Advances in Consumer Research Volume 24, 1997 Pages 229-234
THE PERSUASIVE IMPACT OF AUTOBIOGRAPHICAL MEMORIES IN ADVERTISING: EPISODIC SELF-REFERENCING OR SCHEMA-TRIGGERED AFFECT?
Simani M. Price, Virginia Polytechnic Institute and State University
Danny K. Axsom, Virginia Polytechnic Institute and State University
Eloise Coupey, Virginia Polytechnic Institute and State University
Marketers frequently use ads intended to evoke autobiographical memories to influence consumers evaluations of products. Little empirical evidence exists, however, to explain the processes by which the self-referencing done by consumers during these ads affects persuasion. Moreover, extant studies which build on an episodic, piecemeal view of self-referencing provide inconsistent results on product evaluation.
We propose that autobiographical memories ads are more similar to semantic memory than episodic memory, and that they influence evaluation by invoking schemas that filter self-referencing. A study designed to test the competing explanations supported a schema-based interpretation of the persuasive impact of autobiographical memories in advertising.
Self-referencing has been described as the persuasive mechanism for ads that encourage consumers retrieval of past autobiographical memories (Sujan, Bettman, & Baumgartner, 1993; Baumgartner, Sujan, & Bettman, 1992). The memory is experienced as the representatin of an event at a particular time and location (Brewer & Pani, 1983) and is characterized as having unique cognitive properties of the self (Gergen, 1971). This effect has led researchers to characterize ads which evoke autobiographical memories as representative of episodic rather than semantic memory (Baumgartener et al., 1992). Research has tended to indicate that the presentation of an autobiographical memories ad causes consumers to focus their attention on idiosyncratic experiences associated with a specific past event. This effect leads to an increase in positive affect because people tend to bias their recall in terms of positive memories, and then transfer the positive affect to the product, resulting in a favorable product evaluation (Sujan et al., 1993).
Though this explanation may seem plausible, few studies have actually examined how self-referencing and affect lead to persuasion. Additionally, the studies currently available have found mixed support for the effectiveness of these ads. In one study (Sujan et al., 1993), the researchers found that self-referencing was an effective persuasive technique as it resulted in a higher evaluation of the product compared to a more traditional product focus type of ad. In a different study, the same authors found that the autobiographical memory ad led to a higher evaluation of the ad, but not the product (Baumgartner et al., 1992). In both studies, however, affect levels increased during the presentation of these ads. If self-referencing via autobiographical memories results in a more favorable product evaluation, then one might expect that the more autobiographical memories consumers recall, the more they will like the product. Existing studies, however, have found no link between the proportion of autobiographical memories consumers report and product evaluation. In fact, the results of these two studies suggest that episodic self-referencing of a past event does not occur spontaneously with any great frequency among subjects. Thus, our understanding of the processes which determine the persuasive impact of autobiographical memories ads remains unclear.
In this paper we propose that the persuasive process underlying autobiographical memories ads is more representative of semantic memory or a category-based process rather than episodic memory. Specifically, we use Fiskes schema-triggered affect model to explain how a category-based process might operate and induce persuasion. Further, we hypothesize that retrieving a unique past memory after viewing an autobiographical memories can encourage an attribute-based process and reduces persuasion. In the following sections we develop and test hypotheses about the nature of the processes by which autobiographical memories ads may influence persuasion through affect and cognition. To develop our argument for a semantic memory explanation, we briefly review two models that have implications for the reciprocal processes of cognition and affect.
Affect in a Category-based versus An Attribute-based Process
The prevailing view in social cognition is that people are cognitive misers, and as such, a category-based rather an attribute-based approach is used initially (Fiske & Taylor, 1991). One model that has direct implications for affect in a category-based process is Fiskes (1982) schema-triggered affect model. Fiske suggests that to the extent that an instance is perceived to fit the schema, it will receive the affect linked to that category. Otherwise, the instance receives a moderate positive affect by default, pending its categorization as a good example of something else. In this model, affect is stored at the top level of the schematic structure without access to all the categorys features and their respective evaluations; affect is linked to the initial act of categorization. Thus, stimulus that is perceived as a consistent (vs. inconsistent) schema-match will be evaluated as similar to the category label. This matching effect has been demonstrated for a variety of person schemas: old flames, politicians, college majors, occupations (Fiske & Pavelchak, 1986). In a marketing context, the schema-triggered affect model suggests that if a consumer can easily categorize a product initially into a previously learned category label (e.g., Florida spring break package), then knowledge about the category becomes more accessible than any one attribute of the product (e.g., cost). As a result, product evaluation is determined by the category label, rather than by product attributes.
Positive affect can also encourage a category-based process. Isen and Daubman (1984) report that positive affect tends to broaden categories, leading subjects to see items as more similar than do subjects in a neutral mood. In addition, research suggests that people in a positive mood exhibit little systematic processing of information and instead rely on less effortful judgment heuristics (e.g., category-label) (Mackie & Worth, 1991). This finding suggests that consumers with an increase in positive affect would be more likely to engage in a schema-based process in evaluating a product. The preceding discussion suggests that affect should have significant impact in a category-based process, with schema match influencing consumer judgments. Thus, we would expect a favorable category-label to result in a higher product evaluation for a schema-consistent (vs. inconsistent) product.
In contrast to a category-based approach, affect is not as influential or as extreme in judgments using an attribute-based approach (Linville, 1982). Attention to the stimulus has to be increased to encourage an attribute-based approach (Fiske & Neuberg, 1990). One way to encourage attention is to increase the number of dimensions that are considered during judgments. This manipulation attenuates the role of affect and encourages an attribute-based process according to Linvilles (1982) self-complexity model. Specifically, Linville proposes that the greater the number of dimensions considered, the greater the likelihood that the affect associated with any given dimension is attenuated during a judgment task. Thus, evaluation is less likely to be extreme (either positively or negatively) under an attribute-based process compared to a category-based process. Complexity or the number of independent dimensions underlying a judgment or cognitive structure, can be either manipulated directly (e.g., by instructions to consider x number of factors in a decision) or measured as individual differences. Unlike a category-based judgment, dimensionality is an important component of an attribute-based judgment. For example, a consumer can elaborate extensively on a product category and increase the number of thoughts (i.e., quantity of thoughts) but not the number of dimensions (i.e., quality of thoughts) considered, thus using a schema-based rather than an attribute-based process to evaluate a product. A schemas structure is more intrinsically, coherent, which results in a significant correlation between the schemas dimensions (Tesser, 1978). In an attribute-based process, each dimension independently contributes to the evaluation of the product. Thus, product categorization as schema-consistent or inconsistent and the affect associated with the category-label is less influential in determining product evaluation.
Persuasive Effects of Autobiographical Memories
We suggest that persuasive effects from autobiographical memories ads result from a category-based process. Autobiographical memories ads initially encourage attention to the self or self-focus. This focus simultaneously increases positive affect, as people tend to view themselves positively (Bradley, 1978), thus encouraging a category-based process in judgments. An increase in self-focus is a different explanation than episodic memory retrieval. The episodic memory explanation requires that a unique memory as been retrieved (Gergen, 1971). In addition, reduced persuasion may occur if a unique past experience is retrieved after viewing an autobiographical memories ad because the ad may serve as a reference point and encourage an attribute-based process. Markus & Sentis (1982) note that information is automatically encoded in a top-down manner in relation to the self until a mismatch occurs. We hypothesize that if people compare their experience to the ad, a mismatch would likely occur. Specific hypotheses designed to compare the schema-based, semantic memory explanation with an episodic memory explanation are detailed in the next section. An overview of the study is provided first to facilitate presentation of the hypotheses.
The goal of this study was to provide a theoretical explanation for the persuasive effects of autobiographical memories ads by juxtaposing our current understanding (i.e., episodic memory) with an alternative explanation (i.e., semantic memory). Fiskes schema-triggered affect model was used to examine the semantic memory explanation. Undergraduate subjects were exposed to a Florida vacation package ad for either Spring (schema-consistent) or Thanksgiving (schema-inconsistent) Break. In addition to Schema-match, Self-referencing was also manipulated where subjects either viewed a product-focus or a self-referencing (i.e., autobiographical memories) ad. The self-referencing manipulation resulted in three different variations (low self-referencing, high self-referencing pre-ad, high self-referencing post-ad). The Low Self-referencing condition was similar to previous studies examining autobiographical memories ads (Sujan et al., 1993) and was hypothesized to encourage a category-based approach. The two High Self-referencing conditions varied by whether they required subjects to write an essay about a past break experience before (high self-referencing pre-ad) or after (high self-referencing post-ad) viewing the ad. The High Self-referencing Pre-ad condition was hypothesized to encourage a category-based approach, as subjects could not use the ad as a source of comparison while thinking about their past experience. In contrast, the High Self-referencing Post-ad condition was hypothesized to encourage an attribute-based approach, the ad should serve as a reference point for subjects while thinking about their past experience. The presentation order of two primary dependent measures, the product evaluation measure and a cognitive response measure, was also manipulated to enable assessment of procedural bias. All other dependent variables followed these primary dependent measures.
In the category-based explanation, the Low Self-referencing and High Self-referencing Pre-ad conditions should encourage self-focus initially and as such, simultaneously increase positive affect and the use of a schema-based process relative to the Product-focus condition. As Fiskes (1982) schema-triggered affect model suggests, the affect associated with the category-label should influence both product evaluation and subsequent ad-induced thoughts. Thus, a consistent schema-match (spring break in Florida) should have a higher product evaluation than an inconsistent schema-match (Thanksgiving break in Florida).
A category-based process may operate more strongly for subjects in the High Self-referencing Pre-ad condition as a result of having a schema primed (writing about a past break experience) prior to viewing the ad. The essay for subjects in the High Self-referencing Pre-ad condition should further increase positive affect, as people tend to recall primarily positive memories (Wagenaar, 1986). Because the self-referencing essay was written prior to viewing the ad, the ad could not be used as a source of comparison, thus reducing the likelihood of a mismatch occurring between their unique experience and the ad. As participants engage in a category-based process, their idiosyncratic experiences sould be assimilated in the schema-consistent condition and contrasted in the schema-inconsistent condition. The preceding rational for a schema-based process is summarized as:
H1: In the Consistent Schema-match, the Low Self-referencing and High Self-referencing Pre-ad conditions should have a higher product evaluation than the Product-focus condition.
H2: The Consistent Schema-match should have a higher product evaluation than the Inconsistent Schema-match in both the Low Self-referencing and High Self-referencing Pre-ad conditions
An episodic memory explanation, in contrast, would predict that product evaluation should be influenced by the degree of episodic memory retrieval that occurs, regardless of degree of schema-match. Therefore, it would simply predict that product evaluations would be higher in the High Self-referencing Pre-ad conditions than the Low Self-referencing and Product-focus conditions in both Schema consistent and inconsistent conditions.
To support our hypothesis that persuasion effects results from a category-based process and not due to the retrieval of a unique self-referenced memory per se, we need to demonstrate reduced persuasive effects during an attribute-based process. Specifically, we need to provide evidence that attention to a past memory under a category-based process leads to a higher product evaluation than during an attribute-based process. To encourage attention to a past memory, subjects were asked to write on multiple dimensions of a past break experience either before or after viewing the autobiographical memories ad in this study. As noted earlier, we hypothesize that writing the self-referencing essay before viewing the ad would lead to a category-based process for subjects in the High Self-referencing Pre-ad condition. In contrast, we hypothesize that writing about a past break experience after viewing the ad, would encourage an attribute-based process for subjects in the High Self-referencing Post-ad condition. The ad presents multiple dimensions of a vacation package (i.e., activities, lodging, etc.) and should serve as a reference point for subjects writing about their own unique experience after viewing the ad. Recall that a characteristic of an attribute-based process is dimensionality (i.e., quality of thoughts) (Linville, 1982). Thus, we expect subjects in the High Self-referencing Post-ad condition to discuss more dimensions on the self-referencing essay compared to subjects in the High Self-referencing Pre-ad condition if they are using the ad as a reference point. As subjects in the High Self-referencing Post-ad condition consider multiple dimensions of their past experience, they will likely compare and contrast this with the ad, resulting in a mismatch between the ad and their experience. This should lead subjects to view the ad as less representative of their past experience. Further, as subjects engage in an attribute-based process, product category (schema consistent vs. inconsistent) and the affect associated with the category-label should have less influence on product evaluation. This will result in the product being evaluated less extremely (either positively or negatively) regardless of schema-match. The preceding discussion can be summarized as follows:
H3: In the Consistent Schema-match, the High Self-referencing Pre-ad and Low Self-referencing conditions should have a higher product evaluation than the High Self-referencing Post-ad condition.
H4: The Consistent Schema-match should not differ significantly from the Inconsistent Schema-match in the High Self-reerencing Post-ad condition.
The preceding predictions regarding the influence of the self-referencing essay stand in contrast to what might be predicted from an episodic memory perspective. If the underlying persuasive mechanism of autobiographical memories ads is determined primarily by episodic memory, then increasing attention to a unique self-referenced memory should result in the highest product evaluation. From this perspective, it is assumed that people generally remember positive experiences from their life, and thus increasing attention to a past event would bias people to consider even more positive experiences associated with a past episode. Thus, from an episodic memory perspective, the two High Self-referencing (Pre-ad and Post-ad) conditions would similarly have the highest product evaluation (irrespective of schema-match) compared to the Low Self-referencing condition. In both of these conditions, attention to a unique past experience is increased by writing the self-referencing essay.
Subjects and Procedure
Two hundred and forty undergraduates enrolled in psychology and marketing classes at a southeastern university received extra credit for participation in this study. Stimuli were presented on computers with VGA color monitors. The study used a 4 (Self-referencing: None (Product), Low, High Pre-ad, High Post-ad) x 2 (Schema Match: Consistent, Inconsistent) x 2 (Response Order: Prod Eval First, Cog Resp First) between-subjects factorial design. Subjects were randomly assigned to one of 16 conditions and completed the study in individual cubicles. See Table 1 for experimental design.
Self-referencing. Subjects in the Low, High Pre-ad and High Post-ad Self-referencing conditions were presented with an ad that asked them to think back to a past spring (or Thanksgiving) break and provided information on product features. In the two High Self-referencing conditions (Pre-ad, Post-ad), subjects were required to write on multiple dimensions (e.g., social, planning, travel) of a past spring (or Thanksgiving) break experience. Subjects either wrote the essay before (High Self-referencing Pre-ad condition) or after (High Self-referencing Post-ad condition) viewing the ad. Subjects in the Product-focus condition were also presented with an ad for a Florida spring (or Thanksgiving ) break package but were only provided information on product features with no reference to a past break. Subjects in the Low Self-referencing and Product-focus conditions wrote an essay on the multiple uses and dimensions of paper towels (filler task) after viewing the ad. Subjects had five minutes to complete their essays.
Schema-match. Based on the results of pretests, a Florida Spring break was used as the Schema-consistent manipulation and a Florida Thanksgiving break was used as the Schema-inconsistent manipulation in all conditions.
Response Order. Subjects either received the product evaluation questions first followed by the cognitive response measure or received the cognitive response measure first followed by the product evaluation questions. The other dependent variables (i.e., episodic event rating, schema-match ratings) followed these measures.
Cognitive Response Measure. Subjects were provided two minutes to list all the thoughts that went through their minds while viewing the break package ad. Two independent judges coded thoughts into categories and alence. The average interjudge reliability was 95% across all measures. Disagreements were resolved by discussion.
Product Evaluation. Product evaluation was assessed on four 9-point semantic differential scales (i.e., favorable-unfavorable, good-bad, pleasant-unpleasant, positive-negative). The four measures were averaged to form an overall measure of product evaluation (alpha=.82).
Self-referencing Essay. Two independent judges coded thoughts into dimensions (e.g., social, planning, travel). The average interjudge reliability was 90%. Disagreements were resolved by discussion.
Reaction Time. Response latencies were collected for all dependent measures that required ratings on a Likert-type scales. Latencies were rounded to milliseconds.
Schema-Match Ratings. Subjects made ratings on a 9-point scale on whether college students spending spring and Thanksgiving break in Florida was not at all typical (1) or very typical (9).
Episodic Event Rating. Subjects indicated on a 9-point scale whether, while viewing the ad, they thought about a specific spring (or Thanksgiving) break experience from their past (1) or thought of a general idea of what spring (or Thanksgiving) breaks are like for college students (9).
EXPERIMENTAL DESIGN: SCHEMA-MATCH, SELF-REFERENCING AND RESPONSE ORDER
Personal and Involving the Self Rating. Subjects indicated on a 9-point bipolar scale whether their thoughts while viewing the ad could be described as (1) impersonal not involving the self or personal involving the self (9) (Sujan et al., 1993).
Related to a Past Break Experience Rating. Subjects indicated on a 9-point bipolar scale whether their thoughts while viewing the ad could be described as (1) not related to past spring (or Thanksgiving) break experience or related to past spring (or Thanksgiving) break experience (9) (Sujan et al., 1993).
RESULTS AND DISCUSSION
Schema-Match Ratings. An ANOVA indicated a main effect for Typicality, F (1, 223)=666.89, p <.001. The Florida spring break was rated as significantly more typical (M=7.2) than the Florida Thanksgiving break (M=2.9).
Episodic Event Rating and Reaction Time. An ANOVA indicated no effects for the episodic event rating (grand mean=5.9). However, an ANOVA on the reaction time for the episodic event rating produced a main effect for Self-Referencing, F (3, 224)=3.58, p<.02. Subjects in the High Self-referencing Post-ad condition displayed significantly quicker response latencies (M=9.7) compared to subjects in both the Low Self-referencing condition (M=11.2), t (118)=1.98, p<.05 and the Product-focus condition (M=11.7), t (118)=2.82, p<.01, but were not significantly different from subjects in the High Self-referencing Pre-ad condition. Response latencies are often used as an indication of accessibility of thought processes (Bargh, 1984). Thus, subjects should respond quicker to items they are already thinking about. These results may indicate that writing about a unique past experience made a specific break experience more accessible for subjects in the two High Self-referencing conditions compared to other self-referencing conditions.
Related to a past break experience. An ANOVA on "related to a past break experience" rating indicated a marginal main effect for Schema-match, F (1,224)=2.95, p<.09. Subjects in the Schema-consistent condition rated the ad as being more related to a past break experience (M=4.7) compared to subjects in the Schema-inconsistent condition (M=4.1). Post-hc comparisons between the Schema-consistent and the Schema-inconsistent conditions indicated that the only significant difference occurred in the Product-focus condition. In the Product-focus Prod Eval First condition, subjects with the Consistent Schema-match rated the ad as significantly more "related to a past break experience" (M=6.2) compared to subjects with the Inconsistent Schema-match (M=3.4), t (224), p<.05. Within the other comparable Self-referencing conditions, the Schema-match comparisons did not significantly differ on this rating. These results suggest that the autobiographical memories ads may have encouraged self-focus initially compared to the product focus ads. That is, the autobiographical memories ad may have increased attention to the self regardless of schema-match. The increased attention to the self is hypothesized to increase positive affect and as such, encourage the use of a category-based process by participants initially.
Personal and Involving the Self Rating. An ANOVA on the "personal and involving the self" rating indicated a marginal main effect for Self-referencing, F (3, 222)=2.49, p<.06. Post hoc protected t-tests indicated that subjects in the High Self-referencing Post-ad condition rated their thoughts as significantly more impersonal and not involving the self while viewing the ad (M=5.6) compared to subjects in the Low Self-referencing condition (M=6.6), p<.05, the High Self-referencing Pre-ad condition (M=6.6), p<.05 and the Product-focus condition (M=6.5), p<.05. These results suggest that a mismatch likely occurred for subjects in the High Self-referencing Post-ad condition, leading them to view the ad as less personal compared to the other Self-referencing conditions. A mismatch was hypothesized to encourage an attribute-based process for this condition.
An ANOVA procedure revealed a three-way interaction between Schema-match, Self-referencing, and Response Order, F (3, 221)=4.02, p< .008. Because there were no a priori predictions regarding the impact of Response Order on product evaluation, results were examined separately for the different response orders.
When product evaluations were collected before cognitive responses, results indicated a 2-way interaction between Schema-match and Self-referencing, F (3,109)=4.70, p<.004. Post-hoc protected t-tests provided support for the schema-triggered affect model.
As predicted by hypothesis 1, in the Consistent Schema-match condition, both the Low Self-referencing and High Self-referencing Pre-ad conditions had a significantly higher product evaluation compared to the Product-focus condition (M=6.0), t (221)=2.91, p<.01. Additionally, and as predicted by hypothesis 3 in the Consistent Schema-match, both the Low Self-referencing (M=7.5) ) and High Self-referencing Pre-ad (M=7.3) conditions had a significantly higher product evaluation compared to the High Self-referencing Post-ad condition (M=6.3), t (221)=2.03, p<.05.
Product evaluation comparisons for schema-consistent versus inconsistent cells within each Self-referencing condition also provided some support for the schema-triggered affect model when product evaluations were collected before cognitive responses. There was partial support for hypothesis 2. Product evaluation in the Low Self-referencing condition, was significantly higher in the Consistent Schema-match (M=7.5) than in the Inconsistent Schema-match (M=6.5), t (221)=2.0, p<.05. In the High Self-referencing Pre-ad condition, product evaluation was directionally higher in the Consistent Schema-match (M=7.3) than the Inconsistent Schema-match (M=6.7). The difference was only significant, however, in one of the four product evaluation questions (unpleasant-pleasant dimension), where the High Self-referencing Pre-ad Schem-consistent condition was rated as significantly more pleasant (M=8.0) compared to the High Self-referencing Pre-ad Schema-inconsistent condition (M=6.9), t (221)=2.15, p<.05.
As predicted in hypothesis 4, product evaluation in the High Self-referencing Post-ad condition, was not significantly different in the Consistent Schema-match (M=6.2) compared to the Inconsistent Schema-match (M=6.8). Interestingly, the results for the Product-focus condition indicated that product evaluation in the Inconsistent Schema-match was significantly higher (M=7.4) than the Consistent Schema-match (M=6.0), t (221)=2.69, p<.01. One potential explanation for the higher product evaluation in the Product-focus Schema-inconsistent condition is the novelty of a Thanksgiving break package in Florida. Further, because the self was not initially primed in the product-focus-ad as it was in the self-referencing-ads, the package was not viewed as inconsistent with the self in Florida for a Thanksgiving break; it was merely evaluated as a novel product and not inconsistent with a family vacation experience.
When product evaluations were collected after cognitive responses, there were no significant differences in the product evaluation across conditions.
Self-referencing and Schema-match Post-hoc Comparisons. The results suggested that the presentation of the cognitive response measure prior to the product evaluation measure encouraged all subjects to engage in an attribute-based approach and resulted in similar outcomes across all dependent measures in this study. Further, the schema-triggered affect model was most strongly observed when making self-referencing comparisons in the schema-consistent (vs. inconsistent) conditions. The schema-inconsistent conditions encouraged all participants to engage in an attribute-based process, as a category-label was not readily available. As a result of these factors and for the sake of parsimony, post-hoc comparisons for the Self-referencing conditions are only presented for the Schema-consistent (vs. Schema-inconsistent) conditions all for the product evaluation first condition. Similarly, Schema-match post hoc comparisons are only presented for the product evaluation first condition.
Positive Affect and Product Evaluation Correlation. If subjects in the High Self-referencing Pre-ad and Low Self-referencing conditions were similarly using a schema-based process, then positive affect should strongly influence their judgments. In contrast, the High Self-referencing Post-ad condition subjects were hypothesized to be using an attribute-based process, thus positive affect should be less influential in making judgments. The correlation between positive affect and product evaluation provide empirical support for the two information processing models operating in the different conditions. The following correlations represent the Schema-consistent Prod Eval First condition. The percent of positive thoughts on the cognitive response measure and product evaluation were significantly positively correlated in the High Self-referencing Pre-ad (r=.52, p<.05) and Low Self-referencing conditions (r=.71, p<.01). Subjects in the High Self-referencing Post-ad condition had little correlation between product evaluation and the percent of positive thoughts (r=-.16, p<.70). A Fishers z transformation indicated that the High Self-referencing Post-ad condition was significantly different from the Low Self-referencing condition (z=2.45, p< .01) and the High Self-referencing Pre-ad condition (z=1.72, p<.05).
Percent of Florida Vacation Imagery Thoughts on Cognitive Responses. Florida vacation imagery thoughts represented a general reference to a Florida vacation with no reference to the self (i.e., bikini, beer etc.). Results indicated a main effect for Self-referencing, F (3, 218)=3.15, p<.05 in the percent of Florida vacation imagery thoghts mentioned. Post hoc protected t-tests indicated that the High Self-referencing Pre-ad condition (35%) was significantly higher in the percent of Florida vacation imagery thoughts compared to the Low Self-referencing condition (19%), p<.05 and directionally higher than the High Self-referencing Post-ad (26%) and Product-focus (25%) conditions. These results appeared perplexing if the Low Self-referencing and High Self-referencing Pre-ad conditions were similarly using a schema-based process as the product evaluation data suggested. To test this hypothesis further, self-referencing comparisons were made in those cells where the likelihood of a schema-based process was strongest (Schema-consistent Prod Eval First conditions). Post-hoc protected t-tests indicated that the percent of Florida vacation imagery thoughts in the High Self-referencing Pre-ad condition (42%) was not significantly different from the Low Self-referencing condition (24%) but was significantly higher than the High Self-referencing Post-ad condition, (18%), t (218)=2.15, p<.05 and the Product-focus condition (15%), t (218)=2.15, p< .05. These results suggest that category-related thoughts during the ad in the High Self-referencing Pre-ad and Low Self-referencing conditions were more similar then originally indicated by the main effect. However, the Low Self-referencing condition was not significantly different from any of the self-referencing comparisons. One possible explanation is that a schema-based process was even stronger for subjects in the High Self-referencing Pre-ad condition compared to the Low Self-referencing condition.
Number of Dimensions Discussed on the Self-referencing Essay. An ANOVA indicated a main effect for Self-Referencing, F (1, 109)=7.69, p<.007 in the number of dimensions discussed on the essay. Subjects in the High Self-referencing Post-ad condition discussed significantly more dimensions (M=4.7) compared to subjects in the High Self-referencing Pre-ad condition (M=4.1). Subjects in the High Self-referencing Post-ad condition appeared to use the ad as reference point, leading them to consider more dimensions of their own past experience. An increase in dimensionality was hypothesized to encourage an attribute-based process for subjects in this condition.
Dimensions Discussed and Product Evaluation Correlation. If subjects in the High Self-referencing Pre-ad condition engaged in a category-based process, their unique experiences may be influenced by the schema-label (consistent vs. inconsistent). In contrast, High Self-referencing Post-ad condition subjects were expected to have engaged in an attribute-based process regardless of schema-match. Correlations between the number of dimensions discussed on the self-referencing essay and product evaluation provide evidence for the two information processing models. The following represent correlations for the Prod Eval First conditions. In the High Self-referencing Pre-ad Schema-consistent condition, the number of dimensions of a past spring break experience (M=4.2) was significantly positively correlated with product evaluation (r=.69, p<.01), suggesting that subjects were more likely to assimilate their idiosyncratic past experiences with the ad. In the High Self-referencing Pre-ad Schema-inconsistent, the number of dimensions of a past Thanksgiving break experience (M=4.3) was significantly negatively correlated with product evaluation (r=-.59, p<.02), suggesting that subjects were more likely to contrast their past experiences with ad. The number of dimensions discussed on essay in the High Self-referencing Post-ad condition had little influence on product evaluation in either the Schema-consistent (M=4.7, r=-.05) or Schema-inconsistent (M=4.3, r=.05) conditions. A comparison of the correlations using Fishers z transformation indicated that in the Schema-consistent condition, the High Self-referencing Post-ad condition was significantly different from the High Self-referencing Pre-ad condition (z=2.20, p<.01). Similarly, in the Schema-inconsistent condition, the High Self-referencing Post-adcondition was significantly different from the High Self-referencing Pre-ad condition (z=1.79, p<.05).
The pattern of results suggests that a category-based process can be used to explain the underlying persuasive mechanism in autobiographical memories ads, rather than self-referencing based on episodic memory. Autobiographical memories appear to increase self-focus and positive affect, and as such they encourage a schema-based process. When a unique self-experience is retrieved, the ability to assimilate ones idiosyncratic experiences and engage in a category-based process is the critical component in inducing persuasion. If a unique past experience is retrieved after viewing the ad, the ad is used as a reference point. This increases the likelihood of a mismatch occurring, thus triggering an attribute-based process which may reduce the ads effect on persuasion. The results of this study suggest that autobiographical memories ads can be an effective persuasive technique, but that information relating to the self or to a past experience should be kept at a general level so that consumers can assimilate any of their unique experiences with the ad.
Bargh, J. A. (1984). Automatic and cognitive processing of social information. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition. (Vol. 3, pp. 1-43). NJ: Erlbaum.
Baumgartner, H., Sujan, M., & Bettman, J. R. (1992). Autobiographical memories, affect and consumer information processing, Journal of Consumer Psychology, 1, 53-82.
Bradley, G. W. (1978). Self-serving biases in the attribution process: A reexamination of the factor fiction question. Journal of Personality and Social Psychology, 36, 56-71.
Brewer, W. F. & Pani, J. R. (1983). The structure of human memory. In G. H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 17, pp. 1-38). New York: Academic.
Fiske, S. T. (1982). Schema-triggered affect: Applications to social perception. In M. S. Clark & S. T. Fiske (Eds.), Affect and cognition: The 17th Annual Carnegie Symposium on Cognition (pp. 55-78). NJ: Erlbaum.
Fiske, S. T. & Pavelchak, M. A. (1986). Category-based versus piecemeal-based affective responses: Developments in schema-triggered affect. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: Foundations of social behavior (pp. 167-203). New York: Guilford Press.
Fiske, S. T., & Taylor, S. E. (1991). Social Cognition. New York, McGraw-Hill, Inc.
Isen, A. M. & Daubman, K. A. (1984). The influence of affect on categorization. Journal of Personality and Social Psychology, 47, 1206-1217.
Gergen, K. J. (1971). The Concept of Self. New York: Holt, Rinehart & Winston.
Linville, P. W. (1982). Affective consequences of complexity regarding the self and others. In M. S. Clark & S. T. Fiske (Eds.), Affect and cognition: The 17th Annual Carnegie Symposium on Cognition (pp. 79-109). Hillsdale, NJ: Erlbaum.
Mackie, D. M. & Worth, L. T. (1991). Feeling good, but not thinking straight: The impact of positive mood on persuasion. In J. Forgas (Ed.), Emotion and social judgments (pp. 201-219). Oxford, England: Pergamon Press.
Sujan, M., Bettman, J. R. & Baumgartner, H. (1993). Influencing consumer judgments using autobiographical memories: A elf-referencing perspective. Journal of Marketing Research, 30, 422-436.
Markus, H. & Sentis, K. (1982). The Self in Social Information Processing. In J. Suls (Ed.), Psychological Perspectives of the Self (pp. 41-70). NJ.: LEA.
Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experimental social psychology, (Vol. 11, 289-338). New York: Academic Press.
Wagenaar, W. A. (1986). My memory: A study of autobiographical memory over six years. Cognitive Psychology, 18, 225-252.