JOURNAL
OF
RESEARCH
IN
PERSONALITY
12, 189-192 (1978)
Unclogging the Bogus Pipeline: A Critical Reanalysis of the Cherry, Byrne, and Mitchell Study GERALD
G. GAES,
BARBARA QUIGLEY-FERNANDEZ, AND JAMES T. TEDESCHI
State
University
of New
York
at Albany
The assignment of subjects to conditions on the basis of social desirability scores could not have been done in the manner described by Cherry, Byrne, and Mitchell (Journal of Research in Personality, 1976, 10, 69-75). Even if subject assignment procedures were not challenged the data do not support the authors’ conclusions. Additional statistical analyses were carried out to test the hypotheses of the original study. It is concluded that either the demand cue manipulation was not successful or else paper-and-pencil and bogus pipeline measurement procedures are equally reliable measurement devices for the attitude similarity effect.
An increasing number of studies use the bogus pipeline as a technique to elicit “true” attitudes from subjects (Jones & Sigall, 1971; Gaes, Rivera, & Tedeschi, Note 1; Sigall & Page, 1971). The rationale behind this measurement procedure is that if subjects believe that the experimenter can detect when they are not telling the truth, they will be more frank in revealing socially undesirable attitudes. Presumably, in most circumstances it is better from the subjects’ point of view to be perceived as holding a somewhat undesirable attitude than it is to be perceived a liar. The bogus pipeline has recently come under attack by Ostrom (1973). who suggests that this procedure may not generate more accurate assessments of attitudes but may instead introduce other biases not present in conventional self-report techniques. Cherry, Byrne, and Mitchell (1976) indicate that there is no evidence that the bogus pipeline can overcome positivity biases, social desirability sets, and experimenter demands. Based on the assumption that the bogus pipeline should disinhibit subjects from a reluctance to use the negative ends of an affective scale, they predicted that they would obtain an interaction between attitude similarity and the mode of measuring interpersonal attraction. That is, one would Address reprint requests to James T. Tedeschi, Department of Psychology, State University of New York at Albany, 1400 Washington Avenue, Albany, NY 12222. 189 0092-6X6/78/0122-0189$02.00/0 Copyright @ I?78 by Academic Press, Inc. All rights of reproduction in any form reserved.
190
GAES,
QUIGLEY-FERNANDEZ,
AND
TEDESCHI
expect subjects in the bogus pipeline conditions to indicate more dislike for a dissimilar phantom other than would subjects in paper-and-pencil conditions. Presumably, this predicted difference would be greater in subjects who have a strong need for approval than those who have a weak need for approval. In addition an interaction was expected between demand cues and social desirability. Cherry et al. tested their hypotheses in a 3 x 2 x 2 design. Subjects were told either that similarity leads to attraction, opposites attract, or were not given a demand cue. They were then asked to rate a phantom other whose attitudes were either .24 or .75 similar with their own. The ratings were made on the paper-and-pencil version of the Interpersonal Judgment Scale (Byrne, 1971) or else were obtained by the bogus pipeline procedure. Postexperimentally subjects were given the Social Desirability Scale (Crowne & Marlowe, 1964). The social desirability scores were used to partition subjects and to form a four-factor analysis of variance. The results indicated a main effect for attitude similarity and a four-way interaction, which was apparently confined to a simple interaction showing that high social desirability subjects in the bogus pipeline conditions were more responsive to demand cues than low social desirability subjects. This effect was apparently due to a greater difference between high and low similarity subjects in the Similarity Demand conditions than those in the No Demand (control) conditions. Based on these results, Cherry et al. affirm the imperviousness of the Interpersonal Judgment Scale to demand cues and subjects’ concerns about social approval and suggest caution in using the bogus pipeline to uncover “true” attitudes. Close examination of the data reveals that the post hoc partitioning of subjects based on their social desirability scores is the key to the four-way and simple interactions reported. Unfortunately, it is impossible to determine how subjects were partitioned to conditions. Prior to testing subjects on the Marlowe-Crowne Scale, all were assigned to conditions in a between subjects design (12 cells). Sixteen subjects were assigned to each condition (N = 192). According to Cherry et al. “subgroups of high and low social desirability individuals were formed by dividing the sample at the mean of 12.40” (p. 72). Their Table 1 shows that this criterion allowed a post hoc distribution of subjects such that exactly half in each of the prior 12 experimental conditions were high and half were low in social desirability. The chances against such a distribution of social desirability scores are astronomical. In the discussion of their results Cherry et al. offer the suggestion that the statistical interaction obtained could be due to demand oriented subjects who were temporarily sensitized, and “hence obtained temporarily high scores on the Marlowe-Crowne” (p. 74). However, if this is a possibility, the implication is that the subjects in the bogus pipeline
UNCLOGGING
THE BOGUS PIPELINE
191
demand cue conditions showed enhanced social desirability scores. This points out the correlational nature of the partitioning. Any strong inference must be based on a preexperimental partition and random assignment procedure. ’ In order to clarify the effects of demand cues on subjects’ responses in the paper-and-pencil and bogus pipeline conditions we reanalyzed the results by collapsing across social desirability. To give the demand hypothesis a fair test we decomposed the three-way interaction into two contrasts. The first contrast compared the no demand groups with the demand groups; the second contrast compared the two demand groups against each other. Both contrasts were three-way interaction comparisons and neither was significant, F < I, and F (1,168) = 3.20, p > .05. The conclusion that must be reached is that the only significant finding of the study of Cherry et al. is the familiar one of the effects of attitude similarity on interpersonal attraction. The demand cues manipulated in the experiment were ineffective and did not bias the responses of subjects on either a paper-and-pencil or a bogus pipeline procedure for measuring attraction. It is not possible to determine whether the demand cue manipulation did not work for the range of attraction elicited in the experiment (mostly neutral to positive and no mean indicating strong negative attitudes) or whether both measuring procedures are impervious to demand cues and social approval needs. It is probably more productive to examine the validity of the bogus pipeline within the context of experiments designed to test competing theories. For example, Gaes, Rivera, and Tedeschi (Note I) argued that attitude change in the forced compliance situation is not a manifestation of dissonance reduction but instead is an impression management tactic used by subjects to project a positive and consistent identity to the experimenter. A dissonance effect was obtained in paper-and-pencil conditions where faking could be carried out by subjects without fear of detection but did not occur in bogus pipeline conditions where they believed their lies could be detected. Rivera and Tedeschi (1976) reasoned that people who gain a relative and inequitable advantage in the distribution of rewards in a group may express guilt and distress (Walster, Berscheid, & Walster, 1973) to show their commitment to group norms, but that they are secretly pleased that they gained a large reward. The degree of positive inequity was manipulated; measures of satisfaction and guilt were I A personal communication from the Editor of ./ound of Research in Prrsot~alir~ acknowledges that unequal numbers of subjects were distributed on a post hoc basis according to their social desirability scores. Despite this error of reporting, the original analyses were appropriately based on unequal ns. Although these corrections clarify how the social desirability partitioning was accomplished, the post hoc nature of the partitioning makes the interpretation of any finding relative to social desirability equivocal.
GAES, QUIGLEY-FERNANDEZ,
192
AND TEDESCHI
obtained from subjects either by paper-and-pencil instruments or by bogus pipeline procedures. It was found that in paper-and-pencil conditions subjects expressed greater distress and guilt the greater the inequity their advantaged rewards represented but expressed little guilt and greater satisfaction as the amount of reward increased in the bogus pipeline conditions. As these theory-relevant studies accumulate, the confidence in the widespread applicability of the bogus pipeline will increase. It would be too bad if the path to such theory-relevant research is clogged by doubts produced by misrepresentations such as those reported by Cherry et al. The fact that attitude similarity effects are produced by both measurement modes does not decrease one’s confidence in the bogus pipeline but merely indicates that the effect is reliable over different measurement techniques. REFERENCES Byrne, D. The attraction paradigm. New York: Academic Press, 1971. Cherry, F., Byrne, D., and Mitchell, H. E. Clogs in the bogus pipeline: Demand characteristics and social desirability. Journal of Research in Personality. 1976, 10, 69-75. Crowne, D. T., and Marlowe, D. The approval motive. New York: Wiley, 1964. Jones, E. E., and Sigall, H. The bogus pipeline: A new paradigm for measuring affect and attitude. Psychological Bulletin, 1971, 76, 349-364. Ostrom, T. M. The bogus pipeline: A new ignisfituus? Psychological Bulletin, 1973, 79, 252-259. Rivera, A. N., and Tedeschi, J. T. Public vs. private reactions to positive inequity. Journal
of Personality
and Social
Psychology,
1976, 34, 895-900.
Sigah, H., and Page, R. Current stereotypes: A little fading, a little faking. Journal of Personality and Social Psychology, 197 I, 18, 247-255. Walster, E., Berscheid, E., and Walster, G. W. New directions in equity research. Journal of Personality and Social Psychology. 1973, 25, 151-176.
REFERENCE NOTE I. Gaes, G. G., Rivera, A. N., and Tedeschi, J. T. Cognitive dissonance vs. impression mamagement: A study using rhe bogus pipeline. Unpublished manuscript, State University of New York at Albany, 1976.