Tim Morris
@timpmorris.bsky.social
Biostatistician working on methodology at Novartis. Simulation studies, non-inferiority, missing data, estimands, covariate adjustment…
He/him
https://tpmorris.substack.com/
He/him
https://tpmorris.substack.com/
Reposted by Tim Morris
Don't you f**king dare.
February 3, 2026 at 2:56 PM
Don't you f**king dare.
Reposted by Tim Morris
I've done an analysis where the outcome unit was in hours, and in the results we report decimal hours (e.g. 0.89 hours) rather than in HH:MM. A respected colleague thinks we should use the later. Who is more correct? True pedants only please.
*ahem* @timpmorris.bsky.social
*ahem* @timpmorris.bsky.social
February 3, 2026 at 3:56 PM
I've done an analysis where the outcome unit was in hours, and in the results we report decimal hours (e.g. 0.89 hours) rather than in HH:MM. A respected colleague thinks we should use the later. Who is more correct? True pedants only please.
*ahem* @timpmorris.bsky.social
*ahem* @timpmorris.bsky.social
Thank you to the company who just emailed this reassurance/encouragement 😘
February 3, 2026 at 4:49 PM
Thank you to the company who just emailed this reassurance/encouragement 😘
Please can someone create a bot that replies to all covid commentary papers with the title followed by ‘(needless to say, I had the last laugh)’ appended?
January 29, 2026 at 4:49 PM
Please can someone create a bot that replies to all covid commentary papers with the title followed by ‘(needless to say, I had the last laugh)’ appended?
Reposted by Tim Morris
Worth remembering when looking at the horrific scenes in the US right now that both Reform and the Tories have pledged to bring in ICE-style organisations to the UK if they win power
January 24, 2026 at 10:33 PM
Worth remembering when looking at the horrific scenes in the US right now that both Reform and the Tories have pledged to bring in ICE-style organisations to the UK if they win power
New paper:
‘A fully-integrated Bayesian approach for the imputation and analysis of derived outcome variables with missingness’
Harlan Campbell, me and Paul Gustafson
onlinelibrary.wiley.com/doi/full/10....
‘A fully-integrated Bayesian approach for the imputation and analysis of derived outcome variables with missingness’
Harlan Campbell, me and Paul Gustafson
onlinelibrary.wiley.com/doi/full/10....
A Fully‐Integrated Bayesian Approach for the Imputation and Analysis of Derived Outcome Variables With Missingness
Derived variables are variables that are constructed from one or more source variables through established mathematical operations or algorithms. For example, body mass index (BMI) is a derived varia....
onlinelibrary.wiley.com
January 22, 2026 at 9:35 PM
New paper:
‘A fully-integrated Bayesian approach for the imputation and analysis of derived outcome variables with missingness’
Harlan Campbell, me and Paul Gustafson
onlinelibrary.wiley.com/doi/full/10....
‘A fully-integrated Bayesian approach for the imputation and analysis of derived outcome variables with missingness’
Harlan Campbell, me and Paul Gustafson
onlinelibrary.wiley.com/doi/full/10....
Reposted by Tim Morris
Regretting this post already
Everyone please red this and then vigorously attack @timpmorris.bsky.social 😜
tpmorris.substack.com/p/how-to-rea...
tpmorris.substack.com/p/how-to-rea...
How to read ‘evidence pyramids’
To get past the pointless bit of the arguments
tpmorris.substack.com
January 16, 2026 at 12:50 PM
Regretting this post already
Commenting on a draft manuscript once again leads me to the conclusion that Ian White is a 🤬 genius
January 15, 2026 at 6:22 PM
Commenting on a draft manuscript once again leads me to the conclusion that Ian White is a 🤬 genius
You’re gonna love this one
'How to interpret hazard ratios', with @dominicmagirr.bsky.social and @timpmorris.bsky.social thestatsgeek.com/2026/01/15/h...
How to interpret hazard ratios
Survival analysis of time-to-event outcomes is very commonly performed using Cox’s famous proportional hazards model. The model estimates hazard ratios for the ‘effects’ of covari…
thestatsgeek.com
January 15, 2026 at 3:02 PM
You’re gonna love this one
My 11 year old has jumped quickly to the “humour” stage of grief
January 11, 2026 at 4:10 PM
My 11 year old has jumped quickly to the “humour” stage of grief
An asterisk in the text but no footnote* is one more annoying
There's nothing more annoying than an asterisked footnote at the end of a piece of writing, with no asterisk in the main text to show what it relates to.
* Like this one.
* Like this one.
January 9, 2026 at 8:06 PM
An asterisk in the text but no footnote* is one more annoying
New post on a topic you might have struggled with if you do simulation studies. For a given data-generating mechanism, you might wonder “What is the true value of my estimand?”
This post describes four ways to learn it.
tpmorris.substack.com/p/learning-t...
This post describes four ways to learn it.
tpmorris.substack.com/p/learning-t...
January 9, 2026 at 1:44 PM
New post on a topic you might have struggled with if you do simulation studies. For a given data-generating mechanism, you might wonder “What is the true value of my estimand?”
This post describes four ways to learn it.
tpmorris.substack.com/p/learning-t...
This post describes four ways to learn it.
tpmorris.substack.com/p/learning-t...
Lukas Widmer showed me this years ago:
set.seed(0)
x0 <- runif(5)
set.seed(1)
x1 <- runif(5)
x = cbind(x0,x1)
x
x0 x1
[1,] 0.8966972 0.2655087
[2,] 0.2655087 0.3721239
[3,] 0.3721239 0.5728534
[4,] 0.5728534 0.9082078
[5,] 0.9082078 0.2016819
x[2,1]-x[1,2]
0
Wild.
set.seed(0)
x0 <- runif(5)
set.seed(1)
x1 <- runif(5)
x = cbind(x0,x1)
x
x0 x1
[1,] 0.8966972 0.2655087
[2,] 0.2655087 0.3721239
[3,] 0.3721239 0.5728534
[4,] 0.5728534 0.9082078
[5,] 0.9082078 0.2016819
x[2,1]-x[1,2]
0
Wild.
January 8, 2026 at 10:50 AM
Lukas Widmer showed me this years ago:
set.seed(0)
x0 <- runif(5)
set.seed(1)
x1 <- runif(5)
x = cbind(x0,x1)
x
x0 x1
[1,] 0.8966972 0.2655087
[2,] 0.2655087 0.3721239
[3,] 0.3721239 0.5728534
[4,] 0.5728534 0.9082078
[5,] 0.9082078 0.2016819
x[2,1]-x[1,2]
0
Wild.
set.seed(0)
x0 <- runif(5)
set.seed(1)
x1 <- runif(5)
x = cbind(x0,x1)
x
x0 x1
[1,] 0.8966972 0.2655087
[2,] 0.2655087 0.3721239
[3,] 0.3721239 0.5728534
[4,] 0.5728534 0.9082078
[5,] 0.9082078 0.2016819
x[2,1]-x[1,2]
0
Wild.
Laptop but no screen
iNnOvAtiVe
iNnOvAtiVe
This keyboard is actually an entire computer. Just plug it into a monitor and you’re set. Pricing isn’t finalized yet for the HP Eliteboard G1A, but HP says the starting price should come in under $1,000.
January 8, 2026 at 7:11 AM
Laptop but no screen
iNnOvAtiVe
iNnOvAtiVe
The Matilda from the 90s and the more recent Matilda The Musical are about as good as each other
This place needs some Innocuous Discourse pronto. Quote this with a take that’s not political or aggressive
January 6, 2026 at 8:02 PM
The Matilda from the 90s and the more recent Matilda The Musical are about as good as each other
Complaining that someone’s blogs* are too long and suggesting the authors could use a good editor is an early contender for 2026.
*papers, in this case 7 pages
*papers, in this case 7 pages
January 2, 2026 at 12:25 PM
Complaining that someone’s blogs* are too long and suggesting the authors could use a good editor is an early contender for 2026.
*papers, in this case 7 pages
*papers, in this case 7 pages
NEJM: hold my beer 🥴
www.nejm.org/doi/full/10....
www.nejm.org/doi/full/10....
December 19, 2025 at 5:50 PM
NEJM: hold my beer 🥴
www.nejm.org/doi/full/10....
www.nejm.org/doi/full/10....
JAMA’s Statistics and Methods section must get so mad when BMJ Christmas papers get so much attention because they’re doing this level of comedy all year round!
December 19, 2025 at 1:57 PM
JAMA’s Statistics and Methods section must get so mad when BMJ Christmas papers get so much attention because they’re doing this level of comedy all year round!
Great to see this announced!!
🎉 Big news!
The MRC Centre for Clinical Research Efficiency (CoRE) is here.
By accelerating the testing of new medicines, the MRC CoRE will help bring better treatments to patients sooner.
🔗 Learn more: www.mrcctu.ucl.ac.uk/news/news-st...
The MRC Centre for Clinical Research Efficiency (CoRE) is here.
By accelerating the testing of new medicines, the MRC CoRE will help bring better treatments to patients sooner.
🔗 Learn more: www.mrcctu.ucl.ac.uk/news/news-st...
Groundbreaking £50m MRC Centre will overhaul clinical trials for patients | MRC Clinical Trials Unit at UCL
www.mrcctu.ucl.ac.uk
December 18, 2025 at 4:39 PM
Great to see this announced!!
One journal I sometimes submit to/publish in has a submission system of “email your paper and files to the editors, CCing coauthors”.
This morning, we submitted a revision. Within 30 mins, the editor replied that we had done what was asked; paper accepted.
Beats every other submission system.
This morning, we submitted a revision. Within 30 mins, the editor replied that we had done what was asked; paper accepted.
Beats every other submission system.
December 18, 2025 at 11:50 AM
One journal I sometimes submit to/publish in has a submission system of “email your paper and files to the editors, CCing coauthors”.
This morning, we submitted a revision. Within 30 mins, the editor replied that we had done what was asked; paper accepted.
Beats every other submission system.
This morning, we submitted a revision. Within 30 mins, the editor replied that we had done what was asked; paper accepted.
Beats every other submission system.
Posted this yesterday after being devastated to see one situation where I might be persuaded, by a collaborator who always makes me rethink strongly-held opinions – exhausting!
Once again: please use your brain when considering “meta-models” for simulation study performance and make the right decision to just not. Using one instantly gets you on my naughty list.
open.substack.com/pub/tpmorris...
open.substack.com/pub/tpmorris...
December 18, 2025 at 8:50 AM
Posted this yesterday after being devastated to see one situation where I might be persuaded, by a collaborator who always makes me rethink strongly-held opinions – exhausting!
Once again: please use your brain when considering “meta-models” for simulation study performance and make the right decision to just not. Using one instantly gets you on my naughty list.
open.substack.com/pub/tpmorris...
open.substack.com/pub/tpmorris...
December 17, 2025 at 4:37 PM
Once again: please use your brain when considering “meta-models” for simulation study performance and make the right decision to just not. Using one instantly gets you on my naughty list.
open.substack.com/pub/tpmorris...
open.substack.com/pub/tpmorris...
The opposite applies if you’re using a German word, because everyone loves learning and remembers German-word-for terms
People, short-hand terms are supposed to conjure some meaning for the thing they describe!
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.
December 10, 2025 at 11:18 AM
The opposite applies if you’re using a German word, because everyone loves learning and remembers German-word-for terms
People, short-hand terms are supposed to conjure some meaning for the thing they describe!
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.
December 10, 2025 at 9:51 AM
People, short-hand terms are supposed to conjure some meaning for the thing they describe!
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.
If your short-hand terms are Latin, you may as well be using A and B because, though equally non-descriptive, they are at least shorter.