Cameron Patrick
@cameronpat.bsky.social
biostatistician @ University of Melbourne Statistical Consulting Centre. enjoyer of multiple imputation, RCTs, and DAGs. always graph your data. also runs, bikes, hikes, etc. he/him #BiInSci 🏳️🌈
https://cameronpatrick.com/
https://cameronpatrick.com/
Reposted by Cameron Patrick
*cries in medical statistician*
Sometimes I hear, "we need to frame our research as associational otherwise the MDs reading it will not interpret it correctly."
It took me a while to realize how bonkers this is. It's saying some important primary users of the knowledge created are unable to understand what they're reading.
It took me a while to realize how bonkers this is. It's saying some important primary users of the knowledge created are unable to understand what they're reading.
February 6, 2026 at 11:55 AM
*cries in medical statistician*
Reposted by Cameron Patrick
Many years I ago I read (sorry) that when Queensland mandated helmets, cycle use halved. It saved 2 or 3 TBIs a year and probably caused 100x that loss of QALYs (or whatever the measure is) due to increased use of sedentary modes of transport.
February 6, 2026 at 10:16 AM
Many years I ago I read (sorry) that when Queensland mandated helmets, cycle use halved. It saved 2 or 3 TBIs a year and probably caused 100x that loss of QALYs (or whatever the measure is) due to increased use of sedentary modes of transport.
Reposted by Cameron Patrick
An example of how impacts of interventions can differ at individual and population levels, and are conditional on other events. Given a car hits you, the helmet helps. But mandating helmets could lower the # of bikes on the road & thus might increase the chance of getting hit in the first place.
This is counter-intuitive:
"It is clear that the benefits of cycling outweigh the risks, with helmet legislation actually costing society more from lost health gains than saved from injury prevention." theconversation.com/ditching-bik...
"It is clear that the benefits of cycling outweigh the risks, with helmet legislation actually costing society more from lost health gains than saved from injury prevention." theconversation.com/ditching-bik...
Ditching bike helmets laws better for health
With epidemics of diabetes and obesity threatening to bankrupt state health budgets, governments need to broaden their strategies to encourage physical activity. Allowing cyclists to ride without a he...
theconversation.com
February 5, 2026 at 2:55 PM
An example of how impacts of interventions can differ at individual and population levels, and are conditional on other events. Given a car hits you, the helmet helps. But mandating helmets could lower the # of bikes on the road & thus might increase the chance of getting hit in the first place.
Reposted by Cameron Patrick
Is AI killing publishing reform? Important post by Ulrike Hahn about #scholarlyPublishing. (write.as/ulrikehahn/i...)
I've had the same thought, that the flood of LLM-written junk and fraudulent manuscripts being submitted to journals means that
I've had the same thought, that the flood of LLM-written junk and fraudulent manuscripts being submitted to journals means that
Is AI killing scientific reform?
Recently I tried to post a pre-print on arXiv about what might be going wrong in debate about reasoning in LLMs. arXiv seemed a relevant ...
write.as
February 6, 2026 at 9:51 AM
Is AI killing publishing reform? Important post by Ulrike Hahn about #scholarlyPublishing. (write.as/ulrikehahn/i...)
I've had the same thought, that the flood of LLM-written junk and fraudulent manuscripts being submitted to journals means that
I've had the same thought, that the flood of LLM-written junk and fraudulent manuscripts being submitted to journals means that
Reposted by Cameron Patrick
IMO it's hard to find any part of the broader left as utterly discredited by the events of the last two decades as the anti-trade, anti-globalisation set. Just fundamentally wrong about absolutely everything, and in a way that opened the door wide for the far right to come charging through.
It was missing the point to suggest the objection to globalisation was about things like jobs or wages or living standards which is why the anti globalisation people talked exclusively about those things and got them entirely wrong
February 5, 2026 at 9:19 PM
IMO it's hard to find any part of the broader left as utterly discredited by the events of the last two decades as the anti-trade, anti-globalisation set. Just fundamentally wrong about absolutely everything, and in a way that opened the door wide for the far right to come charging through.
big oof!
Newsletter up. Rant incoming.
Pop sci mostly reaches educated, left-leaning, wealthy readers — including, of course, scientists and science writers.
I increasingly believe the gap between the real general audience and what's called "general audience" in scicomm is huge and worth talking about: 🧪
Pop sci mostly reaches educated, left-leaning, wealthy readers — including, of course, scientists and science writers.
I increasingly believe the gap between the real general audience and what's called "general audience" in scicomm is huge and worth talking about: 🧪
Pop science doesn't reach a "general audience"
Pop sci readers are wealthy, educated, and left-leaning
www.reviewertoo.com
February 6, 2026 at 6:51 AM
big oof!
Woke 2.0: this time we are not, in fact, joking
I think "woke" was at least partly identified with "irritating" and the way we're rectifying that in Woke 2 is by being scary instead
February 6, 2026 at 5:31 AM
Woke 2.0: this time we are not, in fact, joking
Reposted by Cameron Patrick
i swear to god i do not want to have to consider it possible that jeffrey epstein personally caused the 2008 financial crisis
February 6, 2026 at 1:35 AM
i swear to god i do not want to have to consider it possible that jeffrey epstein personally caused the 2008 financial crisis
Reposted by Cameron Patrick
getting permabanned from research dot gov for turning in a 5-page Background section explaining that the effect estimates from a one-way ANOVA are in fact a model and therefore an algorithm and therefore Artificial Intelligence
February 6, 2026 at 12:06 AM
getting permabanned from research dot gov for turning in a 5-page Background section explaining that the effect estimates from a one-way ANOVA are in fact a model and therefore an algorithm and therefore Artificial Intelligence
Reposted by Cameron Patrick
I feel like this is a great time to remind everyone that it's not even technically wrong to say that logistic regression is machine learning
Dawg are you fucking kidding me bsky.app/profile/horm...
February 6, 2026 at 12:02 AM
I feel like this is a great time to remind everyone that it's not even technically wrong to say that logistic regression is machine learning
Reposted by Cameron Patrick
#TodayinHistory #dataviz #Onthisday #OTD 📊
📅Feb 5, 1772 Pierre Simon Laplace presented his first probability memoir to the Acadmie des Sciences
1810: Proof of the central limit theorem-- why sums of variables tend to be close to the normal
📅Feb 5, 1772 Pierre Simon Laplace presented his first probability memoir to the Acadmie des Sciences
1810: Proof of the central limit theorem-- why sums of variables tend to be close to the normal
February 6, 2026 at 2:13 AM
#TodayinHistory #dataviz #Onthisday #OTD 📊
📅Feb 5, 1772 Pierre Simon Laplace presented his first probability memoir to the Acadmie des Sciences
1810: Proof of the central limit theorem-- why sums of variables tend to be close to the normal
📅Feb 5, 1772 Pierre Simon Laplace presented his first probability memoir to the Acadmie des Sciences
1810: Proof of the central limit theorem-- why sums of variables tend to be close to the normal
who else remember the Australian federal election that was Hawke vs Peacock for PM?
February 5, 2026 at 1:44 AM
who else remember the Australian federal election that was Hawke vs Peacock for PM?
Reposted by Cameron Patrick
This is not new, Congressional reports have been floating this for at least the last year or two, but it continues to be a gobsmackingly hubristic and dangerous proposal by the U.S. which Australia should not accept and should push back strongly against www.theguardian.com/world/2026/f...
US congressional report explores option of not delivering any Aukus nuclear submarines to Australia
Report offers alternative of the US navy retaining boats and operating them out of Australian bases
www.theguardian.com
February 5, 2026 at 1:06 AM
This is not new, Congressional reports have been floating this for at least the last year or two, but it continues to be a gobsmackingly hubristic and dangerous proposal by the U.S. which Australia should not accept and should push back strongly against www.theguardian.com/world/2026/f...
RETVRN
February 4, 2026 at 1:55 PM
RETVRN
Reposted by Cameron Patrick
the villains in neuromancer being a family of gross inbred rich weirdos who want to live forever in space feels a little too on the nose
February 2, 2026 at 2:29 PM
the villains in neuromancer being a family of gross inbred rich weirdos who want to live forever in space feels a little too on the nose
possibly my most single-dude trait is that my dining table is covered in ... stuff ... and I eat all my meals from the coffee table instead
February 4, 2026 at 7:25 AM
possibly my most single-dude trait is that my dining table is covered in ... stuff ... and I eat all my meals from the coffee table instead
this is how I was taught fixed and random effects and I think it still holds up well in practice
Gelman's cryptic definition #2 inspired me to look up Searle, Casella, and McCulloch, which to me at least provides some useful terminological context:
February 4, 2026 at 6:17 AM
this is how I was taught fixed and random effects and I think it still holds up well in practice
Reposted by Cameron Patrick
There's more than one way to do it: Perl, R, Stan
There's exactly one way to do it: Python, BUGS
There's at most one way to do it: SPSS, SAS
There's exactly one way to do it: Python, BUGS
There's at most one way to do it: SPSS, SAS
February 4, 2026 at 5:21 AM
There's more than one way to do it: Perl, R, Stan
There's exactly one way to do it: Python, BUGS
There's at most one way to do it: SPSS, SAS
There's exactly one way to do it: Python, BUGS
There's at most one way to do it: SPSS, SAS
starting to think that R is 21st century Perl:
- there's more than one way to do it (tm)
- surprisingly useful for random data manipulation tasks
- eternal flamewars with Python fans
- there's more than one way to do it (tm)
- surprisingly useful for random data manipulation tasks
- eternal flamewars with Python fans
Are you a #Stata user? Maybe you work with one?
Have you ever found yourself copy-pasting from the results window?
It's annoying as hell! And terrible practice. So I wrote a blog post on using #rstats to extract results from Stata log files
benharrap.com/post/2026-02...
Have you ever found yourself copy-pasting from the results window?
It's annoying as hell! And terrible practice. So I wrote a blog post on using #rstats to extract results from Stata log files
benharrap.com/post/2026-02...
Using R to extract results from Stata log files – Ben Harrap
benharrap.com
February 4, 2026 at 5:09 AM
starting to think that R is 21st century Perl:
- there's more than one way to do it (tm)
- surprisingly useful for random data manipulation tasks
- eternal flamewars with Python fans
- there's more than one way to do it (tm)
- surprisingly useful for random data manipulation tasks
- eternal flamewars with Python fans
if you understand this reference, it is time to log off and touch grass
(this applies to me too)
(this applies to me too)
"Valorous Dumpster Fires" is the name of my new music act
February 4, 2026 at 4:54 AM
if you understand this reference, it is time to log off and touch grass
(this applies to me too)
(this applies to me too)
Reposted by Cameron Patrick
e-bikes are the single best consumer technology of the last 20 years and it's driving me mad that there's this sudden backlash against them
February 4, 2026 at 2:58 AM
e-bikes are the single best consumer technology of the last 20 years and it's driving me mad that there's this sudden backlash against them
who needs multiple imputation now, i guess
February 3, 2026 at 3:10 PM
who needs multiple imputation now, i guess
Reposted by Cameron Patrick
yeah for most of us absorbing the chaos void isn't very healthy. i'm fine obviously, i don't have a chaos void problem... but almost none of us should be doing this. anyone up for void summoning later?
February 3, 2026 at 3:47 AM
yeah for most of us absorbing the chaos void isn't very healthy. i'm fine obviously, i don't have a chaos void problem... but almost none of us should be doing this. anyone up for void summoning later?
Reposted by Cameron Patrick
📣 registration for the rainbowR conference is now live 🎉
The conference is for anyone interested in R and/or analysing data to understand LGBTQ+ issues. The majority of speakers and workshop leaders are LGBTQ+, but you do not have to be LGBTQ+ to attend.
pretix.eu/rainbowr/con...
#RStats #LGBTQ
The conference is for anyone interested in R and/or analysing data to understand LGBTQ+ issues. The majority of speakers and workshop leaders are LGBTQ+, but you do not have to be LGBTQ+ to attend.
pretix.eu/rainbowr/con...
#RStats #LGBTQ
rainbowR Conference 2026
Feb. 25 – 26, 2026
pretix.eu
February 2, 2026 at 2:45 PM
📣 registration for the rainbowR conference is now live 🎉
The conference is for anyone interested in R and/or analysing data to understand LGBTQ+ issues. The majority of speakers and workshop leaders are LGBTQ+, but you do not have to be LGBTQ+ to attend.
pretix.eu/rainbowr/con...
#RStats #LGBTQ
The conference is for anyone interested in R and/or analysing data to understand LGBTQ+ issues. The majority of speakers and workshop leaders are LGBTQ+, but you do not have to be LGBTQ+ to attend.
pretix.eu/rainbowr/con...
#RStats #LGBTQ
Reposted by Cameron Patrick
this is violence
I regret to inform you that “1979” by Smashing Pumpkins was released in 1996, which is now 30 years ago. Meaning a similar song released in 2026 would be called “2009”
February 3, 2026 at 12:04 AM
this is violence