Also this week: Ada Lovelace’s bicentenary, philosophy of science humor, and more.
Andrew Hendry on ecological and evolutionary research topics about which lots of people are enthusiastic, but for which the data are mostly noise, with the true signal or effect size being very weak or small. Fluctuating asymmetry is one (in)famous example. Andrew’s suggestions:
- genome scans to identify genes for some trait of interest (another infamous example, I think, though it’s far from my expertise)
- parallel evolution (Andrew says it’s usually not very parallel)
- animal personalities and behavioral syndromes (rife with significant p-values associated with biologically-weak effects, sez Andrew)
- biodiversity and ecosystem function (this is definitely one you could argue about, since the R^2 for biodiversity may be low, but so is the R^2 for many other predictors of various ecosystem functions)
I can think of another example (see here and here). Under the surface here is an implicit assumption about what effects one ought to study. It would be interesting to make that explicit: what are the good, and bad, reasons why one might choose to study a “small” or “weak” effect? (And what do “small” and “weak” mean here, exactly? I suspect there’s no one-size-fits-all definition…)
The bicentenary of Ada Lovelace is coming up!
Two high-profile plant biologists working on plant root exudates, allelopathy, and rhizosphere interactions have been temporarily banned from receiving NSF funding for (in one case) falsifying and fabricating data and (in the other case) continuing to capitalize on the falsified data after learning they were false.
And finally, this week in philosophy of science humor: Popper, Kuhn, and Carnap play poker. Cameos from Freud and Marx.🙂