The Dangers in False Assumptions

Roger Pielke, Jr. has an interesting article in the current issue of Bridges. He quotes himself:

“most preventable errors of policy analysis stem from the analyst’s perspective; as the analyst simplifies a problem to make it tractable for analysis and action, some important part of the relevant context is misconstrued or overlooked altogether.”

His examples are drawn from the field of climate change. He notes that most models predicting global warming assume that technological changes will reduce greenhouse gas emissions, and that the recent growth of energy demand in Asia has not seen the expected conservation technologies. He also notes a recent analysis that used fairly conservative estimates of the effect of global warming on hurricane generation to estimate the increased cost to insurers of global warming.

Comment: I remember a wise, perhaps apocryphal, story from my disertation adviser. He told me that a model of transportation needs for the Washington DC area had suggested that rapid transit was not as much needed as had been thought. It took a long time to discover that the model assumed much faster rush hour speeds on the city's highways that were reasonable. The offending parameter had been casually estimated by a couple of modelers, and never exposed to discussion or outside comments.

The example adds to those of Dr. Pielke, suggesting that assumptions of the modeler may be wrong and hidden. Good practice calls for sensitivity analysis of parameters and assumptions, to bring attention to those which are really important, and to allow them to be brought forth for further examination. JAD