Lloyd describes J. J. C. Smart’s opinion that biologists use statistics only to do significance testing on experimental data and not to extract trends and create generalizable models of underlying fundamental processes as scientists in other disciplines (notably physics and chemistry) do.
In my experience of learning statistics and talking with others about statistics, this is not an entirely unfair accusation. The only statistics I learned during my undergraduate studies in biology were related to basic significance testing (e.g. t-tests and chi-squared tests) and how to formulate hypotheses testable by these methods. It wasn’t until after completing my masters degree that I have started to learn in depth, through self-motivated study, about linear modeling, how that relates to principles of experimental design and power analysis, and how to generalize responsibly.
I am more and more of the opinion that a decent understanding at least of the logic underlying these methods is critical in doing experimental research, and that this kind of training is not sufficiently carried out during typical studies in biology.
Lloyd also describes (before arguing against) the rationalization put forth by Smart, as well as Sir Karl Popper and Thomas A. Goudge, that evolutionary biology is not real science because the data and contexts are all historical and therefore not replicated or reproducible.
To that, I simply say: experimental evolution.
(I acknowledge that after Dallinger’s 1880s experiment, experimental evolution took about a century to become a coherent practice. However, if those three wanted to make such sweeping claims about a field of study, it would have been nice for them to familiarize themselves with its entire potential first.)