Abstract
This paper examines the technique of decision analysis in the presence of an incompletely specified matrix of probabilities concerning the different possible outcomes of market research type investigations into the various states of nature being considered. The approach is based upon Fishburn's technique which exploits the existence of dominance among the various alternative courses of action open to the decision maker. It is also shown that, based upon a particular assumption regarding the structure of this matrix of probabilities, it is possible to determine maximum and minimum values of the "expected value of market research" without the need for specific point estimates of the relevant probabilities.