Hillel J. Einhorn

Order effects in belief updating: The belief-adjustment model

Abstract Much literature attests to the existence of order effects in the updating of beliefs. However, under what conditions do primacy, recency, or no order effects occur? This paper presents a theory of belief updating that explicitly accounts for order-effect phenomena as arising from the interaction of information-processing strategies and task characteristics. Key task variables identified are complexity of the stimuli, length of the series of evidence items, and response mode (Step-by-Step or End-of-Sequence). A general anchoring-and-adjustment model of belief updating is proposed. This has two forms depending on whether information is processed in a Step-by-Step or End-of-Sequence manner. In addition, the model specifies that evidence can be encoded in two ways, either as a deviation relative to the size of the preceding anchor or as positive or negative vis-a-vis the hypothesis under consideration. Whereas the former (labeled estimation mode) results in data consistent with averaging models of judgment, the latter (labeled evaluation mode) implies adding models. Conditions are specified under which (a) evidence is encoded in estimation or evaluation modes and (b) use is made of the Step-by-Step or End-of-Sequence processing strategies. The theory is shown both to account for much existing data and to make novel predictions for combinations of task characteristics where current data are sparse. Some of these predictions are examined and validated in a series of five experiments. Finally, both the theory and the experimental results are discussed with respect to the structure of models of updating processes, limitations and extensions of the present work, and the importance of developing a procedural theory of judgment.

Decision Making under Ambiguity

Abstract : Ellsbergs paradox demonstrates that ambiguous or vague probabilities derived from choices between gambles are not coherent. A descriptive model of judgement under ambiguity is developed in which an initial estimate serves as a starting point and adjustments are made for abbiguity. The adjustments involve a mental simulation in which higher and lower probabilities are considered and differentially weighted. Implications of this model include ambiguity avoidance and seeking; sub- and superadditivity of complementary probabilities; dynamic ambiguity; and reversals in the meaning of data. Three experiments involving Ellsbergs paradox and the setting of buying and selling prices for insurance and warranties test the model. A choice rule under ambiguity is developed that implies a lack of independence between ambiguous probabilities and the sign of payoff utility. The applicability of the model to the case where probabilities are explicitly stated is considered, including the handling of context effects. Keywords: Ambiguity, Decision making, Insurance.

Unit weighting schemes for decision making

Abstract The general problem of forming composite variables from components is prevalent in many types of research. A major aspect of this problem is the weighting of components. Assuming that composites are a linear function of their components, composites formed by using standard linear regression are compared to those formed by simple unit weighting schemes, i.e., where predictor variables are weighted by 1.0. The degree of similarity between the two composites, expressed as the minimum possible correlation between them, is derived. This minimum correlation is found to be an increasing function of the intercorrelation of the components and a decreasing function of the number of predictors. Moreover, the minimum is fairly high for most applied situations. The predictive ability of the two methods is compared. For predictive purposes, unit weighting is a viable alternative to standard regression methods because unit weights: (1) are not estimated from the data and therefore do not “consume” degrees of freedom; (2) are “estimated” without error (i.e., they have no standard errors); (3) cannot reverse the “true” relative weights of the variables. Predictive ability of the two methods is examined as a function of sample size and number of predictors. It is shown that unit weighting will be superior to regression in certain situations and not greatly inferior in others. Various implications for using unit weighting are discussed and applications to several decision making situations are illustrated.

Use of nonlinear, noncompensatory models as a function of task and amount of information

Abstract The present research was designed to assess the effect of two variables as they affect the use of nonlinear, noncompensatory models in decision making. These two variables were type of decision task and amount of information. The former variable was found to have a marked effect on the kind of combination model used by subjects, while the latter variable had a significant effect on the accuracy with which the various models could adequately represent the decision process. The major findings showed considerable use of the nonlinear, noncompensatory models with differential use in the two decision tasks used. The results for amount of information suggest that subjects use complex combinations of models (or compound models) in order to simplify the situation cognitively. In addition, the claim that decision makers do not use additive models in their strategies is given support by these data.

Expert measurement and mechanical combination

Abstract The expert can and should be used as a provider of input for a mechanical combining process since most studies show mechanical combination to be superior to clinical combination. However, even in expert measurement, the global judgment is itself a clinical combination of other judgmental components and as such it may not be as efficient as a mechanical combination of the components. The superiority of mechanically combining components as opposed to using the global judgment for predicting some external criterion is discussed. The use of components is extended to deal with multiple judges since specific judges may be differentially valid with respect to subsets of components for predicting the criterion. These ideas are illustrated by using the results of a study dealing with the prediction of survival on the basis of information contained in biopsies taken from patients having a certain type of cancer. Judgments were made by three highly trained pathologists. Implications and extensions for using expert measurement and mechanical combination are discussed.

ALCHEMY IN THE BEHAVIORAL SCIENCES

Access to powerful new computers has encouraged routine use of highly complex analytic techniques, often in the absence of any theory, hypotheses, or model to guide the researchers expectations of results. The author examines the potential of such techniques for generating spurious results, and urges that in exploratory work the outcome be subjected to a more rigorous criterion than the usual tests of statistical significance. Hillel Einhorn is Assistant Professor of Behavioral Science, Graduate School of Business, University of Chicago.

Motivation Theory and Job Design

One element in business life that persists as a management concern is the attention to productivity. Whether supervising an expansion of his firms activities, or cutting back on policies in response to sagging demand, a manager continually strives to draw a better return on his capital investment. When this effort is directed toward the human portion of the firms capital, the manager focuses upon a recurrent problem in business activity -employee motivation. At the blue-collar level, productivity difficulties have been attributed to the well-established practice of maximizing specialization of tasks, that is, attaining efficiency through the development of highly fractionalized, repetitious, and programmed jobs. Blue-collar blues has been the result of this form of job design-a dissatisfaction with work performed, and a justget-by cast to the consequent effort. Motivational problems in managerial positions have also drawn considerable attention. Intense competition for quality in manager talent may have effected a degree of uniformity in administrative ability across business firms. The simple model,

Cue definition and residual judgment

Abstract A major problem in naturalistic studies is the defining of cues. The problem of cue definition leads to a consideration of the residual variance in modelling the judge. The part of judgment that cannot be modelled (i.e., the residual) is assumed to consist of cues not adequately defined and mis-specification of the integration function. The major question asked is whether the residual portion of judgment has functional value, i.e., does it add to the clinicians predictive efficiency. Cue definition is shown to be related to the “bootstrapping” phenomenon. The conditions where bootstrapping will work are delineated. The issue is shown to depend on the correlation of the residual and the criterion. Furthermore, psychological achievement is shown to be a weighted sum of the modelled and residual parts of judgment where the weight parameters are interpreted as “environmental complexity.” Two naturalistic studies are examined to investigate the residual contribution to achievement. It is shown that the residual enters into achievement most prominently in situations where the model of man is poor. Moreover, for certain cases one gets a reverse bootstrapping, i.e., the residual correlates higher with the criterion than either the model of the man or the global judgment. Further results are discussed and their implications noted.

Decision Making Under Ambiguity: A Note

In 1961, Daniel Ellsberg used the following choice problem to demonstrate that expressions of uncertainty are not totally captured by the concept of “probability.” Imagine two urns, each containing red and black balls. Although it is known that Urn 1 contains 100 balls, the proportions of red and black balls are unknown. Urn 2, on the other hand, contains 50 red and 50 black balls. Furthermore, imagine that you can win

Multidimensional models for the evaluation of political candidates

100 by naming the appropriate color (red or black) of a ball to be drawn at random from Urn 1. Would you bet on red, black, or be indifferent between the two? What would your answer be if the drawing were to be made from Urn 2?