Aydin Örstan told “A tale of universal parsimony.” He ends his tale,

Mejzlik would have approved of Atkins’ method. However, although Mejzlik solved Divisek’s mystery, Čapek hints at the end of his tale that the duo hadn’t followed “proper scholarly methodology”. Perhaps, what they lacked was sufficient data. Then again, scientists trying to explain historical events rarely ever have enough information, but must often rely on statistical reasoning. It seems that we can not be sure of the causations of even the simplest happenings.

You can read Aydin’s post to find out who Mejzlik, Atkins and Čapek are and, more importantly, the place of parsimony in all research worthy of the name. But let me say two things about history and statistical reasoning.

First, the answers to all the abnormally interesting questions in history are “underdetermined; they “lack sufficient data.” “Statistical reasoning” provides a way to overcome this problem. This statistical reasoning may be quite formal but is more often informal. When we “weigh the evidence,” we are engaging in informal statistical reasoning. Such informal statistical reasoning is fraught with problems made worse by the fact that most historians have no formal training in the mathematics of statistics or probability. Without such formal training, historians lack intuitions that might better control what are, no matter how informally reasoned or stated, in the end matters of statistics and probability. I know I’m conflating statistics and probability but they are near relatives.

Second, on Sunday Shirley and I heard Jared Diamond pitch *Natural Experiments of History*, the new book he co-edited with James A. Robinson. In the course of his lecture, Diamond bemoaned the fact that historians are not taught statistics, that they even eschew it. Diamond is right and historians are wrong to avoid formal training in statistics and probability.

As I see you are dealing with statistical research: I have put one of the most comprehensive link lists for hundreds of thousands of statistical sources and indicators on my blog: Statistics Reference List. And what I find most fascinating is how data can be visualised nowadays with the graphical computing power of modern PCs, as in many of the dozens of examples in these Data Visualisation References. If you miss anything that I might be able to find for you or if you yourself want to share a resource, please leave a comment.

You may enjoy listening to the podcast interview with the physicist Sean Carroll here:

http://www.guardian.co.uk/science/blog/audio/2010/feb/15/science-weekly-extra-podcast-sean-carroll-time

He thinks the universe is deterministic; if we knew all the laws & the exact state of the universe at a given instant, we could predict the future with absolute certainty. That’s not a new idea, of course. But, I don’t know if we could also use that info to deduce past events with 100% certainty.

But, but … prob and stats was a required course when I went to college.

Aydin,

That discussion with Carroll is interesting. I’ve hear and read him say most of these things before. Even if the universe is indeed completely deterministic, and it were possible know how it works in every detail, I think there remains a problem, the problem of sensitivity to initial conditions. Assuming a set of processes that, when looked at in aggregate, appear to be stochastic, we might be able to explain each element in those processes one at a time. But could we do it with sufficient precision in every case to understand the aggregate behavior? While there are likely many cases were we could, I doubt that it is true in general. There will always be something for which we need greater precision than is achievable at the time. I think this is particularly a problem with phase changes. There might really be times when knowing pi to one more order of magnitude is required or when a butterfly fart in Brazil does cause a typhoon in Japan. That typhoon, that butterfly, can change history.

Rochelle,

It was a requirement for me too. I was an engineering major. In fact, engineering majors took more math units than did math majors. But except for whatever was offered in the required math survey course, humanists were not required to take prob and stats. The same was true when my kids when to school. You can get a PHD in history and philosophy at many (most?) major universities without knowing so much as the difference between the mean and the median. By the way, our son who teaches philosophy in top ten department thinks this is a crime. But no more than one or two of his colleagues would agree with him. Diamond used UCLA’s history department as a prime example for the lack of training in statistics among historians.