Limitation of Self-Referencing As a Persuasion Strategy

ABSTRACT - Research has determined that self-referencing strategies (i. e., stimulating associations between message information and the individual's prior experiences and self-concept) are effective mnemonic devices but limits to the technique have recently been identified. The results of an experiment demonstrating that overstimulating self-referencing hinders persuasiveness are discussed.

Citation:

Richard F. Yalch and Brian Sternthal (1984) ,"Limitation of Self-Referencing As a Persuasion Strategy", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 71-74.

Advances in Consumer Research Volume 11, 1984      Pages 71-74

LIMITATION OF SELF-REFERENCING AS A PERSUASION STRATEGY

Richard F. Yalch, University of Washington

Brian Sternthal, Northwestern University

ABSTRACT -

Research has determined that self-referencing strategies (i. e., stimulating associations between message information and the individual's prior experiences and self-concept) are effective mnemonic devices but limits to the technique have recently been identified. The results of an experiment demonstrating that overstimulating self-referencing hinders persuasiveness are discussed.

INTRODUCTION

The personalization of marketing communications through the use of new technology is becoming more prevalent. Direct mail consultants have promoted the technique with claims of up to 25 per cent sales increases in response to computer-generated personalized appeals; e.g., each message is written with the individual's name inserted in various places (Anderluh 1982). Improvements in computer-controlled voices will soon enable other media like telephones, cable television, and point of purchase displays to be Personalized.

One factor underlying advocacy of these procedures is the belief that they are likely to stimulate the audience to engage in self-referencing when processing the marketing communication and this will enhance message recall (Leavitt, Greenwald and Obermiller 1982). Self-referencing refers to the use of personal experiences or personal knowledge as an organizing device for the assimilation of new information. For example, one may learn about an another's shopping trip by thinking of one's most recent trip. By relating the other's trip to a personal one, the individual is easily able to visualize the experience.

The belief that inducing consumers to self-reference message information will enhance communication effectiveness receives support from the theoretical view that self-referencing invokes self-schemata which provide for more elaborate encoding than other schemata. Self-schemas are "cognitive generalizations about the self that organize and guide the processing of self-related information contained in the individual's social experience" (Markus, 1977, p. 64). Because much of the information that we are exposed to is about ourselves, it is relatively easy to relate new information to this information and to create many more associations than would be true with other organizing thoughts. Generally, the more associations that one has for new information the better one's ability to remember this information.

Evidence for enhanced communication effectiveness is found in research demonstrating that self-referencing induces greater memorability of stimulus information than structural, phonemic, or semantic referencing (Rogers, Kuiper and Kirker 1977), or referencing to other objects (Lord 1980). However, it has also been found that self-referencing may undermine the memorability of stimulus information (Lord 1980). This was true when individuals were asked to use imagery rather than propositionally, verbally, or semantically encoded knowledge about themselves. An understanding of when self-referencing might not be effective would provide strategists with guidelines for its proper implementation. Further, it should provide insight into the more general concern about how information is processed.

A useful starting point in addressing this issue is the research conducted by Lord (1980). Subjects were shown a large number of target words. In experiment I, the subjects' task was to answer one of four questions about each word; was the word descriptive of themselves, their father, Walter Cronkite, or a tree. The findings indicated greatest recall when the words were self-referenced. In Experiment II, subjects were instructed to construct a mental picture that included the target word and one of four referents. When these instructions were followed, self-referenced words were recalled most poorly.

Lord interpreted these data as evidence for the fact that information pertaining to self is semantically processed whereas information related to other objects can be imaginally processed. Further, it is difficult for individuals to construct mental images of themselves. Viewed from this perspective the superior recall of self-referenced information found in Experiment I occurred because the task of determining whether a word was descriptive of a referent was semantic in nature. The reversal found in Experiment II is explained by interpreting the task of constructing a mental image as an imaginal one. This explanation, however, is at odds with another finding reported in experiment II. Specifically, when asked whether their images were fuzzy or vivid, subjects claimed that the images they had constructed in relating words to themselves were more vivid than the images generated in relating words to other referents.

A more promising explanation for Lord's findings is the availability hypothesis. This hypothesis is based on the premise that individuals have a greater number of associations to self in memory than they have to most other objects. Thus, stimulus information is more readily related to self than to other objects and hence more available for recall. This appears to be what occurred in Lord's first experiment. In contrast, instructions to construct a mental image that included the target word and self may have caused subjects to flood memory with self-referencing information that only incidentally included the target word. This interference interpretation is not only consistent with the poorer recall of self-referenced target words found in Experiment I but also with subjects' report of more vivid images when self-referencing.

The purpose of the present research was to examine the effects of self-referencing on individuals' attitudes toward an unfamiliar product. Subjects listened to a radio commercial that described the product and then evaluated it. Two independent variables were manipulated. One was a label the experimenter conferred on subjects which either described them as individuals who would be highly interested in one of the attributes of the product advertised in the ensuing radio commercial or this label was absent. Receipt of the label was expected to enhance the availability of message information by virtue of its association to self. Thus, it was predicted that judgment would be more favorable toward the advertised product on the attribute mentioned in the description when the label was present than when it was absent. Some empirical grounds for this prediction are found in studies showing that conferring a label increases the likelihood of label-consistent actions (Kraut 1973; Miller, Brickman and Bolen 1975; Tybout and Yalch 1980). A possible explanation for this is that the label causes subject to rehearse self thoughts about engaging in this behavior and this makes them more available at the point when individuals must decide whether or not to engage in the behavior.

The second independent variable was the instructions given to the subjects. In the instructions conditions, they were told to relate what they were hearing to their own experiences and to themselves, whereas in the no instruction condition, this directive was absent. It was anticipated that the instructions would stimulate the processing of message information in relation to oneself. The effect of this information on product evaluation was expected to depend on whether or not individuals had been labeled.

Those who were not labeled and who were asked to self-reference would engage in processing using personal experiences. Relative to the unlabeled subjects who were not instructed to use self-referencing, they would assimilate more of the positive taste information in the message, be influenced by it during their product evaluation, and this would result in a more favorable evaluation. In contrast, for those who were labeled as being particularly interested in a product's taste, the self-referencing instruction was expected to stimulate scrutiny of this hypothesis in assessing the unfamiliar product. In pursuing this strategy, subjects were expected to generate associations that interfered with memory of the positive description of the product and this would reduce the favorableness of their product evaluation. Relative to the single treatment groups (those receiving a label but not self-referencing instructions and those not receiving a label but requested to use self-referencing), they should be less favorable toward the product. For them, the message advocacy would have little influence on their product evaluations.

METHOD

Sixty students enrolled in an introductory marketing course were recruited to participate in a beverage taste test. The subjects participated in groups ranging from five to ten persons in a seminar room near their classroom. Three experimenters blind to the predictions conducted the studY.

The procedure involved responding to a questionnaire booklet in a controlled manner. Initially, subjects read the first page which described the study as being conducted by a beverage manufacturer. Next, they completed one page of questions about their demographic characteristics and preferences for beverages. At this point, half the subjects received a label from one of the experimenters reporting that their initial responses indicated a high interest and concern about the taste of beverages. The other half of the subjects received no feedback about their responses.

The next part of the procedure involved the subjects being given a glass of the test beverage that contained 200 ml. of Ambrosia juice. They were asked to taste a small amount of the juice and then listen to a radio commercial that was to be used to promote the product. Prior to listening, subjects were given one of two written instructions as to how to process the information contained in the message. Those in the self-referencing instructions condition were told "As you listen to the following commercial message, please try to relate what you are hearing to your own experiences and to yourself." In the no instructions condition, they were told to listen to the commercial as they normally would listen to an advertising message.

The radio message was recorded by a semi-professional actor and mentioned six flavor-related characteristics of the product (e.g., refreshing, good flavor, not sweet) intermixed among nine other product characteristics (e.g., no preservatives, no artificial ingredients, available in three sizes).

Immediately after the commercial, the participants were given three minutes to list all the thoughts that occurred to them about the advertised product. Twelve numbered lines were provided on one page for this purpose. Next, they were instructed to return to the thought-listing page and evaluate the thoughts as being favorable, neutral or unfavorable toward the product. A score was later calculated by subtracting the number of unfavorable thoughts from the number of favorable thoughts for each subject.

The last part of the questionnaire required subjects to evaluate the product on seven, seven-point bipolar adjective scales. Responses on these scales were subjected to a principal components factor analysis with varimax rotation. One factor emerged from this analysis. It included the following five items: good tasting/bad tasting, good aftertaste/bad aftertaste, good flavor/bad flavor, pleasant/unpleasant, and enjoyable/not enjoyable. Subjects' responses to the five items were summed into an overall TASTE index, that was highly reliable (Cronbach's alpha = .91).

RESULTS

The means and standard deviations for the TASTE index and thought measure, categorized by the independent variables are presented in the table. An analysis of variance indicated the absence of a label or instruction main effect on the TASTE index (Fs < 1), but a significant interaction between these factors (F(1,56) = 4.81, p < .033. The predicted negative effects were examined using a series of treatment comparisons. It was found that the instructions with a label induced a less favorable evaluation on the TASTE index than did instructions without a label [t(28) = 1.92, p < .05]. It was also observed that a label with instructions produced a less favorable evaluation of TASTE than did a label without instructions lt(28) = 1.64, p < .06]. These comparisons suggest that the presence of two self-referencing strategies undermines the favorableness of the TASTE evaluation in relation to the presence of either self-referencing strategy alone.

Comparisons were made to examine the prediction that the presence of a single self-referencing strategy would enhance evaluation in relation to the absence of a self-referencing strategy. It was found that the presence of a label but no instructions enhanced the favorableness of the TASTE evaluations in relation to the control group (normal instructions and no label). However, this effect did not reach conventional levels of statistical significance [t(28) = 1.22, p = .11]. The comparison of the self-referencing instructions group that did not receive a label with the control group also indicated enhancement but the difference was only marginally significant [t(28) = 1.50, p = .07].

TABLE

MEANS AND STANDARD DEVIATIONS FOR TASTE INDEX AND THOUGHTS CATEGORIZED BY THE INDEPENDENT VARIABLES

The process by which self-referencing might affect persuasion was studied by looking at the effects of the treatments on the thought measure. An analysis of variance indicated no main effects (Fs < 1) and a significant label by instruction interaction (Ft(1,56) = 3.92, p < .05]). Treatment comparisons yielded a pattern similar to that found for the TASTE index, but the significance tended to be smaller. Within the instruction condition, the presence of a label induced less favorable thoughts than when the label was absent [t(28) = 1.65, p < .05]. And, within the label conditions, subjects produced more favorable thoughts when instructions were absent than when they were present [t 28) = 1.93, p < .05]. The remaining comparisons were not significant (ts < 1).

DISCUSSION

The results of the study support the prediction that self-referencing is not always an effective device to enhance the persuasiveness of an advertising message. Those subjects who had been labeled as interested in the taste of the product prior to being instructed to use a self-referencing strategy during the presentation of the advertising message were less favorably disposed toward the advertised product than unlabeled subjects using this strategy and than labeled subjects who were not following self-referencing instructions. Further, they were no more favorably disposed toward the product than unlabeled subjects who were not using self-referencing.

These data provide some evidence consistent with current notions of memory operations. According So this view, the self is a rich network of associations. Thus, stimulating people to self-reference such as by instructing them to self-reference or by introducing a label, enhances the likelihood that message information will be encoded in memory and activated when rendering a judgment. As a result, self-referencing increases influence. However, prompting people to focus too much on information as it relates to themselves, such as by introducing multiple self-referencing strategies, may cause people to activate information about themselves that interferes with the processing of message information and hence undermines the persuasive impact of a communication.

This explanation is consistent with Lord's findings that self-referencing that involves a very rich network of association (use of visual imagery) was detrimental to the recall of words linked to this network. Research by Sullivan (1981) provides additional support for the belief that self-referencing may result in many associations that interfere with information processing. In his study, subjects were exposed to one of five want ads for a position in financial management for a glass and plastics manufacturer which had appeared in the College Placement Annual. The ads were modified so that one group saw the original advertisement and the others versions which substituted varying amounts of the word YOU for words like employee, his or her, person, and other impersonal references. The messages had either four, seven, ten or fourteen you's inserted. Subjects were asked immediately after exposure to list all their thoughts and then to indicate which ones were favorable to the product. The results were that there was an increase in support and total thoughts listed when the messages contained four you's, but that there was a decline when the messages contained ten or fourteen you's relative to seven you's.

This might be considered evidence that those subjects presented with many cues to self-reference were stimulated to rehearse thoughts of little relevance to the message and which were therefore difficult to recall when requested to list their thoughts.

Support for this "interference" explanation is found in a variety of ways in the present study. First, the thought measure analysis revealed that subjects' thoughts were more unfavorable when they were exposed to a dual strategy compared to the single strategy conditions. In memorial terms, the multiple self-referencing strategies interfered with the generation of associations to the message information, if for no other reason than because the message was constructed to be highly persuasive. The thoughts data pertaining to the effect of a single self-referencing strategy in relation to the absence of this strategy are also in the predicted direction. However, treatment differences in thought favorableness were nonsignificant. This outcome is probably a reflection on the difficulty in producing an effect of a single self-referencing strategy on flavor evaluations once subjects had tasted the product.

The final evidence in support of the interference explanation is the pattern of correlations between the subjects' attitudes and the amount of the juice they consumed. As expected, these correlations were higher when individuals had used a self-referencing strategy in processing the advertising message than when they had not. Apparently, self-referencing evokes a network of associations that guides both evaluative and behavioral responses, whereas in the absence of self-referencing different associations may be accessible in rendering evaluative and behavioral judgments.

From a practical perspective, the present research suggests that self-referencing should not always be assumed to be an effective method for enhancing communication effectiveness. Although it may induce consumers to encode information in a rich network of associations, it is not always clear that this network will be congenial to the new information. Even when the new information is not "lost" in the self thoughts, subjects may become more aware of the inconsistency of the message advocacy with their previous experience.

The interaction of self-referencing and initial attitudes has been shown in a recent article by Tybout, Calder and Sternthal ( This). They found self-referencing to effective in neutralizing attitudes toward a fast food restaurant which were created by the presentation of an untrue and disbelieved rumor. Subjects who were exposed to the rumor that McDonald's used worms in its hamburgers reported less favorable evaluations of the restaurant chain than subjects unexposed to this rumor. However, subjects who were exposed to the rumor but who were also asked to recall personal knowledge about McDonald's (the location of the restaurant they frequent most often, how often they visit it per year, and whether or not it had indoor seating), were not less favorably disposed toward McDonald's than those unexposed to the rumor. Apparently, invoking personal experiences with McDonald's which were inconsistent with the worm rumor eliminated its influence.

Together, Tybout, et al. ( 1981 ) and this study show that self-referencing can have quite different results depending on the message and the audience's initial attitude. In their study, the personal information was more favorable than the message information and invoking self-referencing had a positive result. In the present study, when the personal information was less positive than the message information, extensive self-referencing resulted in less favorable evaluations. Future research should investigate the relationship between existing knowledge and the use of self-referencing strategies.

Additional research is needed to determine how instructions might be used to limit or stimulate self-referencing to a desirable level, and perhaps to better focus it. It should be possible to demonstrate that single strategies might result in the same detrimental effects as found with the dual strategy in the current study. On the other hand, dual strategies might be very effective under different circumstances. Finally, it would be useful if a general theory of information processing could be expanded to account for the positive and negative persuasive effects of mnemonic devices usually thought to enhance recall. For example, Calder and Sternthal (1980) have shown that exposing an audience to high levels of repetition and Kisielius and Sternthal (1983) have demonstrated that vividly presenting product information do not always increase persuasion. These findings present difficulties for conventional models of advertising response but are consistent with the availability hypothesis discussed in this paper.

REFERENCES

Anderluh, John R. (1982) "Computerized Ink-jet Printing Lets Marketers Personalize, Target Direct-Response Packages," Marketing News (March 19), p. 10.

Calder, Bobby J. and Brian Sternthal (1980), "Television Advertising Wearout: An Information Processing View," Journal of Marketing Research, 17, 173-186.

Kisielius, Jolita and Brian Sternthal (1982) "Detecting and Explaining Vividness Effects in Social Judgment," Unpublished Manuscript, Northwestern University.

Kraut, Robert E. (1973), "Effects of Social Labeling on Giving to Charity," Journal of Experimental Social Psychology, 551-562.

Leavitt, Clark, Anthony Greenwald and Carl Obermiller (1981), "What is Low Involvement Low In?" in Kent Monroe, ed., Advances in Consumer Research, Vol. 8, Ann Arbor: Association for Consumer Research,

Lord, Charles G. (1980), "Schemas and Images as Memory Aids: Two Modes of Processing Social Information," Journal of Personality and Social Psychology, 38 (February), 257-269.

Markus, Hazel (1977), "Self-Schemata and Processing Information about the Self," Journal of PersonalitY and social Psychology, 35 (February), 63-78.

Miller, Richard L., Philip Brickman, and Diana Bolen (1975), "Attribution versus Persuasion as a Means of Modifying Behavior," Journal of Personality and Social Psychology, 31, 430-441.

Rogers, T. B., N. A. Kuiper and W. S. Kirker (1977), "Self-Reference and the Encoding of Personal Information," Journal of Personality and Social Psychology, 35 (September), 677-688.

Sullivan, Jeremiah J. "The Impact of a Self-Schema on Cognitive Responses to a Persuasive Message and on Affect," unpublished manuscript, University of Washington, 1981.

Tybout, Alice M., Bobby J. Calder and Brian Sternthal (1981), "Using Information Processing Theory to Design Marketing Strategies," Journal of Marketing Research, 18 (February), 73-79.

Tybout, Alice M., and Richard F. Yalch, "Explaining the Effects of Experience: A Matter of Salience?" Journal of Consumer Research, 6 (March), 406-413.

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