gwern comments on Causal Diagrams and Causal Models - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (274)
I deny that the study had people all "doing it right". In Eliezer's case, I gave him the benefit of the doubt that he was intelligent enough to avoid obvious confounders.
If someone gets sick (for example) towards the end of the study and then shows a "negative 8 percent " fitness level then their data is crap.
If the study did not control for intensity then it is crap.
The difference between someone actually doing an effortful workout and someone just being present at the gym for a period of time is astronomical, and an extremely common occurrence.
The study had an age range from 40 and 67...
This study is garbage.
And they could have been sick at the start, as well, producing pseudo gains... You're postulating things which you have no reason to think happened to explain things that did happen; nowhere is anything indicated about that and you are arguing solely that because you dislike the results, the researchers were incompetent.
Why should there be any control for intensity? They did an intervention; there should be a non-zero effect. If any level of exercise does not show any benefits, then you are wrong. And I guess you did not read the link, because several interventions were tested and did not show any difference in terms of exercise resistance.
So? Why do you think that exercise should be entirely ineffective in people age 67? Are 40yos from a different species where exercise does not work? By examining older people, who are much less fit and much more sedentary, shouldn't the effects be even more dramatic and visible?
So, in addition to "Individual responses to combined endurance and strength training in older adults", Karavirta 2011, let me also cite "Endurance training-induced changes in insulin sensitivity and gene expression", "Individual differences in response to regular physical activity", "Effects of Exercise Training on Glucose Homeostasis: The HERITAGE Family Study", "Adverse Metabolic Response to Regular Exercise: Is It a Rare or Common Occurrence?", "Genomic predictors of trainability", "Effects of gender, age, and fitness level on response of vo2max to training in 60–71 yr olds", "Resistance to exercise-induced weight loss: compensatory behavioral adaptations", and "Cardiovascular autonomic function correlates with the response to aerobic training in healthy sedentary subjects", to name a few. (One nice thing about HERITAGE and Bouchard's earlier studies is that they recorded exercise, so spare me the 'maybe they didn't actually exercise'.) In these, too, some people don't benefit from exercise and show individual differences in exercise trainability exist.
Epstein 2014, The Sports Gene, ch6 "Superbaby, Bully Whippets, and the Trainability of Muscle", pg68:
I don't even remember this conversation (4 years of necromancy?). I don't remember the context of our discussion, and it seems like I did a bad job of communicating whatever my original point was and over-exaggerated. I am pretty sure you have a better understanding of the data.
The context was whether exercise resistance was a thing that existed (and hence, whether it was something Eliezer could have). I was revisiting my old comments on the topic to grab the citations I had dug up as part of working on a section for my longevity cost-benefit analysis where I observe that given the phenomenon of exercise resistance, behavioral backlash like lowering basal activity levels, and twin studies indicating various exercise correlations are partially genetically confounded, we should be genuinely doubtful about how much exercise will help with non-athletic or cosmetic things and be demanding randomized trials.
First, we probably should be interested in the amount of total physical activity -- "exercise" implies additional activity besides the baseline and the baseline varies a LOT. Some people work as lumberjacks and some people only move between the couch and the fridge.
Second, as long, as we are expressing wishes about studies, I'd like those studies to focus on differences between groups of people (e.g. run some clustering) and not just smush everything together into overall averages.
Third, there is one more category besides longevity and (athletic and/or cosmetic) -- quality of life. Being fit noticeably improves it and being out of shape makes it worse.