ABSTRACT
Although psychologists have spent a good deal of time in documenting the fallibility of judgment, less attention has been devoted to the effects that fallible judgments have on decision making. In order to investigate this, the number and types of decision errors are considered within the following task: a population of applicants applies for some action (welfare, food stamps, medical care, etc.). Because the action cannot be given to all who apply, a judgment (x) must be made regarding the degree of "deservedness" of the applicants. On this basis, people are either accepted or rejected. At some later point in time, a criterion (y) is used to assess the accuracy of judgment and to define errors. In order to deal analytically with the number and types of errors, a simple probability model is developed. The model allows one to show the relationship between the number of acceptance errors (e.g., welfare cheaters, unnecessary surgeries, etc.) and rejection errors (people cheated out of welfare, etc.). The trade-off relationship between these errors is defined as is the percent of errors made relative to the total number of decisions. These variables are then shown as a function of the fallibility of judgment, the unconditional probability of acceptance, and the base rate, i.e., the unconditional probability of "true need." The results are discussed with respect to the following issues: (a) How valid must judgment be before errors can be reduced significantly? (b) How does the fallibility of judgment and the probability of acceptance affect the trade-off between errors? (c) What defines "good" as opposed to "poor" decision performance? (d) Is it "worth it" to try to increase judgmental accuracy? (e) How does the probability of acceptance reflect the differential costs of acceptance and rejection errors? (f) Is welfare cheating, unnecessary surgery, etc., necessary? (g) If one is truly deserving (or 143undeserving) how likely is it to be accepted? (h) How do outcomes based on decisions at one time affect the fate of programs over longer time periods?
