Skip to main content


SYMP 4 – The Two Cultures of Statistics In Ecology: Prediction Vs. Explanation
Tuesday, August 7, 2012: 8:00 AM-11:30 AM
Ecologists often use “prediction” and “explanation” as near synonyms; after all, if one can explain the processes driving an ecological system one ought to be able to predict its future behavior, and vice versa. However, there is a surprising tension between these goals reflected across many different areas of science, most famously by Breiman (2001) in statistics and Peters (1991) in ecology. The predictive (algorithmic) approach, using statistical tools such as random forests, support vector machines, or classification and regression trees, emphasizes computational tractability and robustness; the explanatory (model-based) approach, using tools as such as maximum likelihood estimation and multilevel modeling, emphasizes interpretability and connection with ecological theory. With the emerging use of big ecological data sets from NEON, remote sensing, telemetry, genomics, and citizen science (among other sources), and with the ever-growing need for practical answers in conservation and management highlighted by the Meeting Theme, the pendulum seems to be swinging toward prediction. This symposium aims to bring together practitioners from across the spectrum of approaches to discuss how we can best achieve a fusion in ecological statistics that provides a good combination of tractability, robustness, interpretability, and relevance to ecological theory. Breiman, Leo. 2001. “Statistical Modeling: The Two Cultures.” Statistical Science 16 (3) (August): 199-215. Peters, Robert Henry. 1991. A Critique for Ecology. Cambridge University Press.
Organizer: Benjamin M. Bolker
Co-organizers: Andrew J. Tyre
Moderator: Benjamin Bolker
Endorsement: ESA Theoretical Ecology Section, Statistical Ecology Section
SYMP 4 Introductory remarks
SYMP 4-1 Exploratory analysis and inference with broad-scale citizen science data
Daniel Fink, Cornell University; Wesley M. Hochachka, Cornell University; Theodoros Damoulas, Cornell University; Jaimin Dave, Cornell University; Steve Kelling, Cornell Lab of Ornithology
SYMP 4-2 Predictive or explanatory? Is that the question?
Haiganoush K. Preisler, US Forest Service; David R. Brillinger, University of California Berkeley
SYMP 4-3 Hierarchical statistical models for ecological data: Combining explanation and prediction
Andrew M. Latimer, University of California Davis; Cory Merow, University of Connecticut; Adam M. Wilson, University of Connecticut
SYMP 4-4 Bridging the two cultures: Latent variable statistical modeling with boosted regression trees
Thomas G. Dietterich, Oregon State University; Rebecca A. Hutchinson, Oregon State University
SYMP 4-5 Physics envy vs. computer sciences envy: Shifting theoretical paradigms in ecology
Jorge Soberon, University of Kansas; Andres Lira-Noriega, University of Kansas; Narayani Barve, University of Kansas; A. Townsend Peterson, University of Kansas
SYMP 4-6 An analysis of approaches to presence only data
Trevor Hastie, Stanford University; Will Fithian, Stanford University
SYMP 4 Discussion