Self-Perception Development and Consumer Choice Criteria: Is There a Linkage

Chris T. Allen, University of Massachusetts/Amherst
William R. Dillon, University of Massachusetts/Amherst
ABSTRACT - Self-perception theory has been presented as highly germane to consumer researchers because it specifically can accommodate the dynamic nature of consumption. Yet there is little compelling evidence that the cognitive activity prescribed by the theory does, in fact, mediate the influence of one behavior on another. This paper examines the link between self-perception development and consumers' choice criteria, and supports a causal inference regarding its existence.
[ to cite ]:
Chris T. Allen and William R. Dillon (1983) ,"Self-Perception Development and Consumer Choice Criteria: Is There a Linkage", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 45-50.

Advances in Consumer Research Volume 10, 1983      Pages 45-50

SELF-PERCEPTION DEVELOPMENT AND CONSUMER CHOICE CRITERIA: IS THERE A LINKAGE

Chris T. Allen, University of Massachusetts/Amherst

William R. Dillon, University of Massachusetts/Amherst

[Data collection in this project was made possible by a grant from the Arthur D. Hill Research Support Fund, Department of Marketing, The Ohio State University, and by In Energy Research Initiation Grant from the Graduate School or The Ohio State University.]

ABSTRACT -

Self-perception theory has been presented as highly germane to consumer researchers because it specifically can accommodate the dynamic nature of consumption. Yet there is little compelling evidence that the cognitive activity prescribed by the theory does, in fact, mediate the influence of one behavior on another. This paper examines the link between self-perception development and consumers' choice criteria, and supports a causal inference regarding its existence.

INTRODUCTION

Self-perception theory has been represented as providing at least an appropriate starting point for developing a cognitive theory of how individuals learn from their experiences (Scott 1981). Such a theory has considerable potential relevance for consumer researchers if one conceives of consumption as a dynamic process wherein current actions are affected by prior behaviors (Scott 1978; 1981). Most would grant consumption's dynamic character and the resultant importance of crying to explicate the cognitive linkages between behaviors performed over time for explaining consumption. The self-perception paradigm has been the focus of a number of consumer research studies; limited insights about how performed behaviors influence beliefs and subsequent behavior have been offered, along with some tentative guideLines for generating marketing strategy (Scott 1981).

Bem's (1972) self-perception paradigm does presuppose some level of cognitive activity which mediates the influence of one behavior on another, but is quite imprecise in specifying its nature. Generally, the self-perception process is viewed as beginning with performance of a behavior which may then stimulate causal analyses. If the performer does not perceive the behavior to be caused by a situational/circumstantial pressure or cue, then the process is likely to lead to an altered internal stare (e.g., attitudes, self-perceptions, motives, and/or emotions) which in turn presumably drives subsequent behavior.

Despite their popularity, however, the self-perception paradigm, and attribution theories generally, have a skeleton in the closet. Researchers have been able to demonstrate behavior changes predicted by these theories more reliably and persuasively than the attribution/ self-perception changes (i.e., self-report measures on internal states) that are, theoretically. supposed to be mediating them (cf., Bem 1972; Scott 1975; Tybout and Yalch 1980; Wilson, Hull, and Johnson 1981). One explanation for these largely equivocal findings involves the recognition that the proposed cognitive changes may not, in fact, be mediating the behavioral outcomes. Accepting this explanation necessarily implies an abandonment or self-perception and attribution theories (Wilson, Hull, and Johnson 1981), and would also imply that consumer researchers should be investigating alternative perspectives like behavior modification, which deliberately avoid speculations about the mediating role of cognitions (Nord and Peter 1980).

Indeed, in the consumer behavior literature, there is woeful lack of evidence about the link between self-perception development and decision making. Behavioral effects predicted by Bem's paradigm have been demonstrated (e.g., Dodson, Tybout, and Sternthal 1978; Scott 1976), and self-perception development has been investigated in "untraditional consumer contexts" like participation in publicizing a-recycling program (Scott 1977), and voting participation (Tybout and Yalch 1980), but there is no evidence that directly links self-perception development with consumer decision making. This is a matter of concern not only because of the general skeleton in the attribution theorist's closet, but also because of the growing number of researchers who recently have expressed doubts about the usefulness of self-perception-based influence strategies in pure applied contexts (cf., Allen, Schewe, and Wijk 1980; Furse, Stewart, and Rados 1981; Tybout, Sternthal, and Calder 1982; Tybout and Yalch 1980).

The goal of this paper is to establish whether internal states potentially altered by self-perception processes are causally linked to individuals' consumption priorities. The specific focus of the investigation is the link between self-perception development and individuals' importance ratings on specific product performance attributes.--Attribute-importance ratings, of course, underlie consumption decisions not only because of their potential direct influence on brand choice (e.g., Myers and Alpert 1968), but also because they may reflect what information a consumer would seek when evaluating brands (e.g., Holbrook and Maier 1978).

SPECIFYING THE NATURE OF SELF-PERCEPTION DEVELOPMENT

Bem's (1979) lack of specification regarding the exact nature of the self-perception changes which occur after performance of a behavior has certainly created problems for those crying to demonstrate the mediating role of these cognitions on subsequent behaviors (Dejong 1979). However, as researchers have attempted to explain their failure to demonstrate such effects, they have generated a number of cogent arguments which are beginning to add precision to Bem's paradigm (cf., Bem 1979; Scott and Yalch 1978; Tybout and Yalch 1980; Wilson, Hull, and Johnson 1981). While an exhaustive treatment of these arguments is beyond this paper's scope, one issue must be considered because of its methodological implications: it concerns the specific type of cognitive construct altered by self-perception Processes.

Somewhat ambiguously, Bem (1979) refers to cognitions altered by self-perception processes as attitudes and/ or self-attributions. It seems warranted to distinguish this process output from a general attitude conception because it involves heavy emphasis on self-awareness and self-evaluation (Dejong 1979). The most developed conceptualization concerning the nature or cognitions impacted by processes like self-perception appears to be Markus' (1977) self-schemata notion. She defines self-schemata as "cognitive generalizations about the self, derived from past experience, that organize and guide the processing of self-related information contained in the individual's social experience" (Markus 1977, p. 64). Since she is specifically considering the nature of cognitions formed as a result of "attempts to organize, summarize, or explain one's own behavior in a particular domain" (Markus 1977, p. 64), her conception does seem directly interpretable as self-perception output. Indeed, Markus (1977) offers her conception as one to fill the void of cognitive structure specification in theories like self-perception (see p. 64).

Markus' self-schemata or "cognitive generalizations about the self" come very close to a self-concept type of conception, and her work seems to indicate that in attempting to trace self-perception process output, one should look to self-concept variables. Interestingly, in empirical studies where both attitudinal and self-concept measures were employed in gauging self-perception development, experimental manipulations designed to alter self-perceptions generally registered stronger effects on the self-concept measures (Chaiken and Baldwin 1981: Scott 1977; Wilson, Hull, and Johnson 1981). It should be emphasized that this prior work was not developed specifically to examine the issue of whether attitude or self-concept is most appropriately conceived as self-perception process output; yet it does suggest that in trying to monitor such output, it will be more productive to have individuals make judgments about themselves and their prior behaviors, rather than asking for their feelings toward an attitude object. In this project, both attitudinal and self-concept measures were used to gauge self-perception development, but it was anticipated that the experimental manipulations would register more strongly on the self-concept variable.

STIMULATING SELF-PERCEPTION PROCESSES VIA LABELING

As mentioned previously, consumer researchers have begun to express doubts about the usefulness of some self-perception-based influence strategies (principally, foot-in-the-door), but the problem again can be traced to a lack of specification in Bem's paradigm. There is no guarantee, and indeed, it is highly unlikely, that individuals are always able or motivated to engage in causal analysis (Scott 1978; 1981; Wilson, Hull, and Johnson 1981); Bem (1979) however, has little to say concerning the specific conditions that will foster spontaneous causal analysis. Without such knowledge influence techniques like foot-in-the-door, which rely on spontaneous causal analysis, might be expected to yield inconsistent effects across research contexts. This has been the case in applied settings where behaviors like responding Lo a telephone interview (Allen, Schewe, and Wijk 1980), or trying a product (Scott 1976), may no: spontaneously stimulate causal analysis.

In exploring the mediational role of self-perception development experimental manipulations that actively drive the self-perception process may be required. As demonstrated by Tybout and Yalch (1980), labeling appears t. be an appropriate device for this purpose and it is employed in the present project. Bem's paradigm recognizes that a trait-label provided by another person may serve as an important source of information about one's own dispositions, or may at least signal that a behavior and its implications for self-image should be reflected on (Dejong 1979). Stimulating such self-reflection, of course, amounts to activating the causal analysis process critical in self-perception development.

METHOD

Context and Overview of the Experiment

A laboratory experiment was devised to study the linkage between self-perception development and individuals' consumption priorities. Labeling was used in the experimental manipulations and to enhance external validity, labels were delivered via specially-developed television advertisements. Energy conservation was the research context and the ads were developed to influence individuals' perceptions of effectiveness as consumers. Perceived consumer effectiveness is a disposition that has been linked to energy efficient consumption (e.g., Good 1979; Seligman et al. 1979); such perceptions have also been conceived as "cognitive output" that evolve from self-perception processes (Allen 1982; Scott 1977).

In the first phase of the experiment subjects were exposed to one of four levels of the promotional treatment and then furnished self-report measures of their self-perceptions concerning consumer effectiveness. To create a control group to detect demand artifacts, roughly one in every five subjects was processed without these measures. In the second phase participants reported, in the context of a product evaluation exercise, how much importance they would place on energy utilization if they were purchasing a major appliance.

Subjects

Participants were 358 women recruited from various organizations in Columbus, Ohio; organizations were compensated for their members' participation. To increase the likelihood that subjects had purchasing experience with the focal experimental product (refrigerators), only home-owners were recruited, and to generate diversity, subjects were recruited from two sections of the city that differed markedly in social status. The sample included strong representation from the upper and lower middle classes, and from the upper lower class. Additional description of the sample is furnished in Allen (1982).

Procedure

Subjects were cold they would be participating in two studies: this cover was used to disassociate the conservation advertisement delivered in a "political opinions study" from the product evaluation task for refrigerators performed in the "consumer decision-making study." Fifty-four groups of five to nine persons were processed--each group was assigned randomly to a level of promotion.

The first study or phase began with participants responding to several political opinion items in a warmup questionnaire. They next watched a program that they were told had been videotaped from a local TV network; it was a documentary on property tax issues chosen to be consistent with the experimental guise and with the home-owner subject screening requirement. During the twelve minutes of video subjects saw the treatment and two dummy ads and afterwards responded to 13 items concerning property tax issues and 19 about energy (including the self-perception measures) on a six-point agree-disagree scale. They were then told they had finished the first study and could relax while their questionnaires were being collected.

As they were relaxing, an information display board (Jacoby et al. 1976) was placed in front of each participant along with a questionnaire and sec of instructions. Study two was introduced as an investigation of how consumers use information about products when evaluating them, and the information display boards were described as devices that would facilitate this purpose. Subjects ultimately evaluated two product groups--spray deodorants and refrigerator/freezers--using the boards. As a preliminary, they gave attribute importance ratings for these two products; the study's key dependent measure was generated here.

Experimental Manipulations

The four promotion levels consisted of two appeals emphasizing society's needs, a functional appeal stressing personal financial benefit, and a no appeal condition. The two societal messages were developed to influence self-perception development on the perceived consumer effectiveness dimension. The persuasive societal appeal argued that individual action can help resolve the energy problem. The attribution societal appeal labeled the American consumer a willing participant in solving the energy problem; examples of conservation behaviors individuals say they are performing (Milstein 1977; O'Neill 1975) were referenced to enhance the label's credibility. This persuasive/attribution distinction was patterned after the work of Miller, Brickman, and Bolen (1975). Treatment ads were filmed in color using a professional actor to deliver the messages.

Measurement

Self-Perceptions. Self-perceptions on the perceived consumer effectiveness dimension were operationalized to reflect both an attitudinal and a self-concept conception. The attitudinal component, referred to as general feeling of effectiveness (GFE) throughout the discussion, is made up of five items (e.g., "energy is really not my problem because there is simply nothing I can do about it"). The self-concept component, referred to as perceived change in consumption (PCC),-consists of two items ("I've changed the way I purchase and use products as a result of the nation's energy problem," and "When I buy products, I seldom consider how my use of them will contribute to reducing the severity of the nation's energy problem").

GFE is treated as an attitudinal variable since the items tap sentiments about an attitude object (in this case, the issue of individual responsibility for the energy problem). PCC taps self-awareness and self-evaluation--this emphasis on self reflects a self-concept variable. Factor analysis results from this data base (Allen 1982) and others collected in the U.S., Sweden, and Canada (Allen, Calantone, and Schewe 1982) support a conclusion that these two sets of items represent separate cognitive dimensions. Notably, literature dealing with the general phenomenon of socially-conscious consumption also portrays perceived consumer effectiveness as involving both attitudinal and self-content components (Henion 1976).

Attribute Importance. Importance ratings were taken for ten refrigerator/freezer attributes on a six point scale (extremely important to not at all important). A normalized score for the energy usage attribute was employed in analyses.

RESULTS

Tests for Demand Artifacts

As typifies laboratory research, there was a concern that demand artifacts presented a potential confound to valid inference. The focus of concern involved the energy questionnaire items and how the act of filling them out might affect a subject's subsequent responses on the importance ratings. Tv help minimize the problem the two key rating tasks were disassociated through the two-study experimental guise. Moreover, processing roughly a fifth of the subjects as controls made a direct test for the confound problem possible. In an analysis of variance that included presence/absence of the energy items and promotion as separate factors, no main effect for presence of energy items, and no promotion by energy item interaction effect were found on the attribute importance dependent measure. Thus, perhaps due to the two-study experimental cover, the feared artifact problem did not materialize.

Promotion's Effect on Attribute Importance

To legitimately examine the mediational role of self-perception development on importance ratings, it must be established that the treatment did, in fact, influence subjects' ratings. It was expected that the attribution appeal would have more impact on these ratings than the societal persuasive appeal, following the example of Miller, Brickman, and Bolen (1975). It was also anticipated that the functional appeal would affect these ratings, but there was no a priori basis for making a prediction about the relative impact of this appeal.

Promotion's impact was assessed in a design that included subjects' social class status as a blocking variable to increase the efficiency of the main effect's test. Analyses utilized just participants who furnished absolutely complete response sets: deletion of subjects was not related to experimental treatments (X2(6) = 9.6;, p > .80). Promotion did affect the importance ratings (F(3,189) = 2.62, p < .052). Newman-Keuls tests demonstrated that persons exposed to the attribution a peal raced the energy use attribute as more important than chose in the no appeal group (p < .01), but the persuasive and functional appeals had no apparent impact on the ratings.

The Mediating Role of Self-Perceptions

The focal research issue then entails establishing whether self-perception development mediated the impact of the promotional manipulation on participants' importance ratings. This issue was examined using two complementary analyses. First, the data were examined with an analysis or covariance procedure utilized extensively in documenting the mediating role or cognitive responses on attitude development (Wright 1980). Second, more formal causal modeling was used to confirm the results of the covariance analysis. Subjects exposed to the functional appeal were dropped from these analyses since this message was not designed to influence self-perceptions.

The covariance procedure entails a specific analytical sequence (see Wright 1980, p. 163) which can suggest a modest causal inference about the mediational role of one variable on another. The prescribed sequence was executed for PCC and GFE with summated scores as data input for these variables. Summaries of these analyses are in Table 1. (The PCC & GFE interaction was also examined in this stage of the analysis, but the results revealed no significant findings.)

Table 1 shows the mediational role of self-perceptions is supported, but the pattern of prescribed effects is only in place for PCC. This result was not unanticipated, of course, because PCC represents the self-concept-oriented conception. Parallel effects were demonstrated; that is, just as the attribution treatment enhanced attribute importance ratings relative to the no appeal treatment, it also enhanced PCC when compared to the no appeal group (p < .05). As desired, using PCC as a covariace eliminated the treatment effect on the importance ratings, while with attribute importance as a covariate, the treatment effect on PCC was maintained. As shown in Table 1, this desired sequence of effects was not in place for GFE.

The findings indicate that the influence of the attribution message on importance ratings operated through PCC. In contrast, the covariance analysis did not provide compelling evidence that the same was true for GFE and thus GFE was eliminated from subsequent analysis.

TABLE 1

ANALYSIS OF COVARIANCE SUMMARIES

The Mediating Role of PCC: A Causal Analysis

A more direct analysis of PCC's mediating role was conducted using the model illustrated in Figure 1. The model consists of: a single manipulated independent variable x, which relates to the experimental treatment groups defined here to have only two levels (the attribution and no appeal groups); two manipulation checks y1 and y2 which refer to the two measurement items for the PCC variable; and y3, which is the single operationalization of the attribute importance variable. Figure 1 represents a causal model explicitly incorporating the notion of manipulation checks, measurement error, and the relationship between independent and dependent variables in a single overall structure.

Letting n1 and n2 represent the true but unobservable PCC and Choice Criteria constructs, respectively, the causal system depicted in the Figure can be written in terms of the following structural form equations:

EQUATION (1)

or

EQUATION (2)

The reduced form is

EQUATION (3)

where x = x. Standardized maximum likelihood parameter estimates have been inserted in the Figure; all estimates were obtained using LISREL IV (Joreskog and Sorbom 1978).

The X2 goodness-of-fit for the model with 2 d.f. is 0.32, an acceptable fit (p = .32), and all parameter estimates are significant (p < .05). Thus, the experimental manipulation (p = 0.24) had a significant positive effect on PCC. The parameter estimate for b, which gives the direct effect of the true independent variable on the true dependent variable, indicates that a one unit change in the true independent variable (PCC) will produce a change of b = 0.29 units in the true dependent variable (Choice Criteria). Notice, finally, that l2 is significant which supports a conclusion that the PCC construct is adequately reflected by the observed scores on y1 and y2.

An alternative model was fit that included a direct link between the experimental treatment, x, and the true dependent variable (Choice Criteria), h2. The X2 goodness-of-fit for this model is 0.45 and with 1 d.f., the probability level evaluates to p = 0.50. Though this is also an acceptable fit, the parameter estimate for the link between treatment and Choice,Criteria was nonsignificant; also, the difference in X2 values for the two models is 1.80 with 1 d.f., which is nonsignificant. This further upholds the initial model which proposed that the only effect of the experimental manipulation on the importance ratings comes about through the mediating PCC variable.

DISCUSSION AND CONCLUSIONS

Given the tightly-controlled experimental procedure which seemed to foster no major demand artifact problems, and the consistent findings from the two forms of analysis, a legitimate causal inference seems justified: self-perception development affected subjects' consumption priorities. The study, of course, does not furnish-direct inferences about self-perception processes; such cognitive processes are difficult to examine, and at minimum some form of thought verbalization data (Wright 1980) would be required for making more direct inferences about them. However, the PCC variable gauges an internal state that seems appropriately conceived as cognitive output from the self-perception process. This cognitive state variable was clearly linked to choice criteria, and changes in PCC apparently altered attribute importance ratings. Thus, the research is supportive of Bem's self-perception theory as a basic framework for developing a cognitive theory of how individuals learn from their behavior in consumption settings.

The study offers two implications for researchers interested in developing the self-perception paradigm both in terms of its theoretical precision, and its pragmatic value. First, as Dejong (1979) has observed, perhaps the key to assessment of self-perception development is identifying the issue or task specific trait dimension on which the output of the self-perception process is likely to register. In this study identification of the appropriate dimension was facilitated by literature in the energy area dealing with perceived consumer effectiveness. It may be that some of the trait/self-schema dimensions (e.g., health-conscious) individuals use in thinking about themselves develop as they observe their behaviors in the marketplace. This study indicates that as these self-perceptions (e.g., health-conscious) form, general choice priorities also develop, with potential implications for brand choice across a variety of product categories (e.g., high-bran cereals, caffeine free beverages, low salt/sugar, high nutrient snacks). The challenge then to the commercial marketer is to position his/her brand so that the individual would perceive consumption of the brand as a behavior consistent with a developed self-schema trait.

FIGURE 1

BASIC MODEL AND MAXIMUM-LIKELIHOOD PARAMETER ESTIMATES (STANDARDIZED)

It would be premature on the basis of one limited study to accept self-concept variables (like PCC) and reject attitudinal ones (like GFE) as the cognitive construct that acts as the output variable for self-perception processes. However, the finding that PCC and not GFE mediated the impact of the attribution message on participants' importance ratings does seem consistent with work (e.g., Markus 1977; Rogers, Kuiper, and Kirker 1977) that points to a central role for the self in processing information. There is an appealing "dynamic" to Markus' (19,7) self-schema notion which suggests that these trait-like variables simplify information processing in the highly-complex social environment by predisposing individuals to think first about their behavior in terms of its implications for the self. The attitudinal implications of one's behavior may entail another "cognitive step," and this may be a step individuals are less likely to take. Salancik and Conway (1975) have observed that one implication or self perception theory is that attitudinal judgments are not naturally-evolving predispositions, but rather judgments developed to respond to the question: What is your attitude? Although speculative, the notion that individuals are "self-biased" in their information processing explains the finding that an experimental treatment which purportedly stimulated self-perception processes registered effects on a self-oriented measure (PCC), but not on an attitudinal one (GEE).

A second aspect of these findings also seems quite noteworthy: a single exposure to a television advertisement which delivered an attribution/labeling message was apparently enough to activate the self-perception processes of some participants. The self-perception paradigm's principal applied contributions are likely to come in the area or developing behavioral influence strategies (Scott 1981). In generating and testing such strategies, consumer researchers may be well-advised to give more attention to techniques that actively drive or stimulate the self-perception process (like labeling), and less to techniques (like foot-in-the-door) which rely on spontaneous occurrence or the process.

REFERENCES

Allen, Chris T. (1982), "Self-Perception Based Strategies for Stimulating Energy Conservation," Journal of Consumer Research, 3, 391-390.

Allen, Chris T., Calantone, Roger J., and Schewe, Charles D. (1982), "Consumers' Attitudes About Energy Conservation in Sweden, Canada, and the United States, with Implications for Policymakers," Journal of Marketing and Public Policy, 1, 57-67.

Allen, Chris T., Schewe, Charles D., and Wijk, Gosta (1980), "More on Self-Perception Theory's Foot Technique in the PreCall/Mail Survey Setting, " Journal of Marketing Re- search, 17, 498-502.

Bem, Daryl, J. (1979), "Self-Perception Theory," in Advances in Experimental Social PsYchology, ed. L. Berkowitz. New York. NY: Academic Press. 1-62.

Chaiken, Shelly and Baldwin, Mark i'. (1981), "Affective-Cognitive Consistency and the Effect of Salient Behavioral Information on the Self-Perception of Attitudes," Journal of Personality and Social Psychology, 41, 1-19.

DeJong, William (1979), "An Examination of Self-Perception Mediation of the Foot-in-the-Door Effect," Journal of Personality and Social Psychology, 37, 2271-2239.

Dodson, Joe A., Tybout, Alice M., and Sternthal, Brian (1978), "Impact of Deals and Deal Retraction on Brand Switching," Journal of Marketing Research, 15, 79-81.

Furse, David H., Stewart, David W., and Rados, David L. (1981), "Effects of Foot-in-the-Door, Cash Incentives, and Followups on Survey Response," Journal of Marketing Research, 18, 473-478.

Good, Walter S. (1979), "Canadian Attitudes Toward Energy Conservation: Implications for Public Policymakers," in Public Policy Issues in Marketing: Vol 1, eds. C. Frey et al., Ann Arbor, MI: University of Michigan, 146-159.

Henion, Karl E. (1976), Ecological Marketing, Columbus, OH: Grid, Inc.

Holbrook, Morris B. and Maier, Karl A. (1978), "A Study of the interface Between Attitude Structure and Information Acquisition Using a Questionnaire-Based Information-Display Sheet," in Advances in Consumer Research: Vol. 5, ed. H. Hunt, Ann Arbor, MI: Association for Consumer Research, 93-98.

Jacoby, Jacob, Chestnut, Robert W., Weigl, Karl C., and Fisher, William (1976), "Prepurchase Information Acquisition: Description of a Process Methodology, Research Paradigm, and Pilot Investigation," in Advances in Consumer Research: Vol. 3, ed. B. Anderson, Cincinnati, OH: Association for Consumer Research, 306-314.

Joreskog, Karl C. and Sorbom, Dag (1978), LISREL IV: Analysis of Linear Structural Relationships by the Method of Maximum Likelihood, Chicago, IL: National Educational Resources, Inc.

Markus, Hazel (1977), "Self-Schemata and Processing Information About the Self," Journal of Personality and Social Psychology, 35, 63-78.

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

Milstein, Jeffrey S. (1977), "Attitudes, Knowledge and Behavior of American Consumers Regarding Energy Conservation With Some Implications for Governmental Action" in Advances in Consumer Research: Vol. 4, .d. W. Perreault, Atlanta, GA: Association for Consumer Research, 315-321.

Myers, James H. and Alpert, Mark I. (1968), "Determinant Buying Attitudes: Meaning and Measurement," Journal of Marketing, 39, 13-20.

Nord, Walter R. and Peter, J. Paul (1980), "A Behavior Modification Perspective on Marketing," Journal of Marketing, 44, 36-47.

O'Neill, Harry W. (1975), "The Effects of Energy Availability and Costs on Consumer Attitudes and Behavior," in Advances in Consumer Research: Vol. 8, ed. "1. Schlinger, Cincinnati, OH: Association for Consumer Research, 863-878.

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

Salancik, Gerald R. and Conway, Mary (1975), "Attitude Inferences from Salient and Relevant Cognitive Content About Behavior," Journal of Personality and Social Psychology, 39, 899-840.

Scott, Carol A. (1976), "Effects of Trial and Incentives on Repeat Purchase Behavior," Journal of Marketing Research, 13, 963-269.

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

Scott, Carol A. (1978), "Self-Perception Processes in Consumer Behavior: Interpreting One's Own Experiences," in Advances in Consumer Research: Vol. 5, ed. H. hunt, Ann Arbor, MI: Association for Consumer Research. 714-790.

Scott, Carol A. (1981), "Forming Beliefs from Experience: Evidence from Self-Perception Theory," in Perspectives in Consumer Behavior, eds. H. Kassarjian and T. Robertson, Glenview, IL: Scott, Foresman and Co., 296-306.

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

Seligman, Clive, Kriss, M., Darley, John M., Fazio, R.H., Becker, L.J., and Pryor, J B. (1979), "Predicting Summer Energy Consumption from Homeowners' Attitudes," Journal of Applied Social Psychology, 9, 70-90.

Tybout, Alice M., Sternthal, Brian, and Calder, Bobby J. (1982), "Information Availability as a Determinant or Multiple Request Effectiveness," unpublished manuscript, Northwestern University.

Tybout, Alice M., and Yalch, Richard F. (1980), "The Effect of Experience: A Matter of Salience?" Journal of Consumer Research, 6, 406-413.

Wilson, Timothy D., Hull, Jay G., and Johnson, Jim (1981), "Awareness and Self-Perception: Verbal Reports on Internal States," Journal of Experimental Social Psychology, 40, 53-71.

Wright, Peter (1980), "Message-Evoked Thoughts: Persuasion Research Using Thought Verbalization," Journal of Consumer Research, 7, 151-175.

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