AN INVESTIGATION OF DECISION RULES AND INFORMATION PROCESSING STRATEGIES IN PREFERENTIAL CHOICE AMONG SINGLE- AND MULTI-ATTRIBUTE ALTERNATIVES

DOUGLAS EUGENE LANDON, Purdue University

Abstract

Human preferential choice was examined within a new theoretical framework. Previous analyses of choice have either concentrated on purely normative aspects of decision, such as utility models, or purely descriptive aspects of decision, such as rules and heuristics. The theoretical approach proposed here attempts to combine normative power with descriptive rules through an information processing framework based upon information acquisition and processing strategies at the level of the attribute. This framework is then used as a guide in the examination of preferential decisions. The major assumptions and proposals of the new theoretical framework are as follows. (1) Choice alternatives can be considered as sets of discrete attributes, each of which represents a specific alternative's value on a choice dimension. (2) Choice processes can be examined at the level of the attribute, rather than the level of the alternative. (3) Normative probabilities and other aspects of choice behavior can be empirically derived from large numbers of simple choice trials. (4) These normative aspects have meaning in testing descriptive rules and heuristics of choice behavior. Three subjects were required to make several hundred choice decisions using an information search and evaluation technique in simple multi-dimension, multi-alternative choice situations. The general results were: (1) The subjects produced easily identifiable information search and decision strategies across the several hundred choice trials. Two subjects used a conjunctive type of decision strategy and one used an elimination by aspects type of decision strategy. (2) Tests of the predictions of the normative utility theory models indicated that the utility models can provide good approximations of choice behavior at the level of the attribute. Quantitatively based analyses of attribute evaluation, determined by measuring latencies that occur during the information search procedure of each choice trial, were used to examine the information processing systems of the subjects at the level of attribute comparison within each choice dimension.

Degree

Ph.D.

Subject Area

Psychology|Experiments

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS