stat counnnter

Wednesday, February 22, 2006

The Science Establishment lV

Statistics can be helpful. They have their place. They've even been known to point right at a cause which was later proven by lab tests. But what happens a lot today is a correlation will be found and then hailed as sufficient evidence to warrent government action.
In an ideal society, an informed public would read about the correlation and dismiss it as such knowing that more research is needed before becoming alarmed. But we don't live in such a society. Today, people can become alarmed over such correlations simply because they are referred to as links.

That is one of my biggest peeves with the statistics field, the use of the term "link." I want to urge all epidemiologists and media people to stop using the word link in statistical studies. "Association" and "correlation" are far more accurate and truthful. They convey the sense of a possible but not yet demonstrated, connection.

But the concept "link" projects an image of connectedness, like a link being connected to another link in a chain of causation. The concept "link" suggests causation. This should not be so.

I have some relatives who will not vaccinate their newborn for MMR because they heard that there was a "link" between mercury in vaccines and autism. There have been newspaper reports of other young couples doing the same thing. My relatives are looking for a clinic with vaccines minus the mercury. They are out there. But when I tried to explain that a link is not causation, my words fell on deaf ears. The damage had already been done. Done I think, by the word "link."

To me, this represents an example of intellectual sloppiness by statisticians and reporters and editors. If scientists become indifferent to the issue of truth or falsehood, they will also regard intellectual precision as a non-essential.

Another bit of intellectual sloppiness coming from the science establishment is the notion of truth by consensus. It seems rather obvious that the truth or falsehood of an idea is not to be determined by how many people support or oppose it; that the truth of an idea must stand or fall on its own merits. Yet it is truely amazing how many educated people in the media and in science espouse the truth by consensus doctrine. Of course it's just another "shortcut to truth."

The truth by consensus practice is most popular in the field of climate science. It's common to see scientists and reporters citing a consensus as evidence that global warming is all man's fault. At the website of the Science and Environmental Policy Project Benny Pieser has a good article titled "The Dangers of Consensus Science" which can be found here.

To be concluded

1 comment:

Myrhaf said...

Statistics show that people with gum disease have a higher risk of heart disease. Is that because the bacteria in the gums gets in the blood stream and damages the heart? Or is that people who take care of their gums are also more likely to take care of their cardio-vascular health? As you note, Mike, correlation is not a link.