The following headline from Inside Higher Ed caught my eye:
British Hindus More Likely Than Christians to Enroll in Universities
Is this interesting? Perhaps to a latter day Tawney or Weber. If interesting, reverse the order of Hindu and Christian in the title. Is that interesting? What if Atheists had come out on top? Clearly, some category would have come out on top. An inquiry guaranteed to produce something interesting is uninteresting.
Along the same lines was a recent ESP study by Bem published in the Journal of Personality and Social Psychology. In one experiment he asked 1000 students to pick one of two curtains as the one they thought contained a picture behind it. Students chose the correct curtain 53.1 percent of the time. Suppose they had chosen the wrong curtain 53.1 percent of the time? This is just as informative as picking the correct curtain (being consistently wrong is just as good as being consistently right)…….also evidence of ESP. With a thousand students unless one landed pretty much spot on the 50% mark one is bound to get an interesting result.
Finally, a piece in the August 2nd, 2011 Science Times section of the New York Times, entitled Tracing Social Networks in Elephants. I don’t blame the scientist quoted but the writer of the piece. She put together an article with no content whatsoever (perhaps this is how the Gray Lady is cutting costs?) A scientist followed elephants around noting who they associated with. Here are the three quotes the intrepid reporter chose to relay to the reader.
1) There’s a lot of individual variation.
2) If you think about it, the amount of time you can devote to a friendship decreases with the number of friends you have.
3) It is not clear why the elephants choose to move from one social circle to another.
For this, neither the author of the piece or the scientist conducting the research needed to leave their armchairs. Makes one wonder if the study of social networks has reached the bubble stage.
3 comments
August 2, 2011 at 8:15 am
John
While the Bem results are completely flawed for other reasons, I would point out that 53.1% (either way) is interesting. If the experiments were done correctly (and they were not or the reproductions would have worked) on 1000 students you should by the law of large numbers ‘ landed pretty much spot on the 50% mark’. That is the point.
August 2, 2011 at 10:16 pm
Anonymous
I am confused about the main idea of this blog post (though I probably misunderstood the main idea).
Here is my impression of the main idea:
Case A: The argument for the existence of an optimal policy is trivial (or the likelihood of an one policy outperforming the other according to a given criteria is very high).
MAIN IDEA: Research trying to discover the particular optimal policy in Case A is not interesting.
I think that there are cases where people WOULD be interested in the particular optimal policy. For example, out of n possible treatments for a disease which treatment performs the best (or performs the best according to multiple criteria)? It seems likely that one treatment will outperform the others. If the criteria are respected, what is the problem with this research?
August 2, 2011 at 11:14 pm
Anonymous
Ah, should have been clearer. None of the studies I mentioned were concerned with finding an optimal policy or had a prediction in mind (unless of course the writers of the first intended to convert us all to Hinduism). They reported that if you divided the population into subgroups, some subgroup did better than others on one particular dimension. To illustrate, take a sample of male CEO’s and split them into two groups. One wears boxers and the other briefs (one would have to exclude religious groups that require underwear that are neither boxers or briefs). Pick your favorite measure of company performance. One can expect that one group (boxers?) will have higher average performance than the other group (briefs). No matter the outcome one can put together an amusing little piece on how choice of undergarment can influence company performance.
You could have spun the tale in terms of finding an optimal policy. One has a hypothesis. Perhaps, circulation to the parts of body that do not reach above the table when seated affect decision making. Different types of underwear affect circulation differently etc. If you predicted that boxers would outperform briefs and saw that in the data, that would be
interesting………….but only if you made the prediction BEFORE seeing the outcome. This would not be a proof. There could be another explanation. Drinking Heinekin, which refreshes parts of the body that other beers do not, may be the culprit. Perhaps boxers tend to drink Heineken but briefs don’t.