If you’ve ever attempted a meta-analysis, you’ll know that authors generally do a poor job reporting statistics. If you do this well, you’ll improve your chances of your work being included in a future meta-analysis.

I really really want a reporter to ask Bill Clinton about this and for him to say, verbatim, that he “did not have sexual relations with that man, Mr. Trump”

Um, hell yes! A company wants to sell you their output, but doesn't want to be responsible for that output? We'd laugh out loud if it was a physical product. So why should we accept it for a digital product? #AiIsTheProblem #TLSky #EduSky #SchoolLibrarians

coming to agree with Gen Z's lowered alcohol consumption being far better for health (my body does not bounce back like it used to) but at the same time feel like alcohol-fueled social spaces are broadly Good; certainly have been beneficial for my life and development anyway

if possible we should elect a lesbian. i would like there to be zero more dick-related scandals of any kind after this

pollen: I am not a fan. spring should have less pollen please.

I do love that Melbourne has developed a late night ice cream culture. want fancy gelato at 10pm? we got you! (maybe this is common elsewhere but it was certainly not something I grew up with!)

maybe this is a boomer take but not everything needs access to the internet. arguably most things are not improved by being online, including humans

But if you were moving into a new home and had to assemble a bunch of flatpack furniture back to back you’d be wishing for the screwdriver after the first or second one. Now imagine you have to assemble flatpacks 5 days a week for the rest of your career. It’s electric drill time!

I think that’s pretty good. One thing I would add is that the value of specialisation becomes more obvious at scale. Like if you have a choice of one tool for your home you choose swiss army knife as it will do a passable job at all the odd jobs...

that's not a premium feature though, you can text satan for free on the site formerly known as twitter

In 2017

h/T to @baddestmamajama.bsky.social who did this and really made me laugh: Olivia Nuzzi bio Mad Libs GO

a “Space Force”, if you will

Real talk: to the extent that academics engage in anything resembling either teaching or knowledge-building, they absolutely need people skills.

yes ahaha yessss

last night I dreamed I had a podcast, everyone be glad that that was not reality

This lines up with my experience. LLMs are like truly alien intelligence

My phone has full optical character recognition of 47,000 photos. I can search individual words. I cannot search three words in quotes. Computers used to be powerful. That power meant something. It was power for making your life better in sovereignty to your own interests. And now we have this.

“You have to see how the interests of capital steer it towards certain outcomes based on…“ Shut up, nerd. They all have meetings in Dr. Evil’s underground lair petting hairless cats and cackling.

Visualize your data.

Seconded: it's really hard to fetishise maths after having studied Linear Algebra in any real depth. Like, don't get me wrong, but the mindset you need to cultivate to get good at it is not one that's suited to success in the modern world.

I keep beating this drum: the kind of people one thinks of as Silicon Valley "techbro" are so often finance people with an engineering cosplay habit, not actual technologists Elon Musk was famously *never* a good coder and I can't believe he ever scienced a rocket either. But he craves that respect

A new 'left turn hold' at the start of the traffic signal phase to protect the busy bicycle movement towards Melbourne CBD on Rathdowne Street at Victoria Street.

#ViCBiostat Summer School 13-20 Feb 2026 #Causal inference #Estimands in trials #Cluster randomised trials Advanced #meta-analysis In Melbs and online. More details to come. www.vicbiostat.org.au

…𝐈𝐟 𝐇𝐓𝐄 𝐞𝐱𝐢𝐬𝐭𝐬, 𝗶𝘁 𝗰𝗮𝗻𝗻𝗼𝘁 𝗮𝗳𝗳𝗲𝗰𝘁 𝗮 𝗽𝘂𝗿𝗲 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗴𝗿𝗼𝘂𝗽 𝗯𝘂𝘁 𝘀𝗵𝗼𝘂𝗹𝗱 𝗶𝗻𝗰𝗿𝗲𝗮𝘀𝗲 𝘁𝗵𝗲 𝘃𝗮𝗿𝗶𝗮𝗻𝗰𝗲 𝗼𝗳 𝗬 𝗶𝗻 𝘁𝗵𝗲 𝘁𝗿𝗲𝗮𝘁𝗺𝗲𝗻𝘁 𝗴𝗿𝗼𝘂𝗽 𝗱𝘂𝗲 𝘁𝗼 𝗵𝗲𝘁𝗲𝗿𝗼𝗴𝗲𝗻𝗲𝗶𝘁𝘆 𝗼𝗳 𝘁𝗿𝗲𝗮𝘁𝗺𝗲𝗻𝘁 𝗲𝗳𝗳𝗲𝗰𝘁 𝗮𝗰𝗿𝗼𝘀𝘀 𝘁𝘆𝗽𝗲𝘀 𝗼𝗳 𝘀𝘂𝗯𝗷𝗲𝗰𝘁𝘀.” @f2harrell.bsky.social

“For continuous outcome variables Y where the variance of measurements can be disconnected from the mean, 𝒐𝒏𝒆 𝒘𝒂𝒚 𝒕𝒐 𝒆𝒔𝒕𝒊𝒎𝒂𝒕𝒆 𝒕𝒉𝒆 𝒎𝒂𝒈𝒏𝒊𝒕𝒖𝒅𝒆 𝒐𝒇 𝑯𝑻𝑬 𝒊𝒔 𝒕𝒐 𝒄𝒐𝒎𝒑𝒂𝒓𝒆 𝒕𝒉𝒆 𝒗𝒂𝒓𝒊𝒂𝒏𝒄𝒆 𝒊𝒏 𝒀 𝒊𝒏 𝒕𝒉𝒆 𝒂𝒄𝒕𝒊𝒗𝒆 𝒕𝒓𝒆𝒂𝒕𝒎𝒆𝒏𝒕 𝒈𝒓𝒐𝒖𝒑 𝒂𝒏𝒅 𝒕𝒉𝒂𝒕 𝒊𝒏 𝒕𝒉𝒆 𝒄𝒐𝒏𝒕𝒓𝒐𝒍 𝒈𝒓𝒐𝒖𝒑….

gradually reaching peace with the path of leaving PhD early (with Masters instead). not what I'd originally planned but have learned heaps, there are lots of successful applied statisticians w/o PhDs, and will mean I get to go back to earning actual money sooner

I am turning into my grandfather, example 23197: today I used ye olde Australian phrase "no flies on you" in an email

Any #rstats people on here have experience with hierarchal generalized additive models (HGAM) in ecology? I’m in need of some help in possibly using one with some data! 🦇🌎🧪🧫

not wrong though? (but what is this nightmare diagram)

if they do another Austin Powers in two years the same amount of time will have passed since the first one as between the first one and when he was frozen.

"Schrödinger's causal inference" (n): The practice of making causal claims or interpretations within a scientific article - typically in the title, abstract, implications, or conclusion - while simultaneously warning that the study design is unsuitable for causal inference.

my new AI powered lamp is trained on millions of lamp interactions so it understands that 50% of chain pulls are to turn it on and the other 50% are to turn it off. and honestly, it's a little spooky how it knows exactly what i want almost half of the time

how had I never heard this particular Musk Horror Story before

Dark Souls 2 pigs

This one has both:

worth reading even if you're not a programmer and don't know what a Rubygems is, the important bits are about community and power and brainworms and being a good person

Ah, the classics.

You categorised patients as responders or non responders by dichotomising a change from baseline? You triple criminal!

Because everything is shit, I give you ... Puggles. #Platypus #AustralianWildlife

Yeah :D

Approiximately my Belz Lecture in Melbourne from a couple of weeks ago (there may be a recording later?) notstatschat.rbind.io/2025/10/30/l...

the only instance in recorded history of a traffic collision between a car and a u-boat!

Yeah, ggview() looks exactly like what I need. It would save me constantly having to reopen files.

Rubin's gonna Rubin, but Robins Rubins harder.

virtue signalling is good actually

This, to me, is Blueskyism

someone call georg

sometimes i like to lecture kids about the lost discipline of human factors: 'they used to watch users interact with the program and then they'd *make the program work the way the users needed it to* instead of making users adapt to random, meaningless changes'