Extending the Self-Perception Explanation: the Effect of Cue Salience on Behavior

ABSTRACT - A 2 x 2 factorial experiment was conducted in which consumers received either an incentive or no incentive for choosing one brand (focal brand) over another brand (nonfocal brand). Then, they were given several pieces of information about the brands. The order of these pieces of information was varied so that either participants' behavior was the only highly salient cue or multiple cues were salient. Although no significant effect of the treatments was observed on subsequent choice of the focal brand, a highly significant incentive by cue salience interaction was obtained for choice of the nonfocal brand. When behavior was the only highly salient cue, the presence of an incentive operated as a discounting cue resulting in less choice of the nonfocal brand than no incentive. In contrast, when multiple cues were salient, the incentive served as a reinforcing cue, resulting in more choice of the nonfocal brand than no incentive.

Citation:

Carol A. Scott and Alice M. Tybout (1979) ,"Extending the Self-Perception Explanation: the Effect of Cue Salience on Behavior", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 50-54.

Advances in Consumer Research Volume 6, 1979      Pages 50-54

EXTENDING THE SELF-PERCEPTION EXPLANATION: THE EFFECT OF CUE SALIENCE ON BEHAVIOR

Carol A. Scott, University of California, Los Angeles

Alice M. Tybout, Northwestern University

ABSTRACT -

A 2 x 2 factorial experiment was conducted in which consumers received either an incentive or no incentive for choosing one brand (focal brand) over another brand (nonfocal brand). Then, they were given several pieces of information about the brands. The order of these pieces of information was varied so that either participants' behavior was the only highly salient cue or multiple cues were salient. Although no significant effect of the treatments was observed on subsequent choice of the focal brand, a highly significant incentive by cue salience interaction was obtained for choice of the nonfocal brand. When behavior was the only highly salient cue, the presence of an incentive operated as a discounting cue resulting in less choice of the nonfocal brand than no incentive. In contrast, when multiple cues were salient, the incentive served as a reinforcing cue, resulting in more choice of the nonfocal brand than no incentive.

INTRODUCTION

During the past decade, interest has emerged in understanding how consumers' experiences influence their subsequent attitudes and actions. Much of the inquiry pertaining to this issue has been motivated by self-perception theory (Bem, 1972). According to this formulation, individuals examine their own past behavior and its circumstances in an effort to determine the cause of their behavior. When no plausible external cause for the behavior is present, the behavior is interpreted as a veridical representation of the actors' internal states and a positive attitude toward that behavior is inferred. This inferred attitude, in turn, may guide subsequent behavior. If, however, plausible external causes for the behavior are present (i.e., an incentive is offered for the behavior or there is a lack of volition in performance of the behavior), the role of internal states in determining the behavior is discounted and behavior is likely to have little systematic effect on attitudes and future behavior. A substantial number of social psychological investigations have examined the self-perception premise that individuals use their own behavior and its circumstances to infer their attitudes and/or direct their future actions (e.g., Calder and Staw, 1975; Freedman and Fraser, 1966; Lepper, Misbett and Greene, 1973; Snyder and Cunningham, 1975; Valins, 1974). Furthermore, consumer researchers have tested the effectiveness of self-perception based strategies in marketing contexts (e.g., Dholakia and Sternthal, 1977; Reingen and Kernan, 1977; Scott, 1976, 1977; Swinyard and Ray, 1977). These investigations have provided support for self-perception predictions. In contrast, several studies have found that self perception theory predictions are obtained only under certain conditions (e.g., Scott and Yalch, 1978; Tybout, 1978).

This observation has motivated the search for factors that are necessary for self perceptions to occur. One factor identified is the salience of an individual's own behavior. In a field experiment, Tybout (!978) gained compliance with a behavior under two conditions; for some research participants a discounting cue was present, while for others the discounting cue was absent. The salience of the behavior was manipulated by either asking individuals to identify the causes of their behavior (i.e., high salience), or not asking them to complete this task (low salience). When behavior was made highly salient, the self-perception prediction was obtained; the presence of the discounting cue undermined internal attribution and persistence of the behavior. In contrast, when behavior was not made highly salient, no self-perception effect was observed.

Tybout's findings suggest that an individual's own behavior must be highly salient if the self-perception process is to occur. Further evidence for this contention emerges in a study reported by Scott and Yalch (1978). They offered some consumers an incentive for choosing to taste a new soft drink while others were not given an incentive for performing this task. Prior to tasting the soft drink, some participants were given an opportunity to visually examine the drink, while others were not. They found that when individuals were allowed to usually examine the beverage before tasting it, the incentive served as a discounting cue. Product evaluation was more favorable when the incentive was absent than when it was present. In contrast, when no opportunity was given to visually inspect the product selected, the incentive resulted in a more favorable evaluation than when there was no incentive for product choice.

The Scott-Yalch findings can be explained by interpreting the variation in opportunity to examine the product visually as a salience induction. When research participants were given the opportunity to visually inspect the product chosen, their past behavior was highly salient. As a result, they engaged in an attributional analysis of that behavior. When no visual examination opportunity was provided, subjects' own behavior was not salient. Thus, no attributional analysis was undertaken. Instead, cues about the product were simply processed and the positive incentive cue, like other positive cues, served to enhance or reinforce favorable evaluations of the product. Apparently, the effect of both behavior and other cues in the environment, such as incentives, may be mediated by the salience of behavior.

In sum, the evidence reviewed underscores the fact that the salience of individuals' own behavior is a necessary condition for attributional analysis to occur. However, it does not provide an explanation of the mechanism by which salience mediates attributions. This issue is addressed by Pryor and Kriss (1977). They manipulated the timing or sequencing of different attribution cues and observed that the attributions subjects made were related to the relative availability of the cues from memory. Highly available cues guided the attributions made. The implication is that cue salience mediates information processing by influencing the extent to which various cues are available to an individual for acquiring a disposition.

The purpose of the present study is to investigate further the process by which cue salience mediates influence. This entailed having all subjects engage in a behavior. Some subjects received an incentive for performing the behavior, while others did not. Next, all subjects received two pieces of information relevant to their behavior. One was opposed to the behavior. In was intended to motivate subjects to process the information they had pertaining to the object of behavior. The other piece of information provided support for the behavior subjects had engaged in. The timing of these two pieces of information was manipulated so that subjects assigned to the cognitive work immediate treatment were presented with information that opposed their behavior, immediately after they had engaged in the behavior. This induction was intended to make subjects' own behavior salient. In the cognitive work delayed condition, information opposing subjects' behavior was deferred until after they had received information that was consistent with their behavior. In this situation, cues beyond subjects' own behavior should become salient.

The design of this study is a 2 x 2 factorial, with 2 levels of incentive (no incentive, incentive) and 2 levels of timing (cognitive work immediately, cognitive work delayed). The predicted results are shown in Figure 1. When cognitive work was stimulated immediately after thc behavior, it was hypothesized that the salience of that behavior would cause the self-perception process to occur. In the no-incentive condition, response would be more favorable than in the incentive condition, because the incentive would serve as a discounting cue for internal attribution of the behavior. In contrast, when cognitive work was delayed until after additional information was obtained, no causal analysis of behavior was expected because multiple cues would be salient. A more favorable response was expected in the incentive condition than the no-incentive condition because the incentive would serve to reinforce behavior (Scott and Yalch, 1978). Further, it was predicted that within the no-incentive condition, stimulating cognitive work immediately after behavior would lead to more favorable evaluations and behavior than delaying cognitive work until after further information had been received. This was expected because, in the no-incentive condition, behavior should be a positive cue and thus circumstances which increase the salience of this cue (i.e., cognitive work immediately) should lead to more favorable evaluations. Finally, in the incentive condition, delaying cognitive work until further information had been received was expected to produce more favorable evaluations and behavior than stimulating cognitive work immediately after behavior. This result was anticipated because the delay would reduce the salience of behavior leading the incentive to be interpreted as a reinforcing rather than discounting cue.

FIGURE 1

PREDICTED INTERACTION

Participants were 174 female residents of Columbus, Ohio, between the ages of 21 and 60, who were recruited by a commercial firm for a marketing research study. Most of them were members of volunteer or service organizations which received a $2 contribution from the research firm for each person participating. The women were scheduled to participate in the study in groups of 5-10 and these groups were randomly assigned to one of the four experimental treatments. Tables with side panels were used to insure that women in the same group could not see each others' responses or communicate with each other during the study. Twenty women were eliminated from the final analysis either because they failed to complete the questionnaire or because they chose the nonfocal brand. [There was no significant difference in subject attrition for the various treatments.]

Procedure

Upon arrival at the research firm, groups of participants were escorted to a test kitchen and were introduced to two experimenters. One experimenter explained that the purpose of the study was to obtain consumers' evaluations of two new formulations of a diet soft drink for a well known manufacturer. Participants were told that they would receive several pieces of information about these new brands, including a sample of one of the brands to taste, as a basis for making their evaluations. Then, they were instructed to turn over the booklet which had been placed on the table in front of them and read descriptions of each of the brands which purportedly had been taken from advertising campaigns being developed for them. At this point the procedure varied depending upon the incentive treatment to which the participants had been assigned.

No Incentive Condition. Individuals assigned to this treatment were simply instructed to read the two brand descriptions carefully and then select the brand they wished to sample. For purposes of control, it was important that all individuals voluntarily select the same brand for trial. Therefore, the description of one brand, which we shall refer to as the focal brand, was somewhat more favorable than that for the other nonfocal brand. Nearly all (90%) participants in this treatment chose the focal brand.

Incentive Condition. Individuals assigned to this treatment also were told to read the two brand descriptions carefully and then select the brand they wished to sample. In addition, they were informed that in order to encourage trial the focal brand, on that particular day, a coupon worth 50C at a popular fast-food restaurant would be given to those who decided to sample that brand. Ninety-six percent of the participants in this condition chose the focal brand. [There was no significant difference in the number of participants choosing the focal brand in the no-incentive and incentive treatments. All individuals who did not choose the focal brand were deleted from the analysis.] Immediately after participants in this treatment made their choice, one of the experimenters distributed coupons to those who had chosen the focal brand, thereby eliminating any delay in receiving the reward.

Participants were asked to indicate their choice by marking it on their questionnaire. The rationale for this task given to research participants was that this information would facilitate data processing. In reality, it served to reinforce the fact that a decision had been made.

After participants selected the brand they wanted to sample, they turned the page in their booklets and answered some general questions about the product category. Next, individuals received further information about the two brands which included a manipulation of the second independent variable, timing of cognitive work.

Cognitive Work Immediately. At this point, individuals assigned to the cognitive work immediately treatment received information designed to motivate cognitive processing regarding their evaluation of the focal brand. Specifically, processing was stimulated by providing them with information which was inconsistent with their with their choice behavior (i.e., information which indicated that the nonfocal brand was more desirable than the chosen, focal brand). Participants in this treatment heard the experimenter say:

"Before you taste a sample of the brand you have chosen, we thought you might find it interesting to learn how others like yourself have responded to these brands in previous research. Several months ago we ran some tests in which individuals were given both brands to taste and compare. In these studies we found that more than 80% of the people preferred brand 534 (the nonfocal brand) to brand 453 (the focal brand).'' [Three digit numbers were used to identify the two brands in the study to avoid any confounding effect of a brand name or letter.]

This manipulation was felt to be an appropriate way to motivate cognitive processing since it created uncertainty and uncertainty has been found to motivate information search (Sears and Freedman, 1967). After receiving this negative information about the focal brand, individuals were allowed to gather further data by tasting the focal brand. Pretesting of the drink used as the focal brand indicated that most individuals evaluated the taste of the drink quite favorably. Therefore, this second piece of information was considered to be positive or supportive of the initial choice behavior.

Cognitive Work Delayed. Individuals assigned to the cognitive work delayed treatment received the same positive and negative information about the focal brand as those in the cognitive work immediately treatment. However, the ordering of these two pieces of information was reversed to delay cognitive processing until after further information about the focal brand was obtained (i.e., the negative information followed, rather than preceded, the positive taste information).

Following this information, participants evaluated the focal brand by responding to attitudinal questions in their booklets. Then, individuals were informed that a lottery, in which the prize was a case (24 cans) of diet soft drink, would be held for all study participants. They were asked to mark on their booklet the number of units of each of the new brands (focal and nonfocal) and/or a popular national brand they would want in their case of diet soft-drink if they won the lottery.

Finally, the experimenters led an informal discussion of the study with participants in which they answered questions and attempted to uncover any suspicion about the study purpose and procedure. No evidence of suspicion was found. In general, participants appeared to accept the cover story. The majority of their questions focused on issues such as the identity of the manufacturer and when the brands might be introduced in the marketplace. At the end of the discussion, participants were asked to refrain from discussing the study until the research was complete (two weeks) and were thanked for their cooperation.

Dependent Measures

Both attitudinal and behavioral dependent measures were taken. However, due to space limitations and the fact that attitudinal data analysis was not complete in time to be included in this paper, only the measure used to check random assignment and the behavior measure are presented and discussed here.

Affect Toward Product Category. At the outset of the study participants expressed their affect toward the product category by responding to the question: Assume that you were to taste a sugar-free soft drink. How likely is it that you would enjoy it? Individuals drew a vertical line through the 10-inch line scale, ranging from not at all likely to extremely likely, at the point which best expressed their opinion. The responses were later scored using a ruler. This measure was used to check the random assignment of groups of participants to treatments, if the random assignment procedure was successful, no difference between treatments on this measure should be obtained.

Choice of Focal Brand. The number of cans of the focal brand that each individual indicated that she would want in her case of diet soft drink, if she won the lottery, served as a measure of choice of the focal brand.

Choice of the Nonfocal Brand. The number of cans of the nonfocal brand that each participant indicated that she would want in her case of diet soft drink, if she won the lottery served as a measure of choice of the nonfocal brand. [The number of cans of the national brand chosen was not analyzed as a dependent measure in this study because doing so would use up all the degrees of freedom on the choice measure.]

RESULTS

Randomization Check

Analysis of participants' affect toward the product category by treatment revealed no systematic differences between treatments on this measure (F's for all factors and interactions < 1, n.s.). Therefore, the random assignment procedure is judged to have been successful.

Behavior Measures

The impact of the treatments on choice of the focal and nonfocal brand was assessed using the BMDP program for analysis of variance. Affect toward the product category was employed as a covariate in all analyses. [Affect toward the product category was an appropriate covariate for this analysis because it was not influenced by the treatments (see discussion under randomization check), but it did contribute significantly to the high variance in each treatment condition. Use of the variable as a covariate did not change the pattern of the results, but did increase the significance of the results in some cases.] It was anticipated that the interaction of incentive and timing of cognitive work shown in Figure 1 would be obtained for the focal brand.

Counter to our hypotheses, an analysis of variance of the focal brand data in Table i resulted in no significant effects. However, an analysis of variance on choice of the nonfocal brand (Table 1) did yield a highly significant incentive by timing of cognitive work interaction IF (1,169) = 17.52, p < .001]. [The covariate was not significant in this particular analysis (F < 1 ).]

Furthermore, systematic contrasts between treatments for choice of the nonfocal brand reveal that the pattern of data is the one which had been anticipated for the focal brand (Table 2). Thus, the predicted effects of the treatments on choice were not obtained for the focal brand, but did appear for the nonfocal brand.

TABLE 1

CHOICE OF FOCAL AND NONFOCAL BRANDS BY TREATMENT

TABLE 2

PLANNED CONTRASTS FOR SIGNIFICANT INCENTIVE BY TIMING OF MOTIVATION FOR COGNITIVE WORK INTERACTION

DISCUSSION

The major finding of the study is the significant incentive by timing of cognitive work interaction which was obtained for choice of the nonfocal brand. Specifically, it was found that:

1. When cognitive work was stimulated immediately after the choice behavior, more choice of the nonfocal brand resulted under the no-incentive condition than under the incentive condition.

2. When cognitive work was delayed until after further information was obtained, more choice of the nonfocal brand was produced under the incentive condition than under the no-incentive condition.

3. When no incentive was offered for initial choice of the focal brand, more choice of the nonfocal brand resulted under the cognitive work immediately condition than under the cognitive work delayed condition.

4. When an incentive was offered for the initial choice of the focal brand, more choice of the nonfocal brand was produced under the cognitive work delayed condition than under the cognitive work immediately condition.

In contrast, the predicted interaction for the focal brand was not observed.

One plausible explanation for the results observed exists. Prior to the lottery, individuals' evaluations of the focal brand were independent of the nonfocal brand (i.e., they gave ratings of their liking for the focal brand, the degree to which the focal brand possessed various attributes, etc. without reference to the nonfocal brand). Under these conditions, it seems clear that the predicted incentive by timing of cognitive work interaction should be obtained for evaluations of the focal brand. However, the lottery measure of choice forced a dependent relationship between behavior with respect to the focal and nonfocal brands.

Furthermore, when this measure was taken, individuals had two major pieces of information they could use to make their choice: (1) their attitude toward the focal brand, which appears to demonstrate the predicted incentive by timing of cognitive work interaction [In fact, preliminary results of analysis of attitudinal items (not presented here), suggests that such an interaction was obtained for the focal brand when it was evaluated by itself.], and (2) knowledge that others like themselves preferred the nonfocal brand to the focal brand. In view of the dependency between choice of the focal and nonfocal brands and the nature of the information available, it is not surprising that the nonfocal brand exhibits the predicted interaction. Individuals who were relatively positive about the focal brand reasoned that they would like the nonfocal brand even more and therefore chose it, while those who evaluated the focal brand less favorably had little reason to believe that they would like the nonfocal brand much better and therefore were less inclined to choose it. Thus the effect of the treatments on attitudes toward the focal brand is observed on choice of the nonfocal brand.

In summary, this study makes several contributions. It replicates previous research (Scott and Yalch, 1978), demonstrating that effects of behavior and behavior-contingent incentives on subsequent evaluations of a product are mediated by the salience of behavior. When behavior is highly salient, an incentive serves as a discounting cue for internal attribution of the behavior, resulting in less favorable evaluations than when no incentive is given, while when behavior is not highly salient, an incentive serves as a reinforcing cue enhancing evaluations relative to no incentive. Further, the present study implicates cue availability as the process by which salience determines the effect of behavior in subsequent cognitive processing. Finally, the results of this research suggest that when brand choices are interdependent and direct comparisons are made (i.e., side by side test in advertisements), self-perceptions of behavior with respect to one brand may influence subsequent choice of a competitive brand.

REFERENCES

Daryl J. Bem, "Self-Perception Theory," in L. Berkowitz (ed.), Advances in Experimental Social Psychology, (New York: Academic Press, 1972).

Bobby J. Calder and Barry M. Staw, "Self-Perception of Intrinsic and Extrinsic Motivation," Journal of Personality and Social Psychology, 31(1975), 599-605.

Ruby R. Dholakia and Brian Sternthal, "Highly Credible Sources: Persuasive Facilitators or Persuasive Liabilities?'' Journal of Consumer Research, 3(1977), 223-232.

Jonathan L. Freedman and Scott C. Fraser, "Compliance Without Pressure: The Foot-in-the-Door Technique," Journal of Personality and Social Psychology, 4(1966), 195-202.

Mark R. Lepper, D. Greene and R. E. Nisbett, "Undermining Children's Intrinsic Interest with Extrinsic Reward: A Test of the Overjustification Hypothesis," Journal of Personality and Social Psychology, 28(1973), 129-137.

John B. Pryor and Mitchel Kriss, "The Cognitive Dynamics of Salience in the Attribution Process," Journal of Personality and Social Psychology, 35(1977), 49-55.

Peter Reingen and Jerry Kernan, "Compliance with an Interview Request: A Foot-in-the-Door, Self-Perception Interpretation," Journal of Marketing Research, 14(1977), 365-369.

Carol A. Scott, "The Effect of Trial and Incentives on Repeat Purchase Behavior," Journal of Marketing Research, 13(1976), 263-269.

Carol A. Scott, "Modifying Socially-Conscious Behavior: The Foot-in-the-Door Technique," Journal of Consumer Research, 4(1977), 156-164.

Carol A. Scott and Richard F. Yalch, "A Test of Self-Perception Explanation of the Effects of Rewards on Intrinsic Interest," Journal of Experimental Social Psychology, 14(1978), 180-192.

D. Sears and J. L. Freedman, "Selective Exposure to Information: A Critical Review," Public Opinion Quarterly, 31(1967), 194-213.

Mark Snyder and M. R. Cunningham, "To Comply or Not Comply: A Testing of the Self-Perception Explanation of the Foot-in-the-Door Technique," Journal of Personality and Social Psychology, 31(1975), 64-67.

William R. Swinyard and Michael R. Ray, "Advertising-Selling Interactions: An Attribution Theory Experiment," Journal of Marketing Research, 14(1977), 509-516.

Alice M. Tybout, "The Relative Effectiveness of Three Behavioral Influence Strategies as Supplements to Persuasion," Journal of Marketing Research, 15(1978), 229-242.

S. Valins, "Persistent Effects of Information About Internal Reactions: Ineffectiveness of Debriefing," in H. London and R. E. Nisbett (eds.), Thought and Feeling: Cognitive Alteration of Feeling States, (Chicago: Aldine, 1974).

B. Winer, Statistical Principles in Experimental Design, (New York: McGraw Hill, 1971).

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