Salience of Behavior and the Effects of Labeling

Peter H. Reingen, Arizona State University
William O. Bearden, University of South Carolina
ABSTRACT - A field experiment employing label versus no label and high initial request versus low initial request manipulations was conducted to further explore the hypothesis that labeling effects are successful only when the behavior is salient. Verbal compliance data regarding respondent willingness to participate in a subsequent market research survey support the earlier findings of Tybout and Yalch.
[ to cite ]:
Peter H. Reingen and William O. Bearden (1983) ,"Salience of Behavior and the Effects of Labeling", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 51-55.

Advances in Consumer Research Volume 10, 1983      Pages 51-55

SALIENCE OF BEHAVIOR AND THE EFFECTS OF LABELING

Peter H. Reingen, Arizona State University

William O. Bearden, University of South Carolina

ABSTRACT -

A field experiment employing label versus no label and high initial request versus low initial request manipulations was conducted to further explore the hypothesis that labeling effects are successful only when the behavior is salient. Verbal compliance data regarding respondent willingness to participate in a subsequent market research survey support the earlier findings of Tybout and Yalch.

INTRODUCTION

Self-perception theory (Bem 1972) has guided various inquiries regarding the effects of individuals' past behavior on their subsequent action. According to the theory, individuals acquire a positive attitude toward behavior which serves as a guide to their future actions to the extent that the initial behavior is attributed to internal causes. Two streams of inquiry have yielded evidence in support of the theory. One source of evidence arises from research entailing the foot-in-the-door paradigm in which compliance with a small request is solicited before subjects are asked a larger request (e.g., Pliner, Hart, Kohl and Saabi 1974; Reingen and Kernan 1977; Scott 1977; Snyder and Cunningham 1975). From d self-perception viewpoint, it is argued that compliance with a small request leads research participants to view themselves as the kind of people who perform such behaviors. These self-perceptions in turn produce greater compliance with a larger request than if the larger request had alone been presented. Another source of support comes from research utilizing label in as an influence strategy whereby individuals are characterized in the expectation that their subsequent behavior is consistent with the label (e.g., Kraut 1973; Swinyard and Ray 1977; Tybout and Yalch 1980). In self-perception terms, the labeling or people's behavior leads them to view themselves as the kind of people who perform such behaviors which produces subsequent label-consistent responses.

Although these studies support the self-perception position that the effect of past behavior on subsequent action is the product of belief inference processes, "unsuccessful" studies regarding the foot-in-the-door strategy (e.g., Cialdini and Ascani 1976) and the positive labeling technique (e.g., Steele 1975) suggest an under-specification of the theory. However, Tybout and Yalch (1980) have contributed to an understanding of the conditions under which self-perception predictions regarding the effects of D. St behavior will be obtained.

Drawing on research examining salience (Taylor and Fisk 1978) and individuals' use or schema-consistent and schema-inconsistent data in making self-judgments (Markus 1977), Tybout and Yalch suggest that a person's own behavior serves as a basis for making self-inferences which guide future behavior only to the extent that it is salient. The salience of a behavior is, in part, determined by its consistency with a person's self-schema. Generally speaking, self-schema refer to cognitive generalizations about the self which are derived from past experience and which affect the processing or self-related information.

Tybout and Yalch examined the salience argument by employing a labeling procedure in a voting context. Subjects were initially administered a questionnaire which measured their attitudes, self-perceptions and intention with respect to voting (i.e., self-schema). Then, the experimenter purportedly used the questionnaire to confer a label. Some subjects were told that their responses indicated that they were above-average citizens who were very likely to vote in an election (above-average label treatment) whereas others were cold that their data showed that they were average citizens who had an average likelihood of voting (average label treatment). The key dependent variables were subjects' subsequent self-perceptions of whether or not they were voters and whether or not they voted one week after the label was conferred and again eight months later.

Tybout and Yalch theorized that two factors would mediate the salience of the labeling cue; namely, its consistency with an individual's existing self-schema and the availability or other relevant cues. The above-average label was expected to be highly salient only for subjects with an existing self-schema of being a voter and for short-run behavior. Because the above-average label feedback would be consistent with their self-schema, the above-average label should produce an enhanced self-perception of being a voter and (shortrun) voter turnout for these subjects relative to average-label subjects with a voter self-schema. In contrast, for voters with a weak self-schema of being a voter, the above-average label would be unlikely to serve as a basis for self-perceptions and future behavior because of its inconsistency with existing self-schema. Thus, an absence of any effect of labeling on behavior and self-perception was predicted for these individuals. The findings were in substantial agreement with these expectations, especially the data for subjects with a strong voter self-schema.

In augmenting self-perception theory by the salience hypothesis, Tybout and Yalch have made an important contribution to the theory which had been silent on how the effects or past behavior are mediated by persons' initial dispositions or by the passage of time. However, as Tybout and Yalch point out (1980, p. 413), their procedure allowed self-selection to the independent variable of initial dispositions. Since the thrust of their theory testing depended greatly on this independent variable, this shortcoming becomes more critical. The present study attempted to eliminate the self-selection problem by experimentally manipulating initial behavior. Furthermore, a control group was incorporated into the design or the present study to assess the practical implications of labeling as an influence strategy. Since the focus of the Tybout and Yalch study was on the differential impact of two experimental labels, a no-label/no-initial contact control was not included in their design and this resulted in some uncertainty regarding the efficacy of labeling as an influence strategy.

OVERVIEW AND HYPOTHESES

We maintained Tybout and Yalch's central position that a person's own behavior serves as a basis for making self-inferences which guide future behavior only to the extent that it is salient. A field experiment was conducted in a context of interest to consumer researchers, namely one or securing respondent cooperation with a survey request. More specifically, subjects were initially asked to either perform a very small survey request or a large survey request. This served as the

Dependent Variables and Procedure

Subjects in the initial-request conditions were called again 6 to 8 days after the initial contact by a different experimenter who was blind to a subject's condition and who identified himself as representing a different survey organization. [Different experimenters and organizations were employed to rule out alternative explanations of the expected labeling effect. For instance,. had the experimenter administered both requests, a labeling effect could have been due to increased affect toward the experimenter rather that due to the label.] To minimize mortality bias, experimenters followed a three-callbacks procedure. Specifically, the experimenter told subjects in the experimental groups and the subjects in the no-initial-request control:

"Hello, is this Mr. (Mrs. or Miss) speaking? My name is ___ from the Market Research Corporation. We are conducting a survey on people's use of magazines. We're calling you today to find out whether you'd be willing to fill out a questionnaire for us. It's about seven pages long and we would like to send you the questionnaire for you to complete. Would you help us with this survey by filling it out?"

Subjects' agreement to the request constituted the verbal compliance measure. Within 24 hours of the telephone contact, subjects were sent a mailing which included a cover letter, questionnaire, and return envelope. The cover letter asked subjects to complete the questionnaire within 4 days. The questionnaire consisted of a mixture of closed-end/open-end questions and a variety of rating and ranking tasks involving magazines. The address of the fictitious survey firm was imprinted on the outgoing envelope and its stationery. complete with company logo, was employed to enhance realism. The number of completed questionnaires returned within 3 weeks of the administration of the mail-survey request constituted the behavioral compliance measure for all groups. [No questionnaires were received after this predetermined time interval.]

RESULTS

Internal Validity Checks

Preliminary chi-square analysis showed that subject attrition between the first and second request was unaffected by treatments (X2 = 3.53, d.f. = 3, n.s.) and experimenters (X2 = 1.97, d.f. = 4, n.s.); that verbal compliance with the second request did not differ significantly for days elapsed between the initial and subsequent request (X2 = .52, d.f. = 6, n.s.) and for experimenters (with p's ranging from .70 to .91 across conditions); and that compliance with the initial request did not vary significantly across treatments (X2 = 4.30, d.f. = 3, n.s.). Thus, the data appeared free of internal validity problems due to differential attrition races, time and experimenter effects on compliance rates, and varying initial compliance rates across treatments.

Compliance Rate Analyses

The Table summarizes the data of the impact of the label on verbal and behavioral compliance with the second request. For the verbal compliance data, the pattern of findings was consistent with the salience hypothesis as had been anticipated. For large-initial-request subjects, the label resulted in a significantly higher rate of positive behavioral intentions (61% vs. 42.1%; Z = 1.68, p < .05). [One-tailed tests of significance were employed when the specific direction of an effect was predicted a priori.] In contrast, no difference attributable to the label was observed between the one (initial) question groups (42.4% vs. 44.4%; Z = .17, n.s.). Thus, as far as verbal compliance is concerned, these findings are in agreement with the Tybout and Yalch position.

When the behavioral compliance data were examined, no significant labeling effects were found for the fifteen (initial) questions groups (20.5% vs. 18.4; Z = .93, n.s.) or the one (initial) question groups (15.6: vs. 72.2%; Z = .70, n.s.). This finding was contrary to expectations. The control group comparisons regarding the behavioral data yielded meaningful information, however. Tests on the proportions were significant (p < .05), with tie one (initial) question/label and control group comparison marginally significant (p < .10). From an applied viewpoint, these comparisons suggest that both verbal and behavioral compliance with a second request can be significantly enhanced by first compliance with a smaller request (a foot-in-the-door effect), but they do not imply a differential effect on behavioral compliance due to labeling.

DISCUSSION

The main finding that the salience hypothesis was supported only for the verbal compliance data deserves discussion. We suggest five explanations for this occurrence, none of which poses a significant challenge to the Tybout and Yalch salience hypothesis.

First, the salience of a label tends to be reduced with the passage of time (Tybout and Yalch 1980). The time interval between the administration of the label and the criterion variable of verbal compliance with the second request was typical of the studies which have shown a labeling effect (e.g., Kraut 1973; Tybout and Yalch 1980). However, several more days elapsed before subjects could actually perform the requested behavior. Although it is possible that this additional passage of time led to less retrievable or available labeling cue, thus perhaps resulting in the absence of a labeling effect on actual -behavior, it should be emphasized that the additional passage of time was of minor duration.

The second explanation also addresses the salience aspect of the label. It is conceivable that the salience of the label suffered from a weaR manipulation of initial behavior. We do not place much confidence, however, in this argument. The verbal compliance data would not have shown the labeling effect as predicted by the salience hypothesis had the manipulation been too weak.

A third alternative revolves around the personal nature of subsequent requests for compliance. The differential results between the verbal and behavioral compliance rates may in part be attributed to the greater willingness of respondents to refuse when not engaged in personal interactions with persons requestion compliance.

The fourth explanation centers on the size of the second request, which may have become more apparent only when the subjects were actually confronting the requested manipulation of initial behavior and was based on pretesting which suggested that virtually everyone would comply with the small rec lest but that hardly anyone would yield to the larger request when initially informed about the size-of-request involved. Compliance with the initial request was actually obtained for the vast majority of subjects in both groups. Subjects were then either labeled "helpful" or they were not labeled. About one week later, the subjects were contacted again and asked a second survey request. A fifth group or subjects received only the second request. Thus, a foot-in-the-door experiment was embedded in the overall design for the two unlabeled groups. The behavioral intention to perform the second request and actual behavioral compliance with the second request served as the dependent measures.

Initial compliance with a larger-request compared to initial compliance with a very small request should make a subject more aware that he/she is a "helper" (Seligman, Bush and Kirsch 1976). This is particularly likely in the case when a subject's attention has been directed to his/her behavior by a labeling message. A behavior may not be given much thought in the absence oz .1 label, suggesting the importance of labeling as a method to encourage individuals to use their past behavior to develop self-perceptions. Thus, the salience hypothesis predicted a greater rate of compliance with the second request for subjects in the large-initial-request/label group than for subjects in the large-initial-request/no label group. In contrast, for subjects who complied with the initial small request, no significant difference in compliance rates was expected between the labeled and nonlabeled subjects. It is doubtful that a label would increase the salience of behavior when initial compliance is very small because "everybody would have done this" [Kelly's (1973) discounting principle].

METHOD

Overview

A field experiment was conducted in which labeling and initial request size were manipulated in a 2x2 design via initial telephone contacts. A second request (to complete a 7-page mail questionnaire) was administered about one week after the initial contact for these groups plus a fifth (control group which received only the second request). Behavioral compliance (i.e., completion of the questionnaire) was assessed for all of the five groups.

Subjects and Experimenters

The subjects were 300 (male and female adult residents of (greater Columbia, S.C., randomly chosen from the telephone directory and randomly assigned to one of the five conditions with 50 subjects per condition. The experimenters, five undergraduate students, were blind as to the hypotheses, which was assessed in a post-experimental debriefing session. Experimenters followed a treatment schedule which listed the five treatments in random order and which was systematically varied across experimenters. Experimenters were thoroughly instructed in extensive training sessions. Great care was taken to ensure that the label would be administered in a sincere fashion

Independent Variables and Procedure

Initial-ReQuest Size. Subjects in the four initial contact conditions were called between the hours or 10 a.m. and 5 p.m. Initial subject behavior was manipulated by asking a subject to answer one or several (15) typical survey questions about soft drinks over the phone. Specifically, the experimenter said:

"Hello, is this Mr. (Mrs. or Miss) ___ speaking? My name is ___ from Survey Incorporated. We, are a professional marketing research firm and we are conducting a survey on the use of soft drinks. Would you mind answering a (several) question(s) for us?"

If a subject refused, the subject was thanked for listening and the contact was terminated. If a subject agreed, the experimenter proceeded with asking the question(s). This represented the measure of compliance with the initial request.

The levels for this independent variable were determined based on a pretest. Pretest subjects (n=30), who were similar to the subjects in the main experiment, were asked in telephone contacts the greatest number of questions (about soft drinks) that they would be willing to answer for a professional marketing research firm over the phone and how many questions they thought other people like themselves would be willing to answer. Pretest subjects indicated that on the average the greatest number of questions they would be willing to answer was about 10. (An insignificantly lower average was obtained for similar others.) At the 15-question level (mean plus 3 standard errors), 93% or the subjects indicated that they would refuse to comply with such a request, whereas 95% of the pretest subjects said that they would comply with a 1-question request. Thus, for a subject in the main study who agreed to the large-initial request, it can be inferred with a high degree of confidence that a "helpful" label would be accepted by them as an accurate characterization or initial behavior. Conversely, for subjects who agreed to the small-initial request, it is unlikely that the label would have its intended effect on enhancing the Subsequent behavior because nearly everyone would have performed the initial behavior. [To minimize non-compliance with the large-initial request, subjects in the 15-question groups were not informed about the request size at the outset. Rather, the experimenter asked whether a subJect would mind answering several questions and then proceeded with the interview. The "trick" was, of course, to obtain similar compliance frequencies in the various groups. Differential initial-request compliance rates would have had a confounding effect. In fact, initial-request compliance (see Table) did not vary as a function of condition in the main study (X) = 4.30, d.r. = 3, n.s.) and was uniformly high as was expected based on telephone survey experience. It should be emphasized that this was a desired outcome. What matters in the present study is whether subjects "thought" they performed a large favor, not whether their initial favor was de facto more substantial relative to others' favors because that data was not available to them.]

Label. After a subject had performed the request, the experimenter said: "Thank you very much for answering this (these 15) question(s) for us." For subjects in the label conditions, the experimenter added: "You are a very helpful person and I wish more or the people we call were as helpful as you are." The label was similar to the ones employed in Previous studies (e.v.. Kraut 1973).

TABLE

EFFECTS OF LABELS

behavior. The second request may have been too large (as is suggested by the extremely low behavioral compliance race in the control group) for labeling to have had a differential impact on behavior. [In a previous study (Reingen and Kernan 1979) which employed a similar subject population, a 6-page questionnaire produced a 41% compliance rate in a control group. This high response suggested a ceiling effect in that study. To prevent a ceiling effect in the present study, we made the second request more demanding by increasing its size and by incorporating more demanding tasks into the questionnaire.] With a less demanding task, the behavioral data would perhaps have shown a labeling effect in accordance with the salience hypothesis.

The final, and perhaps most plausible, explanation argues that the type of labeling done in this study was not sufficient to produce a major change in subjects' self-perceptions. That is, unlike Miller, Brickman and Bolen's (1975) subjects, our subjects were labeled only once in a general way by an experimenter perhaps lacking, in high credibility and who did not cite concrete evidence to substantiate the label (Sternthal and Craig 1982, p. 976). Although this appears to have been sufficient in producing greater behavioral intentions, the self-perception induced by labeling was not adequate to motivate completion of a demanding survey.

Although none of these possibilities poses a threat to the Tybout and Yalch salience hypothesis, they do provide some considerations for future research. For instance, the sensitivity of the salience of a label to the passage of time, the effects of the size or the requested criterion behavior on the efficacy of labeling as an influence strategy, and single versus multiple-label strategies need to be investigated.

REFERENCES

Bem, D. J. (1979), "Self-Perception Theory," in Advances in Experimental Social Psychology, Vol. 6, ed. L. Berkowitz, New York: Academic Press.

Cialdini, R. and Ascani, K. (1976), "Test of a Concession Procedure for Inducing Verbal, Behavioral and Further Compliance with a Request to Give Blood," Journal of Applied Psychology, 61, 995-300.

Kelley, H. H. (1973), "The Process of Causal Attribution," American Psychologist, 28, 107-28.

Kraut, R. E. (1973), "Effects of Social Labeling in Giving to Charity," Journal of Experimental Social Psychology, 3, 551-62.

Markus, H. (1977), "Self-Schema and Processing Information about Self," Journal of Personality and Social Psychology, 35, 63-78.

Miller, R., Brickman, P. and Bolen, D. (1975), "Attribution vs. Persuasion as a Means of Modifying Behavior" Journal of Personality and Social Psychology, 31, 430-41.

Pliner, P., Hart, H., Kohl, J. and Saabi, D. (1974), "Compliance without Pressure: Some Further Data on the Foot-in-the-Door Technique," Journal of Experimental Psychology, 10, 17-22.

Reingen, P. H. and Kernan, J. (1977), "Compliance with an Interview Request: A Foot-in-the-Door, Self-Perception Interpretation," Journal of Marketing Research, 14, 365-9.

Reingen, P. H. and Kernan, J. (1979), "More Evidence on Interpersonal Yielding," Journal of Marketing Research, 16, 588-93.

Scott, C. A. (1977), "A Self-Perception Approach to Modifying Socially Conscious Behavior: The Foot-in-the-Door Technique," Journal of Consumer Research, 4, 56-64.

Seligman, C., Bush, M. and Kirsch, K. (1976), "Relationships Between Compliance in the Foot-in-the-Door Paradigm and Size of First Request," Journal of Personality and Social Psychology, 33, 517-20.

Steele, C. M. (1975), "Name-calling and Compliance," Journal of Personality and Social Psychology, 31, 361-9.

Snyder, M. and Cunningham, M.R. (1975), "To Comply or Not to Comply: Testing the Self-Perception Explanation for the 'Foot-in-the-Door' Phenomenon," Journal of Personality and Social Psychology, 31, 64-7.

Sternthal, B. and Craig, C. S. (1982), Consumer Behavior, Englewood Cliffs: Prentice-Hall.

Swinyard, W. R. and Ray, M. L. (1977), "Advertising-Selling Interactions: An Attribution Theory Experiment," Journal of Marketing Research, 14, 509-16.

Taylor, S. and Fisk, S. (1978), "Salience, Attention and Attribution: Top of the Head Phenomena," in Advances in Experimental Social Psychology, Vol. 11, ed. L. Berkowitz, New York: Academic Press.

Tybout, A.M. and Yalch, R.R. (1980), "The Effect of Experience: A Matter of Salience?", Journal of Consumer Research, 6, 406-13.

----------------------------------------