BLACKSBURG, Va.--Ecologists and policy makers make recommendations and decisions with important, long-term social and environmental impacts based upon the results of research done in dynamic environments where findings have to be interpreted in the face of many variables, probabilities and unknowns. "And there are deep philosophical disagreements among ecologists and other scientists about how to quantify the uncertainty, and even about how to even interpret the evidential meaning of the data we collect, to determine what the data are saying, such as about ecological impacts of various technologies," says Virginia Tech philosophy professor Deborah Mayo.
At the 83rd annual Ecological Society of America meeting at the Baltimore Convention Center Aug. 2-6, Mayo will discuss her proposed framework for learning from evidence in the face of error. Her presentation is Monday, Aug. 3, at 3;10 p.m. as part of the "Nature of Scientific Evidence" symposium (1-5 p.m., room 316).
The symposium will include representatives from applied fields, statistics, and philosophy of science who will discuss and debate their different positions about the most fruitful and most reliable methods to use in making scientific inferences in ecology. "We will debate such issues as how to interpret statistical tests, how to get objective inferences or whether to rely on subjective assessments by experts, what is the best way to quantify uncertainty, and how should uncertain assessments be reported to policy makers or the general public," Mayo explains.
"But the issue is not just how much risk of error should we allow in making policy decisions? The issue is also the deeper one: How do we even evaluate and quantify the risk of error using the data we are able to collect?" she says.
"That is why the ecologists are not just grappling with policy issues, such as how conservative to be when one's data analysis will affect the environment, but they are also being drawn into philosophical issues about scientific data and the nature of scientific inference and scientific knowledge," Mayo says. "There are competing methods for interpreting data and each may yield very different assessments of what the risks are in the first place. That is why this session of ours at the ESA is so fascinating and is truly interdisciplinary. Philosophers of science whose business it is to understand and explain scientific data and uncertain inference are being called upon to reflect together with statisticians, biostatisticians, ecologists, and others to explore the most fundamental issues about how to interpret, quantify, and base decisions upon data."
Mayo's own "error statistical philosophy of evidence" is based on learning from mistakes by using familiar statistical hypotheses to recognize common types of experimental mistakes, then developing rules and techniques for uncovering and avoiding errors. "I propose reinterpreting standard statistical tests as tools for obtaining experimental knowledge," she says.
"I represent an orthodox approach with a lot of reinterpretation aimed deliberately at avoiding common misuses and criticisms." But the value of the symposium for her go beyond promoting her own philosophy, she adds. "My goal is to learn much more about the issues from the scientists at the conference."
For more information contact symposium organizer Nicholas Lewin, GIS Support and Research Facility, Iowa State University, nlewin@iastate.edu, 515-294-2279.