I point this out because of the depressing regularity with which I see a raw correlation presented as though it represents a linked cause/effect relationship without the source meeting his scientific burden of evidence.
Science isn't based on hand-waving -- falsifiable empirical evidence is required. Any number of assumptions like this have fallen by the wayside over the decades, undermined by overlooked correlations that weren't included in the original analysis.
The scientific bottom line is easy to state -- is the assumption falsifiable using empirical evidence? If this condition cannot be met, it's not science.
Related: https://youtu.be/b240PGCMwV0?t=37
> Some studies will never be run because they would be deeply unethical ...
That cannot be offered as an argument to justify dispensing with scientific discipline. It's as simple as that.
> ... so we satisfy significance through a preponderance of correlation ...
Yes, but that is not science, it's politics or numerology. Using the method you just described, I can prove that there's a connection -- a "link" -- between sour cream sales and motorbike accidents:
http://www.dailymail.co.uk/sciencetech/article-2640550/Does-...
But, as is often said, correlation is not causation.
Does this mean I can claim there's no connection (or a poor connection) between childhood asthma and air pollution? No, for the same reason that you cannot say there is -- it's not scientific.
I ask that you think about your position. If we allow public policy to be guided by the kind of pseudoscience you're advocating, if we dispense with the falsifiability criterion, then religious beliefs can substitute for science, and Creationism can be taught as science in public-school science classrooms.
Consider cigarettes --- has there been a 40-year controlled study on smoking? Yet, at some point, the observational evidence starts to pile up and one has to make decisions based on that.
Certainly -- I can name a dozen, each of them demonstrated using scientific methods. Asbestos. Ionizing radiation. Sunlight. Tobacco. And so forth. Each of these can be shown to cause health issues in vitro as well as in vivo -- using properly designed, scientific animal studies.
> So we're always at the mercy of spurious correlations.
Not for a scientifically educated population, which -- if it existed -- would reject this sort of pseudoscience.
A recent example of a shaky correlation that will doubtless shape public policy is this one:
Link: https://news.northwestern.edu/stories/2017/01/shootings-us-s...
Title: "Shootings in U.S. schools are linked to increased unemployment"
Quote: "A rigorous Northwestern University study of a quarter-century of data has found that economic insecurity is related to the rate of gun violence at K-12 and postsecondary schools in the United States. When it becomes more difficult for people coming out of school to find jobs, the rate of gun violence at schools increases." (Emphasis added.)
Note the terms/phrases "linked", "is related to" and the tendentious final sentence that tries to nail a nonexistent cause-effect relationship.
Obviously the original study cannot support this conclusion -- it reports a correlation, not a link, not a cause followed by an effect.
This is a practice called "data mining" -- underemployed researchers look for correlations between things, then publish papers describing the correlations, usually but not always with appropriate disclaimers ("Correlation is not causation, much more research is needed"), but irresponsible journalists then turn correlations into cause-effect relationships, as in this case.
In popular articles like this one, the word "linked" is often used to magically turn a correlation into a cause-effect relationship, because it means two things are bound together and move as one. Coincidences become pseudoscience.
> Consider cigarettes ...
Not a good example for your viewpoint. Rigorous, double-blind falsifiable animal studies demonstrate the carcinogenic properties of cigarettes.
> Yet, at some point, the observational evidence starts to pile up and one has to make decisions based on that.
Are you arguing that we should do without science? Without the benefit of science, the U.S. concluded that Eastern Europeans were less intelligent than others, and a resulting changed immigration policy condemned a huge number of Jews and others to death during the Nazi era. Later, more objective intelligence tests came to the opposite conclusion, and Mr. Yerkes, the responsible psychologist, wrote a confessional mea culpa that nearly no one read. This is just one example, there are many.
When we do without science, we should at least acknowledge that we're doing without science, and show an appropriate amount of skepticism about the conclusions we draw.
Right, but then you're still assuming that whatever you've demonstrated on animals will hold on a human population. If that were always true, we would never bother doing studies on humans at all.
Science isn't based on hand-waving -- falsifiable empirical evidence is required. Any number of assumptions like this have fallen by the wayside over the decades, undermined by overlooked correlations that weren't included in the original analysis.
You have to make assumptions to reach a conclusion. "Humans are similar to animals" is one, and "There is no other cause that can reasonably explain this observational evidence" is another. I fully agree that observational data is often abused to reach sensational conclusions, but IMO when it is done correctly, it is a vital ingredient of modern science. Pearl's book on causality is basically devoted to this topic.
Science isn't about assumptions, it's about evidence. People make assumptions every day, that's unavoidable, but we should be alert to the fact that it's not science, and be appropriately skeptical. We need to be able to separate the wheat from the chaff. That's all I'm saying.
Apropos: https://youtu.be/b240PGCMwV0?t=37
Yes, but the point I have been trying to make is that, when we don't do science, we need to say we haven't done the science. Too much public information is pseudoscience without anyone saying to.
> ... but in the mean time -- Kids will stop getting asthma from car exhaust.
I understand that you're not getting this, but any number of times people have gone down the wrong path, let astray by pseudoscience, and calling it science all the while. As it happens, asthma is as likely to be caused by an overly clean environment as a polluted one -- observations support either conclusion.
"Is Cleanliness Among the Causes of Allergies?" : http://www.everydayhealth.com/allergies/cleaning-and-allergi...
Again, we cannot draw a conclusion without doing some science. My only point in bringing this to your attention is so you realize this issue is by no means settled, and no issue can be settled without science.
Environmental pollution is a serious problem, one I happen to think should be aggressively addressed. But if we offer a bogus reason, opponents in industry will use that bogus reason to fight a needed change.