Automatic Activation of the Self in a Persuasion Context

ABSTRACT - It was hypothesized that priming the self-concept in a persuasion context results in more peripheral information processing because people would rely more on their self-concept. Subjects were presented an ad for an unknown brand of sport shoes that either primed or did not prime subjects’ self-concept. Results showed that when the self was primed, subjects elaborated more on atypical arguments as compared to typical arguments, whereas total amount of elaboration remained unaffected. Moreover, attitudes towards the brand were more extreme (and favorable), while confidence in these attitudes was lower. These results are consistent with our hypothesis.

Citation:

Anick Bosmans, Peter Vlerick, Patrick Van Kenhove, and Hendrik Hendrickx (2000) ,"Automatic Activation of the Self in a Persuasion Context", in NA - Advances in Consumer Research Volume 27, eds. Stephen J. Hoch and Robert J. Meyer, Provo, UT : Association for Consumer Research, Pages: 274-278.

Advances in Consumer Research Volume 27, 2000      Pages 274-278

AUTOMATIC ACTIVATION OF THE SELF IN A PERSUASION CONTEXT

Anick Bosmans, Ghent University

Peter Vlerick, Ghent University

Patrick Van Kenhove, Ghent University

Hendrik Hendrickx, Ghent University

ABSTRACT -

It was hypothesized that priming the self-concept in a persuasion context results in more peripheral information processing because people would rely more on their self-concept. Subjects were presented an ad for an unknown brand of sport shoes that either primed or did not prime subjects’ self-concept. Results showed that when the self was primed, subjects elaborated more on atypical arguments as compared to typical arguments, whereas total amount of elaboration remained unaffected. Moreover, attitudes towards the brand were more extreme (and favorable), while confidence in these attitudes was lower. These results are consistent with our hypothesis.

INTRODUCTION

Consumer research has paid little attention to the role of automatic processes in a persuasion context. Although people have a limited processing capacity (Kardes 1998), they are overwhelmed with all sorts of information. This problem is also observed within a marketing-related context: consumers are overloaded with all sorts of information and it has become extremely difficultfor marketers to capture consumers’ attention. Given the extreme costs associated with most communication attempts, the need to develop more effective communication strategies arises.

In this paper we investigate the role of automatic processes in a persuasion context. Automatic processes can be considered as helping devices that enable people to deal with these large amounts of information on a rather efficient way. Indeed, automatic processes, as opposed to conscious or control processes, are not restricted to limited information processing capacities and their cause is independent of, and parallel to other cognitive processes (Logan 1979; Shiffrin and Schneider 1977). Taken from this point of view, automatic processes (although unconscious) can be considered as heuristics that enable people to deal with the complex society we live in.

Relying on the current persuasion literature, the purpose of the present study is to investigate the influence of these automatic processes on both message elaboration and attitude formation. More specifically, we will concentrate on automatic processes associated with self-relevant information. It was hypothesized that confronting people with self-relevant information in a persuasion context will result in automatic activation of their self-concept. As a consequence, this will result in relatively more reliance on their self-concepts in both message elaboration and attitude formation. Because to our knowledge, no previous studies are conducted with regard to the influence of automatic processing of self-relevant information on persuasion, we will first review some relevant insights derived from the social psychology literature.

AUTOMATIC ACTIVATION OF SELF-RELATED CONSTRUCTS

Social psychologists have found that when the self-concept or the self-schema is confronted with information relevant to it (i.e. self-relevant information), it is automatically activated, and the self-relevant information is processed automatically (Bargh 1982; Symons and Johnson 1997). The self is described in the literature as a cognitive schema that represents knowledge of oneself that is rich and plentiful, and that contains many associations that can be related to the incoming information, such as the information in the ad (Markus 1980). By a network of spreading activation, the self provides meaning to incoming stimuli if those stimuli fit one’s knowledge structures in memory (Klein and Loftus 1996). Bargh and Pietromonaco (1982) found that when a personality trait was unconsciously presented to subjects, subjects later on judged an unknown person more conform to this trait when this trait was self-relevant to the subjects in comparison to when it was not. This automatic processing of self-relevant information can be explained by the fact that the self-concept, or the self-schema, automatically screens and/or processes that information that is relevant for it (Kuiper and Derry 1981). Moreover, the self as a cognitive structure can be seen as a chronic accessible category (Higgins and King 1981; Higgins, King and Mavin 1982). Certain categories of social stimuli are indeed that frequently experienced such that they are eventually able to automatically activate their abstract mental representations. Indeed people constantly have experiences with the self as a relevant focus.

THE SELF AND PERSUASION

The aim of the present study is to explore the effects of this automatic activation of self-relevant information in an advertising context. Modern society is suffering from an advertising overload, and because of this overload it has become extremely difficult for marketers to reach the public through advertising campaigns. Indeed consumers, with their limited processing capacities, are unable and/or unwilling toattend to all marketing stimuli presented to them. Given that self-schemas can be automatically activated or primed, the need arises to explore the effects of self-relevant information on both message elaboration and attitude formation.

Because automatic activation of the self can be seen as a heuristic processing device, we can expect that when people are confronted with self-relevant information they will follow a more peripheral processing route as opposed to people who are not confronted with this self-relevant information. They will in other words, rely more on their self-schema, because it has recently been primed through confrontation with information relevant to it. Indeed, mental representations are more likely to be used if their accessibility is increased (Wyer and Srull 1989). According to the Elaboration Likelihood Model of persuasion (e.g. Petty and Cacioppo 1986; Petty, Cacioppo and Schumann 1983), adopting a more peripheral processing route will result in less elaboration, and as a consequence, less arguments of the ad will be remembered in comparison with when a more central processing route is adopted.

Central or systematic processing will lead to attitudes that are relatively enduring, resistant to change, and predictive for behavior whereas peripheral or heuristic processing leads to attitudes that are not very enduring, not resistant to change and relative unpredictive for behavior. Whether the central route is followed depends on both motivation to process (such as involvement) and ability to process (such as skills or professional knowledge). These conditions are non-compensatory (Batra and Ray 1986; MacInnis and Jaworski 1989).

It should be noted however, that we do not expect that relying on mental representations or general knowledge structures in the persuasion process will reduce elaboration of all kinds of arguments. Given the fact that elaboration of the message content requires a considerable amount of processing, and it is more economic for subjects to rely on already existing knowledge (because it has proven to be useful), subjects will elaborate less on knowledge that is experienced as #already known’ (i.e. typical information) (see Bless, Mackie and Schwarz 1992). However, when confronted with atypical information, it can be expected that subjects will allocate more processing resources to this atypical information, because this information does not fit into their knowledge structures. As argued by Bless (Bless et al. 1992), the use of different representations may depend on the degree to which the perceiver chooses to or is able to allocate resources to the persuasion process. In other words, people are not blind followers of their mental knowledge structures, in contrast they are to a certain degree able to choose whether or not they will rely on these mental representations.

In addition, research on the use of general knowledge structures, as is the self-schema, suggests that attitudes formed through reliance on these well associated mental constructs are more extreme in nature than those formed through extensive elaboration (Judd and Lusk 1984; Linville 1982). As is argued, relying on schemata will prevent moderating or inconsistent information to mitigate attitude favorability. We can thus expect that priming the self through presentation of self-relevant information will lead to attitudes that are more extreme than attitudes that are formed when the self is not primed.

Based on these considerations, we made the following hypotheses in order to gain more insight into the role of automatic processes within a persuasion context. First, when the self is primed by confrontation with a self-relevant personality trait, subjects will follow a more heuristic route to persuasion. This will result in less elaboration for arguments that are considered as typical for a given product category (because they already are associated with the self-concept), whereas no priming effect will be expected for atypical arguments. Because the latter arguments are new to subjects, they will allocate more processing resources to these arguments. We can thus expect that the ratio of recalled atypical arguments to typicl arguments (atypical / (typical + atypical)) (see also Batra and Stayman, 1990) will be greater when the self is primed versus when it is not. In contrast, no effect is expected of self-relevance on the total amount of elaboration (typical + atypical).

Second, with regard to attitude formation it can be expected, as argued before, that attitudes will be more extreme when people rely on their general knowledge structures, because in this case moderating or inconsistent information is less likely to mitigate the attitudes.

Finally, because it is argued that activation of the self will result in a more heuristic processing mode, it can be expected that, although attitudes formed under this condition will be more extreme, they will be held with less confidence. Indeed confidence by which attitudes are held is lower when these attitudes are formed through heuristic processing whereas confidence will be greater when a more systematic processing route is followed (Berger 1992).

METHOD

Design

Subjects were 151 undergraduate university students who participated in partial fulfillment of a class requirement. We decided to consider only those subjects in the study who were highly involved with the product category (sport shoes were chosen as the relevant product category). This is because it is expected that self-relevance will have no or little effect when subjects are low involved with the product category. After all, because they are low involved, it can be assumed that their self-schemas have no or little associations that are related to the product category (Klein and Loftus, 1996). Because we used university students in our sample, we argued that all subjects would be sufficiently able to process the information presented to them. Involvement was measured using 5 seven-point items as suggested by Mittal (1995). Scores on these items were then averaged (alpha=.97). Only subjects with an involvement score greater than 4 were included in the study. This resulted in an eventual sample size of 63 subjects.

The independent variable was self-relevance. A personality trait was presented that was either highly descriptive for people’s self-concept or not. We choose the trait #flexible’ as the trait of interest. A pilot study showed that this personality trait had a normal distribution with regard to an undergraduate student sample: measured on a 7-point scale the median was situated around the value 4 (in other words, this is a trait that a great part of the sample would find highly descriptive for themselves, while the opposite is true for the other part). After measuring self-relevance of the personality trait #flexible’, a median split was performed in order to come to two between subject conditions (as done in prior research, see Batra and Stayman, 1990).

The dependent variables were (a) total amount of elaboration (typical + atypical arguments recalled), (b) ratio of atypical arguments recalled (atypical / (typical + atypical)), (c) attitude formation towards the brand, and (d) confidence in the attitudes as a predictor of behavior (Berger, 1992). The data were analyzed using t-tests for independent samples using self-relevance as a grouping variable.

Procedure

An ad was presented to subjects that both contained the self-relevant prime used in the head liner (#You are flexible, you wear Roots’), whereby Roots was the fictive brand name for the sport shoes, as well as a number of typical and atypical arguments for sport shoes. These arguments were derived from some prior done qualitative interviews. Subjects were asked to reason about features that were and weren’t typical for sport shoes. These interviews delivered a great number of typical and atypical argumnts that were later on classified by a sample of experts. They were asked to rate the degree of typicality of a large sample of arguments (#how typical is for sport shoes’) on a 7-point scale. Only those items that had the lowest and highest average scores on typicality were used in the experiment (intraclass correlation=.67). In total, 12 arguments were used in the ad whereof 6 typical and 6 atypical arguments. Typical arguments for sport shoes are for example that they are comfortable, that they have buoyant soles, Atypical arguments are for example that they are sold in a sportive handbag, that they are delivered with a nice key-ring, In total, there were 4 different versions of the ads. These versions only differed in the fact that argument presentation was rotated to exclude order of presentation effects. The ads were developed in collaboration with a professional copywriter. Subjects were asked to process the ad from a #brand evaluation’ perspective, where effects on attitudes should occur primarily through argument elaboration (Hastak and Olson 1989, p. 453). Moreover, a fictive, for the subjects unknown brand, was used such that it can be expected that no prior existing attitudes existed towards the brand.

After the ads were recollected by the experimenter, and after an unrelated filler task to exclude primacy and recency effects (Guttentag and Carroll 1997), the dependent variables were measured.

TABLE 1

MEAN NUMBER OF ELABORATED ARGUMENTS

Measures

Self-relevance. Self-relevance was measured by asking the respondents on a 7-point rating scale to what degree a number of personality traits in the survey applied to themselves. Of course, only the trait #flexible’ was relevant for our study, but it would not be appropriate for our design to confront subjects directly with the trait of interest (because they had seen it only moments ago in the ad). Instead people were told that our study was accompanied by an other study #investigating personality traits of undergraduate students’. In total, 49 personality traits were presented to subjects. This survey was part of the filler task.

Message elaboration. To investigate message elaboration, subjects were asked to write down everything they remembered about the previously presented ad. From these essays, the total number of typical and atypical arguments was calculated. This resulted in two measures: (a) total elaboration (atypical + typical arguments) and (b) ratio of atypical arguments (atypical / (typical + atypical arguments). These two measures were previously used by other researchers to study elaboration processes in the persuasion domain (e.g. Batra and Stayman 1990; Petty, Cacioppo and Schumann 1983).

Brand attitudes. Brand attitudes were measured using 8 seven-point semantic differential items: love/hateful, delighted/sad, acceptance/disgusted, joy/sorrow, useless/useful, safe/unsafe, perfect/imperfect, and wholesome/unhealthy (Crites, Fabrigar and Petty 1994). Scores on these items were recoded into values ranging from B3 (negative) over 0 (neutral) to +3 (positive) in order to be able to represent the valence of the attitudes. These items were then averaged for analysis (alpha=.80).

Confidence in attitudes. Confidence in attitudes as a predictor of behavior (Berger 1992) was measured by asking subjects how certain they were about their evaluation of the brand. This 7-point scale ranged from very certain to very uncertain.

RESULTS AND DISCUSSION

Elaboration

Total amount of elaboration. As expected, no effect was found of self-relevance on total message elaboration (t(61)=.78; p>.10). When the self was primed, no significant differences were found in comparison with when the self was not primed (M=553 versus M=5.66).

Ratio of atypical arguments. Although no effect was found of self-relevance on the total amount of elaboration, a significant effect was found on reproduction of the number of atypical arguments as compared to the total amount of recalled arguments (t(61)=2.09; p<.05). When the self was primed, subjects recalled more atypical arguments in comparison with when the self was not primed: the ratio was higher in the self-relevant condition (M=.61) compared to the not self-relevant condition (M=.51). This is consistent with our predictions: when the self is primed people will rely more on their general knowledge structures compared to when the self is not primed. This however is only the case when the information presented is typical, i.e. when this information is perceived by subjects as part of already stored knowledge. As a result of this bias towards the processing of typical information, people whose self-schema is primed will elaborate more on atypical information.

Table 1 shows the mean number of recalled arguments for both total elaboration and ratio of atypical information together with the mean results of recall of typical and atypical information.

The elaboration results show that automatic activation of the self has a substantial effect in an advertising context. Priming the self however does not lead to mindless processing of information: no effects were found of self-relevance on the total amount of elaboration. However, priming the self does lead to a processing bias: more atypical information is processed as compared to typical information. This suggests that through automatic activation of the self-schema, information associated with this schema (typical information) is processed in a more automatic fashion in comparison with information not associated with the self-schema. This leaves the subject with more processing capacities to elaborate on the atypical arguments.

Attitudes

Consistent with our expectations, a significant effect was found of self-relevance on brand attitudes (t(60)=2.16; p<.05). Brand attitudes were more extreme, and more positive when the self was primed (M=.64) compared to when it was not (M=-.005).

It is important to notice that in our previous investigation of typical and atypical arguments, subjects rated the atypical arguments as less convincing as the typical arguments. Indeed arguments such as #gives you a feeling of concentration’ and #are cheap’ can not be considered as very strong arguments to buy the advertised sport shoes. These atypical arguments can thus be said to be rather weak arguments, that can even be considered as counter arguments when involvement with the product category is high (e.g. Petty, Cacioppo and Schumann 1983; Petty and Cacioppo 1986). While subjects in the self-relevant condition did elaborate more on these atypical arguments, their attitudes still were more favorable and extreme than those of subjects in the not self-relevant condition. This suggests that when the self is primed, attitudes are more influenced by peripheral cues, such as mental constructs, than by message elaboration. These results are in accordance with studies that found that attitudes of subjects in positive moods are less influenced by quality of argument than were attitudes of individuals in neutral or sad moods, although when explicitly asked to evaluate message quality, happy participants differentiate between strong and weak arguments as well as neutral or sad mood participants (e.g. Bless, Bohner, Schwarz, and Strack 1990; Mackie and Worth 1989). These results suggest that happy participants noticed the quality of the arguments presented to them but did not use this information when making an attitude judgment. Similar for the present study, we can argue that subjects for whom the self was primed did elaborate more on atypical arguments, but did not use this information when making an attitude judgment. Instead, the self-schema can be seen as a peripheral information cue. Notice that all subjects in our study werehighly involved with the product category of sport shoes, so they all have mental constructs of sport shoes that are associated with the self-schema (because of the involvement). It can thus be expected that when the self is primed, activation is spread through a network of spreading inter-correlated associations (Klein and Loftus 1996), and mental representations of #me and sport shoes’ are also made more accessible. This will result in the fact that attitudes will be based more on these existing knowledge structures. Social psychologists have indeed found automatic evaluation effects when the relevant attitude-objects were made accessible (Bargh, Chaiken and Pratto 1992; Bargh, Chen and Burrows 1996).

Confidence in Attitudes

In line with our hypothesis, the effect of self-relevance on attitude confidence was significant (t(61)=2.09; p<.05). Attitudes were held with less confidence when the self was primed (M=2.61) compared to when it was not (M=3.25). This suggests that, as expected, priming the self results in a more heuristic processing mode.

CONCLUSION

In sum, these results of automatic activation of the self-concept are highly relevant for the persuasion domain. Several theoretical contributions derive from these findings. First, it was found that priming the self results in a biased elaboration of atypical arguments: although no effect was found on total message elaboration, subjects did elaborate more on atypical arguments as compared to typical arguments. This indicates that when people are confronted with information relevant to the self (i.e. a self descriptive personality trait), they will rely more on their self-schemas if the arguments are perceived as #already stored knowledge’ (i.e. typical arguments). However, when the arguments are perceived as atypical, people whose self is primed will attend more to these atypical arguments.

Second, because it was argued that these atypical arguments were less convincing (see earlier), it can be concluded that people who did elaborate more on these atypical arguments (because of priming effects), did not use these elaborations in making their product evaluations. They instead relied on heuristics (their self-schema), resulting in attitudes that were more extreme and favorable. This indicates that, when the self is primed, activation spreads to associated constructs, including the attitude construct of the product category (that is indeed present for consumers who are highly involved with the product category, as in the present study). This attitude construct is hereby made more accessible, such that attitudes are more extreme (Bargh, Chen and Burrows 1996).

Finally, in line with the expectation that priming the self will lead to more heuristic processing, it was found that although attitudes were more favorable when the self was primed, they were held with less confidence. We can thus conclude that automatic activation of the self in a persuasion context results in relatively more heuristic processing (Berger, 1992).

A limitation of the present research lies in the fact that we did not control for the effects of strong versus weak arguments. As stated earlier, prior investigation revealed that most atypical arguments were seen as weak or counter arguments. In some cases however, an atypical argument can be extremely relevant for the product judgment such that it can be considered as a rather strong argument. Given the fact that weak and strong arguments can be processed in a different manner, it can be highly relevant for future research to control for the plausible orthogonal effects of typical versus atypical arguments on the one hand and strong versus weak arguments on the other hand.

It is important to notice that our study dealt with typical and atypical informtion to be processed. However, more important is the question whether the same results will be found using new and old information. This can be an important issue for marketers who want to communicate the benefits of new and/or innovative products. The present theoretical framework suggests that priming the self will lead to more elaboration of new material. This issue however awaits further investigation.

Another issue for further investigation is the generalization of the present results. The question arises whether the same results will be found using real consumers as a sample instead of undergraduate university students. Also, we assumed self-relevance to be the only difference between the two experimental conditions. We operationalized self-relevance as the degree to which people see themselves as flexible or not flexible. However, the possibility remains that people who categorize themselves as flexible have different processing styles compared to people who do not see themselves as flexible, resulting in response differences that are not due to effects of self-relevance. In order to account for this possibility it can be interesting for further research to investigate whether or not the same effects will be found when using personality traits that are unrelated to flexibility.

REFERENCES

Bargh, John A. (1982), "Attention and Automaticity in the Processing of Self-Relevant Information," Journal of Personality and Social Psychology, 43 (3), 425-436.

Bargh, John A., Shelly Chaiken, Rajen Govender and Felicia Pratto (1992), "The Generality of the Automatic Evaluation Effect: Unconditionally Automatic Attitude Activation with a Pronunciation Task," Journal of Experimental Social Psychology, 62(6), 104-128.

Bargh, John A., Mark Chen and Lara Burrows (1996), "Automaticity of Social Behavior: Direct Effects of Trait Construct and Stereotype Activation on Action," Journal of Personality and Social Psychology, 71(2), 230-244.

Bargh, John A. and Paula Pietromonaco (1982), "Automatic Information Processing and Social Perception: The Influence of Trait Information Presented Outside of Conscious Awareness on Impression Formation," Journal of Personality and Social Psychology, 43(3), 437-449.

Batra, Rajeev and Micheal L. Ray (1986), "Affective Responses Mediating Acceptance of Advertising," Journal of Consumer Research, 1(2), 103-124.

Batra, Rajeev and Douglas M. Stayman (1986), "The Role of Mood in Advertising Effectiveness," Journal of Consumer Research, 17(2), 203-214.

Berger, Ida E. (1992), "The Nature of Attitude Accessibility and Attitude Confidence: A Triangulated Experiment," Journal of Consumer Psychology, 1(2), 103-124.

Bless, Herbert, Gerd Bohner, Norbert Schwarz and Fritz Strack (1990), "Mood and Persuasion: A Cognitive Response Analysis," Personality and Social Psychology Bulletin, 16(2), 331-345.

Bless, Herbert, Diane M. Mackie and Norbert Schwarz (1992), "Mood Effects on Attitude Judgments: Independent Effects of Mood Before and After Message Elaboration," Journal of Personality and Social Psychology, 63(4), 585-595.

Crites, Stephen L. Jr., Leandre R. Fabrigar and Richard E. Petty (1994), "Measuring the Affective and Cognitive Properties of Attitudes: Conceptual and Methodological Issues," Psychological and Social Psychology Bulletin, 20(6), 619-634.

Guttentag, Robert E. and Donna Carroll (1997), "Recency Judgments as a function of Word-Frequency: A Framing Effect and Frequency Misattributions," Psychonomic Bulletin nd Review, 4(3), 411-415.

Hastak, Manoj and Jerry C. Olson (1989), "Assessing the Role of Brand-Related Cognitive Responses as Mediators of Communication Effects on Cognitive Structure," Journal of Consumer Research, 15(4), 444-456.

Higgins, E. Tory and Gillian A. King (1981), "Accessibility of Social Constructs: Information-Processing Consequences of Individual and Contextual Variability," in Personality, Cognition, and Social Interaction, eds. Nancy Cantor and John F. Kihlstrom, Hillsdale, NJ: Lawrence Erlbaum Associates.

Higgins, E. Tory, Gillian A. King and Gregory H. Mavin (1982), "Individual Construct Accessibility and Subjective Impression on Recall," Journal of Personality and Social Psychology, 43(3), 35-47.

Judd, Charles M. and Cynthia M. Lusk (1984), "Knowledge Structures and Evaluative Judgments: Effects of Structural Variables on Judgment Extremity," Journal of Personality and Social Psychology, 46(6), 1193-1207.

Kardes, Frank R. (1998), Consumer Behavior and Managerial Decision Making, Addison-Wesley.

Klein, Stanley B. and Judith Loftus (1996), "The Role of Episodic Memory in the Development of Trait Knowledge", Social Cognition, 14(4), 277-291.

Kuiper, Nicholas A. and Paul A. Derry (1981), "The Self as a Cognitive Prototype: An Application to Person Perception and Depression," in Personality, Cognition, and Social Interaction, eds. Nancy Cantor and John F. Kihlstrom, Hillsdale, NJ: Lawrence Erlbaum Associates.

Linville, Patricia W. (1982), "The Complexity-Extremity Effect and Age Based Stereotyping," Journal of Personality and Social Psychology, 42(2), 193-211.

Logan, Gordon D. (1979), "On the Use of a Concurrent Memory Load to Measure Attention and Automaticity," Journal of Experimental Psychology: Human Perception and Performance, 5(2), 189-207.

Mackie, Diane M. and Leila T. Worth (1989), "Processing deficits and the Mediation of Positive Affect in Persuasion," Journal of Personality and Social Psychology, 57(1), 27-40.

Markus, H. (1980), "The Self in Thought and Memory," in The Self in Social Psychology, eds. D.M.Wegner and R.R. Vallacher, New York: Oxford University Press.

MacInnis, Deborah J. and Bernard J. Jaworski (1989), "Information Processing from Advertisements: Toward an Integrative Network," Journal of Marketing, 53(4), 1-23.

Mittal, Banwari (1995), "A Comparative Analysis of Four Scales of Consumer Involvement," Psychology and Marketing, 12(7), 663-682.

Petty, Richard E. and John T. Cacioppo (1986), Communication and Persuasion, New York: Springer-Verlag.

Petty, Richard E., John T. Cacioppo and David W. Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research, 10(2), 135-146.

Shiffrin, Richard M. and Walter Schneider (1977), "Controlled and Automatic Human Information Processing: II. Perceptual Learning, Automatic Attending, and a General Theory," Psychological Review, 84(2), 127-190.

Symons, Cynthia S. and Blair T. Johnson (1997), "The Self-Reference Effect in Memory: A Meta-Analysis," Psychological Bulletin, 121(3), 371-394.

Wyer, Robert S. and Thomas K. Srull (1989), Memory and Cognition in its Social Context, Hillsdale, NJ: Erlbaum.

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