The Spectator article on alcohol is annoying. The data are genuinely hard to interpret since the studies are all observational and abstainers are very different than occasional drinkers. For example, they are more likely to be religious. They are simply very different populations so you have to proceed with extreme caution.
Also, heavy drinking is obviously terrible. In light of this, epidemiologists are justifiably hesitant to recommend moderate drinking to abstainers, or vice-versa.
If the article had the same sloppy, unbalanced reasoning about computer science, I don’t think you would’ve linked to it.
Which studies “controlled for population” and what do you mean by that?
The reason I am skeptical is that there are a potentially infinite number of unknown factors. You might control for income, but there could be cultural, genetic, environmental, etc factors. Here is evidence for a gene that increases drinking and reduces heart disease, for example: https://www.ncbi.nlm.nih.gov/pubmed/25011450
These objections could be overcome with a super strong signal. For example, we are pretty certain that smoking causes cancer and heart disease even w/o a randomized trial, because the magnitude is so large.
But any heart benefits from alcohol are (as far as I know) modest enough that it is getting into “fake news” territory to suggest that public health professionals are “suppressing” this message. And ok, that’s not your view, you just link to it, but I would also express caution about “alcohol probably won’t kill you”. Globally you can say only that it is 94% likely that you won’t die from alcohol, since 6% of deaths are attributed to alcohol. http://www.who.int/substance_abuse/publications/global_alcohol_report/msb_gsr_2014_1.pdf
I’m probably going on excessively about this – you’re just linking to interesting articles without digging in deeply. But this sort of sensationalist contrarianism is in my view a bad mental habit.
Sorry, my summary of the genetics article was backwards – the allele reduces drinking. But still the point is that these unobserved variables can bias the measured effect in either positive or negative directions.
That’s the problem. We just don’t know very well whether moderate drinking is helpful or harmful. We know that the effect is likely small on most people, however.
The Spectator article on alcohol is annoying. The data are genuinely hard to interpret since the studies are all observational and abstainers are very different than occasional drinkers. For example, they are more likely to be religious. They are simply very different populations so you have to proceed with extreme caution.
Also, heavy drinking is obviously terrible. In light of this, epidemiologists are justifiably hesitant to recommend moderate drinking to abstainers, or vice-versa.
If the article had the same sloppy, unbalanced reasoning about computer science, I don’t think you would’ve linked to it.
I expressed my own point of view which differs from the article. I don’t recommend drinking for health purposes.
Some studies did control the population so they not all as bad as you may think.
Which studies “controlled for population” and what do you mean by that?
The reason I am skeptical is that there are a potentially infinite number of unknown factors. You might control for income, but there could be cultural, genetic, environmental, etc factors. Here is evidence for a gene that increases drinking and reduces heart disease, for example:
https://www.ncbi.nlm.nih.gov/pubmed/25011450
These objections could be overcome with a super strong signal. For example, we are pretty certain that smoking causes cancer and heart disease even w/o a randomized trial, because the magnitude is so large.
But any heart benefits from alcohol are (as far as I know) modest enough that it is getting into “fake news” territory to suggest that public health professionals are “suppressing” this message. And ok, that’s not your view, you just link to it, but I would also express caution about “alcohol probably won’t kill you”. Globally you can say only that it is 94% likely that you won’t die from alcohol, since 6% of deaths are attributed to alcohol.
http://www.who.int/substance_abuse/publications/global_alcohol_report/msb_gsr_2014_1.pdf
I’m probably going on excessively about this – you’re just linking to interesting articles without digging in deeply. But this sort of sensationalist contrarianism is in my view a bad mental habit.
Sorry, my summary of the genetics article was backwards – the allele reduces drinking. But still the point is that these unobserved variables can bias the measured effect in either positive or negative directions.
That’s the problem. We just don’t know very well whether moderate drinking is helpful or harmful. We know that the effect is likely small on most people, however.