Inspired by this paper, in section "Is there a change in the preventive influence of ADHD medication on accident/injury risk over the lifespan in ADHD patients?"
Snippet:
ADHD medication thereby reduced the risk of transport accidents by about 58 % in both young and middle-aged males but did not reduce the risk for females in any age group (Chang et al., 2014).
Surprisingly, we still don't have good evidence about this. There's some disagreement about how long stimulant medication works, or if your body just gets used to it. Ideally we'd get a long-term RTC, or similarly-strong evidence.
Whenever such strong evidence is available, will it be found that long-term stimulant medication of ADHD lowers the accident hazard ratio by at least 1.2x, in men, even after 2 years of continued treatment?
Resolves whenever strong causal evidence is finally available, to my satisfaction. By the way I haven't fully read the papers, in fact, you could say the subject matter pertains to me.
1,000
1.00By the way I haven't fully read the papers, in fact, you could say the subject matter pertains to me
Scott aaronson (probably) wrote a blog post about how he reads papers that I can't find. In it, he pointed out that most of the contents of most papers are not particularly useful. You should basically (caricatured) read the title of 90% of papers you see, read the abstract of 90% of what remains, skim the contents of 90% of what remains, read in full specific sections of 90% of what remains, and only rarely read the 'entire paper'. Not 'fully reading the paper' isn't a symptom of ADHD, it's a symptom of not blindly following some vague socially-transmitted sense of diligence that's not useful at all :)
also: this market would be much easier to bet on if it didn't already build in some regression to the mean
@jacksonpolack Hmm, can we just say the evidence must control for that? Such as by comparing the long-term change to similar people who were randomly assigned not-medicated.
Agree that I probably shouldn't expect myself to read papers anyway. Guess I mentioned that since a better version of this market would be written by some familiar with how these things are usually measured, etc.
Hmm, can we just say the evidence must control for that? Such as by comparing the long-term change to similar people who were randomly assigned not-medicated
Oh whoops, that's not what I meant at all. I meant it'd be easier to bet NO on this if you used the original 58% reduction instead of 1.2x, as in regression to the mean of effect size of studies lol
@jacksonpolack Ah got it. Yeah, I tend to have a strong prior that effect sizes get smaller, when more powerful studies are done. So I deliberately baked in some regression of effect size :)