This proposal focuses on how weather forecasters make use of imperfect information in making predictions. The investigators were seen to provide an unusual and appropriate mix of skills needed to test several important theoretical questions about how experts assess the quality of information and important applied questions about how weather forecasts could be made more reliable. The project involves two experiments, using expert weather forecasters as the primary subject group, that will (a) compare the accuracy of forecasts made in limited vs. full information conditions and (b) test the consistency and reliability of information acquisition by examining how different forecasters make use of the same limited information base. The use of operating weather forecasters is viewed as particularly appropriate because, despite the fact that meterorologists often are viewed as the paragon of excellence in real-world probability judgments, their decision making processes and their ability to interpret and make inferences on the basis of incomplete information has not been thoroughly studied. The results will be general interest as a study of how experts make use of incomplete information and how human-computer interactions should be designed to provide assistance to decision makers facing complex forecasting demands under conditions of limited information.
