Statisticians often like to think that non-statisticians don’t know what exactly it is that we do. The truth is of course that not only do they not know, they do not particularly care! With the possible exception of someone like Nat ‘2012 US election’ Silver, what statisticians are thought to do is about as exciting as driving around in cardigan and slippers in a two-tone ’74 Morris Marina with no radio.
But what if statisticians went around ripping up floorboards and counting up spiders? Now you’re talking!!
Back in ’46 a scientist named LC Cole published some data on counts of spiders, and sowbugs, (or woodlice, roly poly’s or slaters).
Cole and various bright sparks ever since, had the idea of fitting the spider / sowbug counts to various types of probability distribution. Voila!, it was found that spider counts could be quite happily fitted by the Poisson distribution, as can the number of typewriter errors made on a page, the number of people killed by horse kicks in the Prussian cavalry, etc etc.
But not sowbug counts, which are better fitted by a ‘contagious distribution’, such as the ‘generalized Poisson’ or ‘generalized Negative binomial’, in which the event of something happening is itself dependent on other events. Sowbugs, it seems are a social breed, and when they notice their numbers dwindling, to the point where there’s only one or two left, they pick up sticks and try the house down the road, in search of other sowbugs, if not adventure.
Spiders, on the other hand, are more individualistic or anti-social and don’t care if they’re left by themselves. (In fact they probably appreciate the peace and quiet after those pesky sowbugs have marched off elsewhere, unless of course the spiders belong to the type known as woodlouse spiders or sowbug hunters, which is a very different kettle of fish, or spiders, altogether, as are ‘shy spiders’and ‘social spiders’)
Finally, a paper published in the journal of the highly prestigious Royal Society in 2010 found that carnivorous wolf spiders (Lycosidae) and pink sundew plants (Drosera capillaris) competed with each other for available food, in statistically interesting ways, indeed the lead author described the study as ‘awfully fun’ http://www.livescience.com/8566-plant-spider-compete-food.html
So, next time someone asks (without really caring what the answer is) ‘just what is it that statisticians actually do.?….’.
[updated, 9 October 2016]
Cole LC (1946) A study of the cryptozoa of an Illinois woodland. Ecological Monographs, 16, 49-86.
Consul PC (1989) Generalized Poisson distributions. Marcel Dekker, New York.
Forbes C, Evans M et al (2011) Statistical distributions. 4th ed. Wiley, Hoboken, New Jersey.
Janardan KG et al. (1979). Biological applications of the Lagrangian Poisson distribution. BioScience, 29, 599-602.
Jennings DE et al. (2010). Evidence for competition between carnivorous plants and spiders. Proc Royal Society B, 277, 3001-3008.
Raja TA, Mir AH (2011). On applications of some probability distributions. Journal of Research & Development, 11, 107-116.