Rules of Thumb for Judging Ecological Theories. Ginzburg, L. R. & Jensen, C. X. J. Trends in Ecology & Evolution, 19(3):121–126, March, 2004.
doi  abstract   bibtex   
An impressive fit to historical data suggests to biologists that a given ecological model is highly valid. Models often achieve this fit at the expense of exaggerated complexity that is not justified by empirical evidence. Because overfitted theories complement the traditional assumption that ecology is `messy', they generally remain unquestioned. Using predation theory as an example, we suggest that a fit-driven appraisal of model value is commonly misdirected; although fit to historical data can be important, the simplicity and generality of a theory - and thus its ecological value - are of comparable importance. In particular, we argue that theories whose complexity greatly exceeds the complexity of the problem that they address should be rejected. We suggest heuristics for distinguishing between valuable ecological theories and their overfitted brethren.
@article{ginzburgRulesThumbJudging2004,
  title = {Rules of Thumb for Judging Ecological Theories},
  author = {Ginzburg, Lev R. and Jensen, Christopher X. J.},
  year = {2004},
  month = mar,
  volume = {19},
  pages = {121--126},
  issn = {0169-5347},
  doi = {10.1016/j.tree.2003.11.004},
  abstract = {An impressive fit to historical data suggests to biologists that a given ecological model is highly valid. Models often achieve this fit at the expense of exaggerated complexity that is not justified by empirical evidence. Because overfitted theories complement the traditional assumption that ecology is `messy', they generally remain unquestioned. Using predation theory as an example, we suggest that a fit-driven appraisal of model value is commonly misdirected; although fit to historical data can be important, the simplicity and generality of a theory - and thus its ecological value - are of comparable importance. In particular, we argue that theories whose complexity greatly exceeds the complexity of the problem that they address should be rejected. We suggest heuristics for distinguishing between valuable ecological theories and their overfitted brethren.},
  journal = {Trends in Ecology \& Evolution},
  keywords = {*imported-from-citeulike-INRMM,~INRMM-MiD:c-2399431,~to-add-doi-URL,100-ecology-articles,ecology,epistemology,heuristics,local-over-complication,modelling,modelling-uncertainty,overfitting,uncertainty},
  lccn = {INRMM-MiD:c-2399431},
  number = {3}
}
Downloads: 0