An experimental study of trust defined as the ability to predict
Date
1975
Authors
Goss, William Richard
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The evidence available in the literature concerning the importance of the concept of trust to the counseling relationship is unequivocal. To date, however, there has been little evidence to suggest that psychologists have adequately defined or understood the concept. The major purpose of this study was to determine if the concept of trust could be at least partially defined simply as the ability to predict. The general hypothesis of interest was that subjects exposed to higher levels of predictability should perform higher on traditional trust-type measures.
The study employed a single factor design with three levels and twelve dependent variables. A total of 45 female high school students were randomly assigned to one of three treatment conditions. The treatment conditions were: (a) 100% Predictable--the subject was always correct in her predictions about the other player's choice; (b) 50% Predictable--the subject was correct in her predictions only 50% of the time; and (c) 0% Predictable--the subject was ever correct in her predictions.
All subjects played 30 trials of a Prisoner's Dilemma game with a confederate of the experimenter. Subjects were required to rate their partner in the experiment on a number of traditional trust-type measures.
A multivariate analysis of variance was performed on the 12 dependent variables. The dependent variables were: (a) the cooperative choice in the Prisoner's Dilemma game; (b) a modified self-disclosure scale; (c) a 7-point bipolar scale concerning the subject's stated desire to continue playing the game with the same partner; and (d) a series of nine semantic differential-type scales.
The results showed significant differences in performance on the 12 dependent variables, collectively, between the three groups (p < .001). Three additional analyses were performed on the data comparing only two of the groups at a time. Results of these comparisons showed a significant difference between the 100% Predictable group and the 50% Predictable group (p < .001). The second analysis showed a significant difference between the 100% Predictable group and the 0% Predictable group (p < .001). A comparison of the 50% Predictable group with the 0% Predictable group did not show a significant difference (p = .079).
Implications of this study for counselors, teachers, and parents were discussed. The relationship of the findings presented here to a theory of behavior presented by Tolman (1932) was suggested, and areas for future research were presented.