Cameron R. Peterson

credible interval temperature forecasts1

Abstract An experiment was conducted in which forecasters expressed temperature forecasts in terms of intervals of variable width and fixed probability. The use of such intervals, called credible intervals, permits forecasters to describe the uncertainty inherent in their temperature forecasts in a meaningful, quantitative way. The results of the experiment indicate that forecasters can use credible intervals to quantify this information, information which may be important to potential users of these forecasts. Several recommendations are made regarding credible interval temperature forecasting on an operational basis.

Multiple probability-learning with shifting weights of cues.

Mans adjustment to an uncertain environment has been thoroughly studied in the classical experiment on probability-learning, in which the stimulus-situation is characterized by some degree of randomness but is stationary through trials. Uncertainty is associated with the event that occurs on any specific trial, but the probability of which event will occur remains constant over trials.1 Many uncertain situations in the real world are not only partially random, but are also nonstationary. Characteristics of such situations are not stable but change over time. A deviation at any time may be either a random fluctuation or the reflection of a genuine change in the situation. Adjustment to a nonstationary, uncertain environment requires that S detect whether a deviant event is a random fluctuation or the signal of a real change. Responses should ignore random fluctuations, but should react to genuine changes. In the classical experiment on probability-learning, the probability of response stabilizes at, or somewhat above, the stationary probability of the corresponding event. When the probability of an event shifts as a function of trials, a corresponding shift occurs in the probability of the response.2 The present experiment is in the context of multiple probability-learning. Whereas previous experiments have compared the degree to which cue-weights (dependence upon cues) for responses correspond with