Premium
Decision making in small groups: A simulation study
Author(s) -
Clarkson Geoffrey P. E.
Publication year - 1968
Publication title -
behavioral science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.371
H-Index - 45
eISSN - 1099-1743
pISSN - 0005-7940
DOI - 10.1002/bs.3830130404
Subject(s) - construct (python library) , psychology , task (project management) , adaptive behavior , test (biology) , decision theory , social psychology , cognitive psychology , group (periodic table) , computer science , artificial intelligence , statistics , mathematics , paleontology , chemistry , management , organic chemistry , economics , biology , programming language
Abstract The objective of this research has been to develop and test an information‐processing theory of group decision behavior. The individual was taken as the basic unit. It was hypothesized that if an adaptive model of individual behavior were developed, then a group's behavior would be a predictive result of the interaction of models of its participants. The theoretical scheme chosen was that of an information‐processing theory of individual decision behavior. To construct such a theory three important problems had to be resolved: 1) How to determine the decision processes of individual subjects; 2) How to identify die leader‐follower relationship and its effects on group decision processes; 3) How to generate the requisite empirical data to test the predictive ability of the proposed theory. The solutions were, first, to construct an experimental task which permitted both individuals and groups to generate observable, sequentially linked, task‐dependent behavior. Second, an adaptive information‐processing program was written which was capable of learning to behave like the observed behavior of individual subjects. Third, hypotheses were introduced concerning the leader‐follower relationship in two‐person groups so that interactions in dyads could be represented. The theory was tested by having subjects perform a number of trials, first as individuals and then as members of two‐person groups. During the group phase, subjects had to agree on what to do on each trial before proceeding to the next Data from the individual phase were used to build programs that mimicked the decision behavior of each subject. These programs were then employed as the basis of the group model, which predicted the observed behavior of each member as well as the resulting behavior of the group, trial by trial. The tests were successful in that the theory's performance was statistically superior to that of a number of alternative models based on data provided by 15 subject groups.