But the path from "we found a correlation" to "we have good evidence of causation" is a very long path where 99% of hypotheses drop off. And this one is especia

But the path from "we found a correlation" to "we have good evidence of causation" is a very long path where 99% of hypotheses drop off. And this one is especially suspect. It's possible that early life happiness is protective against dementia but mid and late life less so, but our model of biology says that's less likely because almost all similar findings that we have found to be casual recency increases the effect. For instances what you eat and how you exercise this year impacts your current health more than what you ate and how you exercised 2/5/10/15/20 years ago. I think "may protect" is misleading because while it might be the case that if you increase someone's happiness in early adulthood it will decrease their risk of dementia, it's certainly not likely. For instance if you found some depression RCT from 50 years ago in 20 somethings, what are the chances the control group and the intervention group are going to have significantly different dementia outcomes? So while it's technically true that it "may protect against dementia", it was also technically true before someone did this study. And honestly studies like this have so many degrees of freedom they are almost always unintentionally p-hacked.