Thanks to an excellent comment by Morgana on my last post, I have managed to broaden somewhat my initially narrow view of the Biologies to consider the vast array of medicine-relate biology departments and programs. (Not medicine itself, mind you. There I know so little I dare not tread.) Thus, for the moment I'm going to stop talking about eco/evo and mole/cell.
Instead, to discuss statistics, which was the original aim of this essay, I will focus on a distinction that I am actually making up as I write(!). I think there may be two fundamentally different approaches to biological questions--or really, two different questions.
1. How does it work?
2. Why does it work?
This is reminiscent of a lesson commonly taught in introductory classes on proximate vs. ultimate causes. Suppose an octopus is caught and devoured by an eel. You ask, perhaps plaintively, "Why did the octopus die?" The proximate cause involves tissue damage, blood loss, and cessation of brain function due to the teeth and digestive juices of the eel. Ultimately, however, the octopus died because it was too slow to escape, because it received poor genetic material from its parents, because its parents, due to the stochasticity* of the environment, lived in a year when eels were scarce and thus they survived despite their slowness . . .
Obviously, the causality of any one event is almost infinitely zoomable in either direction. Nevertheless the set of causes can be divided into two fairly distinct categories. First is the proximate, the how, the mechanistic explanations: how strong must the eel's teeth be to tear the skin; how potent must the chemicals in its saliva and stomach be to lyse the octopus' cells; etc. Second is the ultimate, the why, the big-picture stuff: why the environment brings eels and octopuses together, why predation and escape responses have developed over evolutionary time, etc.
These two sets of questions, or answers, which clearly complement each other, can be pursued by two equally complementary sets of techniques. The techniques for answering mechanistic questions tend to involve laboratories and microscopes, dissecting tools and fluorescent tags**. The techniques for answering big-picture questions are usually field studies and computer models, scuba gear and Matlab. Ecologist count things. They count lots of things, and they do it many times. And then they do a whole lot of data processing.
Which bring me back to statistics, and the somewhat counterintuitive conclusion that, while the eco/evo crowd is stigmatized as pursuing a "soft" science, they present their data in terms of huge sample sizes and incredibly hardcore statistical analysis. Meanwhile, cellular papers can be published with key figures that are simply photographs of stained gels, described qualitatively in the text and used to defend or refute a hypothesis of the author's choice. This is not to say that such data is invalid, but it is rarely quantified and sometimes not even replicated, and is nevertheless published. Ecology papers can't get away with that--and I bet epidemiology ones can't either.
* One of my favorite words. It's right up there with viscoelastoplastic (sound it out slowly--isn't it delicious?), which is, for this summer at least, my absolute favorite word. To be discussed at a later date.
** This reminds me that I was going to make an (obscure and terrible) joke about FISH in the last entry, but didn't get around to it. FISH stands for Fluorescent In Situ Hybridization, which is a way to track genetic material in a cell by adding fluorescent material that can match up, or hybridize, with the stuff you want to look at. However, this lovely acronym could just as well (and more appropriately) describe what happens when two different species of GloFish get together and produce an equally fluorescent offspring.