I would like to comment on two issues raised in today’s very interesting discussion. The first is that of sensitivity analysis. Several comments today have referred to sensitivity analyses in the case studies, and it should be clarified that, in our case study on fuel efficiency, we did not report formal sensitivity analyses. The reason we did not was that, in some sense, we never were in a position where we thought we had a robust answer. And I wonder if this might be the common situation in complex ongoing studies. Pretty sensitivity analyses typically are possible only after one has completely solved the problem, and usually are possible only when there is an abundance of data. I’m not sure how often such a desirable situation actually occurs, especially with policy matters such as we discussed.
KeywordsFuel Efficiency Final Discussion Response Surface Design Bayesian Formalism Policy Matter
Unable to display preview. Download preview PDF.
- Braten, S.L. (1986) The third position — beyond artifical and autopoetic reduction. In Socio cybernetic Paradoxes, F. Geyer and J. van der Zouwen (eds.), Sage: London, pp. 193–205.Google Scholar