In the familiar concept of Batesian mimicry, an innocuous prey species avoids predation by adopting the aposematic traits of a well-defended "model" species. Such mimicry can be effective even when the ruse is not especially convincing.1
The field of ecology2 has developed its own form of Batesian mimicry in which analytical procedures bereft of inferential logic adopt the verbal patterns of scientific theory, thereby avoiding the depredations of editors and reviewers. This would appear to be an evolutionarily stable strategy, reliably enhancing the fitness of its practitioners3 even when the resemblance between the mimic and the model is superficial.4
The word hypothesis, in particular, is the verbal equivalent of black and yellow bands, signaling to readers that the claims to follow are not to be trifled with. Readers bold enough to continue their approach can expect to be met with elaborate warning displays: disorienting model descriptions, flashing information criteria, p-values that for all the world look like the eyes of a snake. Yet if one at last lays hold of the frantic creature — such as by requesting a reproducible analytical workflow — one discovers that the wasp was a hover fly all along. Or perhaps a male.
Lest this critique appear mean-spirited, I want to make clear that it is first of all autobiographical, a confession of my own blundering journey in scientific practice. The procedural problems of our field are so pervasive that none of us can claim innocence, yet at the same time none of us can be regarded as especially to blame. I am willing to believe, with Smaldino and McElreath5, that the miscarriage of scientific inference stems primarily not from conscious guile but from honest confusion and corrupt incentives. We want to do things right, but spanning the qualitative chasm between data and ideas is hard, and nowhere is this challenge more acute than in the bewildering complexity of ecological systems. Yet the core objectives of ecology, as indeed of any science — true understanding and wise intervention — hinge on principled inference, and we must hold ourselves to standards befitting the seriousness of our goals.6
My intent, moreover, is not a vain philippic against scientific decadence — that can wait for another day. Instead, my aim is a constructive demonstration that the disingenuous use of theoretical language is as unnecessary as it is insidious. Theory pervades even the most descriptive studies, and hypothesis-driven experiments unfold inexorably into unforeseen descriptions. There is no need to feign theory or apologize for description, for both these modalities of science are firmly grounded in the logic of causal inference. When a descriptive study is undertaken in the context of an explicit and defensible causal framework7, the result is what I term strong exploration.8
Finally, while it may at first seem tangential to the topic at hand, this essay is also about scientific writing. Writing and thought are mutually constituting; clarity in one yields clarity in the other, while wooliness in either is a failure of both. Similarly, the joy and ease of writing are conditional on clarity of thought, for there is nothing so tedious and demoralizing — and ultimately antithetical to the proper function of language — as to construct from confused thoughts a veneer of words designed to conceal rather than to express. In science, disciplined causal thinking facilitates sound conceptualization, fluid organization, and lucid expression.
To be continued…
Edmunds M (2000) Why are there good and poor mimics? Biological Journal of the Linnean Society. Linnean Society of London, 70, 459–466.
This may be equally true of other fields, but I restrict my comments to what I know.
Let me clarify that this statement is directed at the practice, not the practitioner. I’m not taking shots at successful academics. That should be obvious to anyone in the sciences who knows as well as I do that successful academics are not the ones doing data analyses and writing papers. Science is done by tenuously employed students and postdocs, people who must act in the interest of their professional survival if they are to continue doing science at all. My critique is aimed at the system that teaches and compels them to resort, often unknowingly, to the abuse of theoretical language.
Smaldino PE, McElreath R (2016) The natural selection of bad science. Royal Society Open Science, 3, 160384.
Ibid.
Christakos G (2010) Integrative Problem-Solving in a Time of Decadence. Springer Science & Business Media.
Note that by “causal” I do not necessarily mean mechanistic. Causal inference is not a mode of reductionism; it is simply the use of logic to keep our storytelling in check. I’ll have more to say about this in future posts.
This term is gratefully borrowed from Platt’s (1964) classic: Platt JR (1964) Strong Inference: Certain systematic methods of scientific thinking may produce much more rapid progress than others. Science, 146, 347–353.