosystem. Yet it is a crucial. And time consuming. It needs separate funding.

Peter Tennant @pwgtennant.bsky.social · 1d

Shhh. No-one would fund that!

1 2

Peter Tennant @pwgtennant.bsky.social · 1d

I haven't managed to publish my own explanation, but an old - somewhat incomplete - version is available as a preprint.

arxiv.org/pdf/2302.01822

arxiv.org

Peter Tennant @pwgtennant.bsky.social · 1d

Has to be said, there's a LOT of disagreement about the solution. And we don't all agree that Pearl's explanation is complete.

1 2

Reposted by Peter Tennant

Prof Christina Pagel @chrischirp.bsky.social · 1d

This is a vulnerability map for 8 key bodies: ONS, OBR, Food Standards Agency, MHRA, Met Office, UKRI, Natural England, & UKHSA.

ONS has the most independence. All but ONS have ministerially appointed leadership. 3 could be abolished with no oversight as they have no statutory basis.
6/11

1 23 89

Reposted by Peter Tennant

Prof Christina Pagel @chrischirp.bsky.social · 1d

🧵🚨

The UK’s independent scientific bodies are highly vulnerable to politicisation - over the past 5 months I've been working with @martinmckee.bsky.social to map out their vulnerabilities and it's not good news.

Today our report is published!
www.ucl.ac.uk/policy-lab/n...

1/11

UK’s arm’s length public bodies are highly vulnerable to politicisation

Seven in ten Britons say it is important for top scientific institutions to be independent in exclusive new polling.

www.ucl.ac.uk

20 390 590

Peter Tennant @pwgtennant.bsky.social · 1d

This is exactly the problem! A myopic definition of methodology and a complete ignorance of the need for, and benefits of, translation methodology and meta science.

Reposted by Peter Tennant

Peter Tennant @pwgtennant.bsky.social · 1d

What percentage of the health research funding pot do you think goes on methods and meta science?

Around 2% (if you add together all mention of research design and methodologies across all areas from research.hscni.net/sites/defaul...)

Just 2% spent on the part that underpins EVERYTHING we do.

Dr Will Ball @wpball.com · 1d

Reading some interest stats about health research funding in the UK in recent years.

- Only 7% goes to Mental Health research
- Only 5% of that MH funding goes to studies related to prevention

2 5 17

Reposted by Peter Tennant

Pausal Zivference @pausalz.bsky.social · 1d

I have an interesting case study on this actually. So at the beginning of the semester, I was preparing a bit on the (mis)use of LLMs for the course I co-teach

One of things I did was have it summarize one of my own papers, since people say "it's so good at it"
arxiv.org/abs/2503.02789

Accounting for Missing Data in Public Health Research Using a Synthesis of Statistical and Mathematical Models

Introduction: Missing data is a challenge to medical research. Accounting for missing data by imputing or weighting conditional on covariates relies on the variable with missingness being observed at ...

arxiv.org

2 2 5