Dammit... Looks like I'm Iearning Stan.
10.03.2026 22:29
👍 1
🔁 0
💬 1
📌 0
Dammit... Looks like I'm Iearning Stan.
I've had something similar before where we are arguing past one another only for me to eventually realise that the real difference is completely discordant ontological assumptions that we are beginning from (usually a kind of realism about a concept vs a strong operationism).
But I'd suspect eventually that day a younger person continued training for life, the net effect of that training and ageing would lead to a decline but I'd predict likely slower than the effect of ageing alone.
Haha we were typing at the same time... No I agree the effect could be the same. Actually ageing is small enough if at all over say a typical 12 week intervention that we do see similar magnitude of effects in older untrained adults compared with younger untrained adults undergoing intervention.
Oh course... Given long enough I'd suspect that the intervention effect function isn't logarithmic conditional on exposure duration as ageing eventually starts to involve decline... Though it may still be the intervention effect is the same and just additive with the effect of ageing.
We've looked at longitudinal growth over multiple years in large datasets and found the 12 week predictions from a log function are very close to what we get from RCTs of 12 weeks. Also we've then used this function to make predictions of what would happen in a trained population and been bang on.
Hmm well from what I have seen the treatment effect isn't really conditional on age. But it is on exposure duration. So over a long enough period it's actually a roughly logarithmic function. So of course more trained folks have smaller treatment effect from continued training.
Yeah, this is both strength and hypertrophy operationalisations.
Forgot to add plot 🫠
Unless you happen to work in an area where they actually do happen to be unbiased estimators of the treatment effect and you can justify the assumption? 😉
Rare sure... but across large RCT datasets of resistance training Vs control, the pre-post SMC is a relatively unbiased estimator of the SMD 😆
Publishing in a journal means endorsing it.
Where you publish reflects your values.
Choose wisely.
doi.org/10.52057/erj...
She has to follow Da Rules!
In fact... IIRC Jorgen is canonically married to the Tooth Fairy 😅
Not all fairies...
I think there's a lot of interesting potential for application of psychometric models to problems in measurement of sporting abilities too www.researchgate.net/publication/...
I recall a blog post on Bayesian IRT for climbing ratings www.ethanrosenthal.com/2022/04/15/b...
You can keep supporting even now I've finished... Think of it as support for me to export all my data and do some analysis and and a write up for you nerds 😜
www.justgiving.com/page/james-s...
That's a wrap... Finished up with 495.1 miles covered according to my Apple Health app and with Snowdon today. The support from all quarters has been amazing and we've raised an incredible total raised (over £3500!).
And our current little girl Kiki who came to us a couple of years ago from Cats Protection too 😻
Bottom our little boy Cosmic who came from Cats Protection as a kitten but who sadly was not with us long though had a lasting impact.
Top left our first cat Bigsey who had a really rough time before he was rescued by my brother and then found his way to us.
Day 27... The penultimate day. Weather was pretty grim today but still got most of this done outside before the 400 mile drive to North Wales. Finishing off with Snowdon tomorrow and look at the amazing t shirt my wife made for me to wear!
Donate to support here: www.justgiving.com/page/james-s...
Day 26 down... Bit of a weird one today with some miles on treadmill and outside AM, treadmill at lunch, then outside evening. Trying to see if I can get to 500 miles Saturday... It's gonna be close. Also kitty spotted.
Donate here: www.justgiving.com/page/james-s...
Brits do like a challenge with a payoff... That's Charmander 🔥
This is the way it was always meant to be in a deterministic universe 😜
It's the stories we tell... and the friends we make along the way 😊
Rumination prompted by @dingdingpeng.the100.ci making me get all ontological with her opening from compass.onlinelibrary.wiley.com/doi/10.1111/... 😅
I wonder if there are strong Humeans in the "Causal Inference" world who prefer to call it "Constant Conjunction of Events Inference" behind closed doors 😂
In the end, we can estimate the average of all the (unobservable) individual-level treatment effects by simply comparing the (observable) average outcomes of the two randomly assigned groups. This is the magic of randomization (Collins et al., 2020). The main shortcoming in practice is that it does not always work quite like this. Sometimes, randomizing the treatment is possible but some participants do not follow the instructions or do not provide outcome data. Other times, we cannot directly manipulate the cause of interest (e.g., because it is a psychological variable, such as mood), and indirect manipulations (e.g., mood inductions) come with side effects (Eronen, 2020). At yet other times, manipulation is simply not an option because it is unethical (childhood maltreatment, exposure to environmental pollution…) or unfeasible (race, social class…)—but we may still be interested in these causes, even if they do not fit into a narrow framework of causality that demands manipulability (Krieger & Smith, 2016). In those situations, thinking about the whole causal net underlying the data can be helpful, which in turn is made easier by graphical notation.
This also contains the causal inference meta-take I’ve converged to: randomization definitely the closest thing to magic in causal inference, but it doesn’t always work out like we want it to, so you still gotta have that whole toolkit available to you.
Day 25 done... So close to the finish line. Stopped this evening for a quick outdoor workout with a view too.
Close to the target donations too... Help us get there!
www.justgiving.com/page/james-s...