Senior Research Scientist US Geological Survey Lafayette, Louisiana, United States
Abstract: Ecology has recently witnessed an increase in papers describing techniques for causal inference adopted from econometrics, with many papers recommending quasi-experimental (QE) analysis for treatment/control and event/response studies. While QE methods are gaining a lot of attention, some ecologists point to a need to include analyses with expanded capabilities beyond the characterization of treatment/control differences. In this paper, we illustrate how the analysis of events can be expanded to address more complex questions and reveal underlying mechanisms using structural causal modeling.
In the presentation we first describe the philosophical underpinnings for QE methodology. We then illustrate QE analyses techniques for a study of wildfire impacts of forest dynamics. We then turn to structural modeling methods that relax the restrictions of reduced-form estimands typically associated with the QE approach so as to investigate responses in finer detail. Structural modeling results reveal the existence of a threshold response to fire severity where sites below the threshold recovered to pre-event conditions, but those above the threshold showed increasingly limited recovery. There is a preliminary indication from the fire ecology literature that threshold inflection points may be a general feature of high-severity wildfire effects, which implies the possibility of external validity for the relationship.
As ecologists consider the use of methods and practices adopted from the social sciences, there should be careful consideration given to the appropriate philosophical framework for ecological studies. We recommend ecologists consider the merits of a philosophical framework known as Evidential Pluralism that emphasizes the across-study ambition of building externally valid causal knowledge and that takes into account pre-existing causal information and the need to investigate mechanisms.
