Premium
The computer as experimenter: New results
Author(s) -
Johnson Edward S.,
Baker Robert F.
Publication year - 1973
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.3830180506
Subject(s) - task (project management) , psychology , variance (accounting) , cognitive psychology , developmental psychology , social psychology , computer science , engineering , accounting , systems engineering , business
In order to investigate the effect of replacing a human experimenter by a computer in a concept formation task, 150 subjects were run in three conditions. One condition was run exclusively by a human experimenter, another was run exclusively by a computer, and a third was run using the experimenter solely as a teletype operator. The task, designed to be difficult for the college level subject, required that all except one of the 30 possible solutions be logically eliminated before the solution was acknowledged. Subjects run by a human performed at a slightly higher level than subjects run by computer. This effect, which occurred only in subjects of below average intellectual ability, was attributed to an experimenter cueing effect. Human administered subjects were also found to be significantly less variable regardless of intellectual level. This was partially attributed to a variance dampening effect exerted by human experimenters on their subjects and which is absent from computer administered procedures. One consequence of this effect is that performance in the computerized condition is more highly related to level of intellectual ability. It was concluded that the effects found in this study will probably occur in any computerized procedure characterized by high task difficulty, low time pressure, and wide response latitude.