I don’t find these results to be of much value. There’s a long history of various sleep-duration correlations turning out to be confounds from various diseases and conditions (as your quote discusses), so there’s more than usual reason to minimize the possibility of causation, and if you do that, why would anyone care about the results? I don’t think a predictive relationship is much good for say retirement planning or diagnosing your health from your measured sleep. And on the other hand, there’s plenty of experimental studies on sleep deprivation, chronic or acute, affecting mental and physical health, which overrides these extremely dubious correlates. It’s not a fair fight.
Yes, my primary reason for posting these studies was actually to elicit a discussion about the kinds of conclusions we may or may not be entitled to draw from them (though I failed to make this clear in my original comment). I would like to have a better epistemic framework for drawing inferences from correlational studies, and it is unclear to me whether the sheer (apparent) poor track-record of correlational studies when assessed in light of subsequent experiments is enough to dismiss them altogether as sources of evidence for causal hypotheses. And if we do accept that sometimes correlational studies are evidentially causally relevant, can we identify an explicit set of conditions that need to obtain for that to be the case, or are these grounds so elusive that we can only rely on subjective judgment and intuition?
I don’t find these results to be of much value. There’s a long history of various sleep-duration correlations turning out to be confounds from various diseases and conditions (as your quote discusses), so there’s more than usual reason to minimize the possibility of causation, and if you do that, why would anyone care about the results? I don’t think a predictive relationship is much good for say retirement planning or diagnosing your health from your measured sleep. And on the other hand, there’s plenty of experimental studies on sleep deprivation, chronic or acute, affecting mental and physical health, which overrides these extremely dubious correlates. It’s not a fair fight.
Yes, my primary reason for posting these studies was actually to elicit a discussion about the kinds of conclusions we may or may not be entitled to draw from them (though I failed to make this clear in my original comment). I would like to have a better epistemic framework for drawing inferences from correlational studies, and it is unclear to me whether the sheer (apparent) poor track-record of correlational studies when assessed in light of subsequent experiments is enough to dismiss them altogether as sources of evidence for causal hypotheses. And if we do accept that sometimes correlational studies are evidentially causally relevant, can we identify an explicit set of conditions that need to obtain for that to be the case, or are these grounds so elusive that we can only rely on subjective judgment and intuition?