The elephant's trunk
In desperation I asked Fermi whether he was not impressed by the agreement between our calculated numbers and his measured numbers. He replied, “How many arbitrary parameters did you use for your calculations?” I thought for a moment about our cut-off procedures and said, “Four.” He said, “I remember my friend Johnny von Neumann used to say, with four parameters I can fit an elephant, and with five I can make him wiggle his trunk.”
It came to mind when I read this story in the NYT with the introductory claim What really stimulates economic growth is whether you believe in an afterlife — especially hell.The report is of some estimations done by Rachel M. McCleary and Robert J. Barro (the story notes that the two are married) published in American Sociological Review.
Barro is probably the biggest name in the field of cross-country growth regressions (a field in which I’ve also dabbled), and I’m sure he’s aware that thousands of these regressions have been run and that, with very limited exceptions, results that particular factors are conducive to growth have proved highly fragile. I haven’t read the paper, so for all I know, the results have been checked for robustness in every possible way. But my eyebrows went up when I saw this para
Oddly enough, the research also showed that at a certain point, increases in church, mosque and synagogue attendance tended to depress economic growth. Mr. Barro, a renowned economist, and Ms. McCleary, a lecturer in Harvard’s government department, theorized that larger attendance figures could mean that religious institutions were using up a disproportionate share of resources.
What this means is that at least two parameters have been used in fitting growth to religiosity and that the two have opposite signs – most likely it’s some sort of quadratic. In my experience, there’s always at least one arbitrary choice made in the pretesting of these models (for example once you have a quadratic, the scaling of variables becomes critical). That gives three free parameters, if not more.
I’m not John von Neumann, but with two parameters I can fit a dromedary and with three I can do a Bactrian camel.