Of course he is. Here's a quote from the post to which I replied:
> There is inadequate evidence of a link between the hygiene hypothesis and autism ... the article is claiming a correlation, which is at least weak evidence.
But the above is false. A correlation is not "weak evidence" for a cause-effect relationship, or a "link", as the poster put it. We need to be perfectly clear that a "link" suggests a cause-effect relationship, but without specifying which way cause and effect run.
A correlation is an observation that can only lead to further work, and it is never, in and of itself, evidence of a cause-effect relationship, absent discovery of a mechanism.
This reminds me of one of my pet peeves about science journalism -- use of the word "link" to describe a correlation. It's tendentious and misleading. There are lots of studies that find "links" between utterly unrelated things, by simple data mining (searching for coincidental correlations devoid of meaning). Such meaningless correlations are often published as though they mean something, and they're nearly always described with the word "link" or another tendentious word.
Of course he is. Here's a quote from the post to which I replied:
> There is inadequate evidence of a link between the hygiene hypothesis and autism ... the article is claiming a correlation, which is at least weak evidence.
But the above is false. A correlation is not "weak evidence" for a cause-effect relationship, or a "link", as the poster put it. We need to be perfectly clear that a "link" suggests a cause-effect relationship, but without specifying which way cause and effect run.
A correlation is an observation that can only lead to further work, and it is never, in and of itself, evidence of a cause-effect relationship, absent discovery of a mechanism.
This reminds me of one of my pet peeves about science journalism -- use of the word "link" to describe a correlation. It's tendentious and misleading. There are lots of studies that find "links" between utterly unrelated things, by simple data mining (searching for coincidental correlations devoid of meaning). Such meaningless correlations are often published as though they mean something, and they're nearly always described with the word "link" or another tendentious word.
Intentionally humorous examples:
http://www.businessweek.com/magazine/correlation-or-causatio...
Apophenia (seeing meaningful patterns or connections in random or meaningless data):
http://en.wikipedia.org/wiki/Apophenia