Attitudes, Advertising, and Automobiles: a Functional Approach

Richard Ennis, University of Waterloo
Mark P. Zanna, University of Waterloo
ABSTRACT - Two studies explore a methodology for identifying the psychological functions of attitudes toward a multifunctional consumer product: the automobile. We hypothesize that the needs met by consumer products will better be reflected in the beliefs that underlie attitude rather than in overall attitudinal evaluations. The results are discussed in terms of a functional approach in attitudinal research and some implications for consumer research.
[ to cite ]:
Richard Ennis and Mark P. Zanna (1993) ,"Attitudes, Advertising, and Automobiles: a Functional Approach", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 662-666.

Advances in Consumer Research Volume 20, 1993      Pages 662-666


Richard Ennis, University of Waterloo

Mark P. Zanna, University of Waterloo


Two studies explore a methodology for identifying the psychological functions of attitudes toward a multifunctional consumer product: the automobile. We hypothesize that the needs met by consumer products will better be reflected in the beliefs that underlie attitude rather than in overall attitudinal evaluations. The results are discussed in terms of a functional approach in attitudinal research and some implications for consumer research.

The investigation of the psychological importance of attitudes has been collectively described as the functional approach. Although work within this approach was initiated some thirty years ago (Smith, Bruner and White 1956; Katz 1960), it disappeared from the mainstream of psychological research due to the lack of an adequate methodology. Fortunately, there has been a renewed interest in this perspective and researchers have begun examining the needs met by various attitude objects such as personal possessions (e.g., Abelson 1986; Belk 1988) and the role of attitude function in advertising and persuasion (e.g., DeBono 1989; Shavitt 1989; Shavitt and Fazio 1987).

Although various functional taxonomies have been proposed (e.g., Herek 1986; Katz 1960; Shavitt 1989; Smith, Bruner and White 1956), all agree on a broad distinction between instrumental and symbolic functions. Instrumental attitudes focus on the features and attributes of the target object and their utility in providing better functioning in the environment. Symbolic attitudes represent events, relationships, thoughts or feelings that are important or meaningful to an individual; or provide a means of self-expression. Belk (1988) described the instrumental function as "a capability for doing;" we would be unable to do the same things without the attitude object. He distinguishes the symbolic function as "a capability for being;" we would not be the same person without the target item.

Consumer products such as the automobile can, therefore, be differentiated by the needs that they might serve. Furthermore, the functional capacity of a particular model will be reflected in the attitudes and opinions that one holds. This has led researchers to hypothesize that functionally consistent advertisements will be more persuasive than inconsistent appeals. Recent empirical evidence has supported such a position. For example, Shavitt and her colleagues (e.g., Shavitt and Fazio 1987) have shown that subjects primed by instrumental products will subsequently rate information-based appeals more highly than image-based appeals, while symbolically-primed subjects find image-based ads more convincing.

This apparent propensity for functional consistency has several implications for consumer psychology. If one can identify the psychological functions of a product as reflected in consumer attitudes this would greatly enhance product development, market segmentation, and advertising strategies. An ability to match persuasive appeals with the needs of a market would improve brand name identification and expand targeted audiences. This is especially true for multifunctional products such as the automobile.

We believe, however, that the potential for functional measurement and manipulation is not at the level of attitudes but at the level of beliefs which underlie those attitudes. Overall evaluations provide little, if any, evidence of the needs that a product might meet. Attitudinal dispositions are a summative expression of the salient thoughts, feelings, and behaviours associated with the target item (Ajzen and Fishbein 1980; Zanna and Rempel 1988). It has been demonstrated that the beliefs associated with an object (i.e., the beliefs comprising an attitude) can be readily identified (Ajzen and Fishbein 1980; Cacioppo and Petty 1981) and coded according to psychological function (Herek 1986; Shavitt 1989). We believe that the constellation of functional beliefs associated with a product will indicate the needs that the product is currently satisfying and the needs that it can potentially satisfy. Furthermore, we believe that the functional capacities of a product can be influenced by advertising appeals designed to manipulate the belief structure underlying consumer attitudes toward that product.

To explore these hypotheses, we conducted two experiments. We began by selecting four models of automobiles which we felt, a priori, would serve different patterns of needs. We assumed that subcompact and family models would meet predominantly utilitarian purposes while, in contrast, sport and luxury models would demonstrate a greater multifunctional capacity by meeting both instrumental and symbolic needs. In Study One, we tested our a priori assumptions about the functional distinctions amongst the car models and we looked for evidence that these distinctions are reflected in the attitudinal beliefs associated with each model. In Study Two, we tested the impact of advertising slogans on the functional perceptions of an unknown, hypothetical car.


We began by eliciting the attitudinal beliefs associated with each of the four car models using the method suggested by Fishbein and Ajzen (1980) which has received general acceptance in the literature. In a between-subject design, two different belief elicitation instructions were used to explicitly pull for either instrumental or symbolic thoughts. We sensed that the traditional procedures are biased for utilitarian attributes, therefore might not capture the full richness of the symbolic needs served by an automobile. Using the two sets of instructions as an independent variable, we hypothesized that the symbolic instructions would generate not only more symbolic beliefs, but would also provide empirical evidence of the relative importance of these beliefs.

Having accomplished the elicitation of attitudinal beliefs, we then constructed a manual for coding each idea according to the psychological need that it addresses. We relied on Shavitt's (1989) coding procedures which she has successfully applied to consumer products such as air conditioners, coffee, perfume, and soft drinks. We hypothesized that automobiles, in general, are a multifunctional product. Therefore, we should find that each of the four models generates a substantial number of attitudinal beliefs associated with both instrumental and symbolic functioning. We also assumed that vehicle models differ in their ability to serve expressive needs. We have hypothesized that luxury and sport cars are more multifunctional (i.e., better able to serve symbolic needs) relative to subcompact and family models. By coding each belief according to the function it serves, we can test these assumptions empirically.


Subjects in this study were sixty (34 females and 26 males) introductory psychology students who participated as part of a course requirement. For half the subjects, the initial instructions explained the instrumental function served by automobiles which read, in part: "we continually assess how `useful' objects are; and how certain features or attributes of each item contribute to, or detract from, that usefulness. Attitudes toward everyday objects are based, in part, on our analysis of the good and bad characteristics of the object."



The remaining subjects received an explanation about the symbolic functions that might be served by automobiles which read, in part: "we can learn something about a person by the items and objects that he/she uses and owns. They provide us with a means of identifying what kind of person he or she is C what might be good or bad; likeable or dislikeable about someone who owns or uses a particular item. Attitudes toward everyday objects are based, in part, on our analysis of the good and bad characteristics expressed by the object."

Subjects then described their beliefs about each type of vehicle on separate thought-listing sheets (Cacioppo and Petty 1981) with the order of responding counterbalanced in a latin-square design. Next, subjects' attitudes toward each model were assessed on two 7-point bipolar scales with endpoints labelled "extremely good"/"extremely bad" and "extremely like"/"extremely dislike." Responses on these two evaluative scales were summed.

Finally, subjects completed a direct measure of the functions served by each car model. This form was adapted from a similar measure used successfully by Shavitt (1989). Subjects were asked to rate on a 5-point unipolar scale, with endpoints of "does not contribute at all to my attitude" to "contributes a great deal to my attitude," the contribution of instrumental and symbolic factors to their attitudes toward each model. For each type of vehicle, the instrumental function was described as "my past experiences with the object: how useful I found the item"; the symbolic function was described as "my social identity: what the item tells others about me" and "my values: what the item means to me."


The direct measure of function confirmed our a priori assumptions about the different models of cars. Recall that subcompact and family models were assumed to serve greater utilitarian and less symbolic needs than luxury and sports models. The results presented in Table 1 provide a pattern consistent with this functional ordering. The planned contrasts reveal that the utilitarian factor contributed more toward subcompact and family vehicles than luxury and sport models, t(59)=5.79, p<.001; while the symbolic factor contributed more to luxury and sport vehicles, t(59)=9.43, p<.001. A between-group (instrumental vs. symbolic instructions) analysis revealed no significant differences in these measures suggesting that subjects' perceptions of the functions served by the known vehicles were not influenced by the instructions.

We then created a manual which would allow us to code the individual beliefs according to their relevance to specific psychological needs. The responses provided by the first 30 subjects were classified as either instrumental or symbolic based on the conceptual definitions presented earlier. The resultant themes were used as examplars for the coding manual. Two judges then used the coding manual to independently code the belief statements provided by the remaining 30 subjects. The agreement rate for the 743 beliefs was 90.2% which was superior to that reported as acceptable by Shavitt (1989).

In total, the 60 subjects provided 1,256 beliefs which could be coded for the four models of cars. There were no between-group differences in the number of beliefs elicited suggesting that our instruction manipulation did not adversely affect subjects' ability to generate attitudinal beliefs. Table 2 shows, however, that subjects provided with instrumentally-oriented instructions (those traditionally used in thought-elicitation procedures) generated more beliefs relevant to utilitarian functions and fewer associated with symbolic functions than did subjects receiving symbolically-oriented instructions, t(59)=2,79, p<.01 and t(59)=2.91, p<.01, respectively. Although not shown in Table 2, we repeated these analyses for each of the four cars and found that the differing rates of elicitation were statistically significant for all but the sports model, and these results demonstrated a pattern consistent with the other models.

By tailoring instructions to effectively pull for specific types of beliefs we were able to obtain a more complete picture of the multifunctional nature of automobiles in general, and these four models in particular. This technique allows us to discriminate amongst these similar attitudinal objects in terms of their capacity to fulfill certain needs. Based on our exploration to this point, it would appear that subcompacts, followed closely by family models, are the most unifunctional of these cars, serving predominantly instrumental requirements. The luxury and sport models appear more multifunctional as they possess a greater capacity to serve symbolic, as well as utilitarian, needs.

This first study has provided a repertoire of beliefs associated with automobiles and it appears that the psychological needs met by this product are reflected in the attitudinal belief set. Taken on its own, these findings seem rather obvious: we successfully generated data consistent with instructional demands. However, when we examined the measures of overall attitude toward the four cars, we discovered there were no between-group differences (means: 2.10 vs 1.87 for subcompacts; 1.93 vs 1.20 for family; 2.10 vs 2.00 for luxury; and 2.77 vs 2.97 for sports). Despite being measured following the manipulation, attitudes do not reflect the functional distinctions that are evident amongst the attitudinal beliefs. This offers some tentative support for our contention that advancement in the functional approach will benefit by measurement techniques that focus on the attitudinal beliefs. If our methodology is to have any practical significance, however, it is essential that we establish that an attitude toward a product does indeed serve the needs suggested by the associated beliefs despite this obfuscation. Study Two explores the functional consistency between attitudes, attitudinal beliefs, and psychological needs.




We have theorized that the psychological needs that a product satisfies are reflected in the structure of the attitude rather than the attitude toward that product. In this study we examine this notion by experimentally manipulating the function of the belief set. We hypothesize that such a manipulation will affect the functions met by the attitude but will not be evident in overall attitude.

In order to conduct this experiment, we needed an automobile model that was not familiar to subjects. Study One demonstrated that known cars are perceived as serving distinct patterns of needs so we created a model that would not permit subjects to enter the experiment with set perceptions about its functional capacity. The fictitious car was identified as the "DL-22" which was "still in the design stages." No additional information was provided.

In order to manipulate the belief set associated with this unknown car we adopted a priming procedure similar to that used by Shavitt and Fazio (1987) and DeBono (1989). Our intent was to prime subjects with either instrumental or symbolic beliefs about an automobile then compare the function of their attitudes toward the car. For our priming task we generated two sets of advertising slogans appealing to instrumental or symbolic psychological needs. Each slogan was a brief statement that was worded identically except for a word or brief phrase chosen from Study One to reflect the different needs. We predicted that, in the absence of previous knowledge of the fictitious model of car, subjects would perceive it as meeting the need consistent with the priming slogans. Specifically, subjects given the instrumental version of the beliefs would rate the DL-22 as relatively unifunctional and similar to our subcompact and family models (i.e., predominantly instrumental) while those given the symbolic version would rate the DL-22 as similar to the more multifunctional luxury and sports models.


The subjects were 40 (20 female and 20 male) introductory psychology students who participated for course credit. They were brought into the lab individually and were told that the study was designed to test advertising slogans for a new car which would reach the market in the near future and their task would be to assess a variety of written slogans which might be used in future advertising for the DL-22. Subjects were then provided with 20 slogans, each printed on a separate 3" x 5" card, that were written in one of two functional versions: instrumental or symbolic (see Figure 1 for examples). This manipulation constituted the independent variable.

In order to prime the predominant function featured in the slogan cards, subjects were given a filler task which involved sorting the cards according to how appealing and, then, how convincing each was. Subjects then rated the DL-22 on two evaluative dimensions: good/bad and like/dislike. A 7 point bipolar scale was used for each and the responses were summed to form a measure of overall attitude toward the fictitious vehicle. Finally we obtained a direct measure of attitude function for the DL-22 and the four familiar car models using the same procedure from the previous study.


As in the previous study, we first examined the direct ratings of function for the known models of cars (subcompact, family, luxury and sports) and for the hypothetical DL-22. For the familiar car models, the results were consistent with the data from Study One in that planned comparisons revealed that the subcompact and family models differed significantly from the luxury and sports models for each function measured; t(39)=7.54, p<.001 for instrumental function and t(39)=6.40, p<.001 for symbolic function.

The priming manipulation was, however, effective in producing different functional perceptions of the unknown and unfamiliar DL-22. As can be seen in Table 3, subjects rated the hypothetical vehicle as serving a pattern of needs consistent with our hypothesis regarding the priming impact. A repeated measures ANOVA revealed a significant interaction between priming condition and function, F(1,38)=17.85, p<.001. Generally, the symbolic primes resulted in lower instrumental and higher symbolic ratings of the DL-22 than did the instrumental prime. Furthermore, the priming effect was limited to the DL-22; there were no significant between-group differences for the four known types of automobiles. It would appear that our manipulation had an impact on the functional nature of a novel belief structure without affecting existing attitudinal structures.

We included the four common car models in this study to serve as comparative data for the functional measures of the DL-22. Looking first at the group primed with instrumental beliefs (the first two columns of data in Table 3), we found that these subjects provided function ratings of the DL-22 that were statistically similar to the unifunctional subcompact and family cars. In other words, the DL-22 was significantly different from the luxury model, t(38)=2.70, p<.05 for instrumental and t(38)=3.32, p<.01 for symbolic ratings. The same was true for the sports model, t(38)=5.77, p<.001 for instrumental and t(38)=6.24, p<.001 for symbolic ratings. The DL-22 did not differ significantly from the subcompact and family models.

The opposite was true for subjects primed with expressive ideas (the last two columns of data in Table 3). In this instance the function ratings of the DL-22 did not differ significantly from the luxury and sports models but did differ from the unifunctional cars. These differences were at least marginally significant for the subcompact, t(38)=2.03, p<.10 for instrumental and t(38)=2.78, p<.05 for symbolic ratings. The comparisons with the family car were significant, t(38)=4.66, p<.001 for instrumental and t(38)=2.32, p<.05 for symbolic ratings.





We then examined the measure of attitude toward the DL-22 and found there were no between-group differences in mean attitude (2.05 for instrumental group and 2.00 for value-expressive group). This was also true for the four familiar car models. As hypothesized, the functional variability of the DL-22 appears to take place at the level of beliefs without being evident at the more general level of overall evaluative attitudes.


This research used the automobile as a consumer product that is capable of meeting and satisfying a variety of psychological needs. In Study Two, subjects provided with the instrumental beliefs in the form of advertising slogans came to perceive the DL-22 as being predominantly a unifunctional product. Those primed with symbolic ideas came to perceive the fictional car as being more multifunctional. As support for our thesis, recall that these distinct profiles were created by introducing function-relevant beliefs identified in Study One. Recall further that these distinct perceptions of the DL-22 were not reflected in the attitudes toward the car.

In terms of method, the belief elicitation and coding employed in Study One appears to be both efficient and valid. Once a coding manual is composed it becomes possible to empirically describe and measure the range or capacity of a product to meet consumer needs. Our ability to differentiate between car models suggests these procedures may also allow brand distinctions within broader product categories. By permitting the identification and functional classification of beliefs associated with a target item, this methodology may allow researchers to better understand and, perhaps control, the needs being served by various consumer products.

In Study Two we have evidence that attitudinal belief sets of new products may be especially susceptible to persuasive communications that might influence consumers' functional expectations. In many cases, an advertiser may simply want to know the predominant function of a product and design appeals consistent with it. There are, however, other intriguing possibilities for multifunctional products such as the automobile. If one knows the full functional profile of a product, one can then design persuasive appeals to change or enhance attitudes toward the product that would reflect the variety of needs that might be satisfied. Changing attitude function may suggest new uses for a product; generate new markets; improve brand name loyalty; or increase involvement with the product.

There are two ways that persuasive communications might accomplish this task. First, an advertiser might introduce new functional beliefs not previously associated with the product in the hopes of creating a new potential demand. For example, an economy car may be advertised as showing the owner "conserves energy" or "cares about our planet." These new expressive ideas would suggest that the economy car can serve more than the utilitarian needs such as fuel-efficiency and transportation.

The second tactic available to the advertiser is manipulating beliefs already associated with the product by improving the strength, importance, or salience of certain ideas within the existing belief structure. The targeted beliefs could be chosen on the basis of their functional affiliation thus enhancing the scope of psychological needs that the product might serve. For example, the safety features of a family car might be promoted in expressive rather than instrumental terms as "caring for your loved one" or "being a parent first and a chauffeur second." Similarly, the instrumental nature of a luxury car can be suggested by "allowing you to arrive well-rested and ready for business."

Both advertising tactics require the mapping of attitudinal beliefs and their functional associations that our methodology may provide. This mapping might also permit an ongoing evaluation of advertising and marketing programs by tracking changes in the functional profile of a product over time. Analysts would be able to empirically measure the success of programs designed to change particular thoughts or ideas within the set.

Given the early stage of our exploratory research, we must emphasize that these applications are speculative. Obviously, the studies presented here are only suggestive of the applied potential of the functional approach. Currently, we are examining the compatibility of our functional approach within the more traditional and popular notions of attitude structure as represented by the work of Ajzen and Fishbein (1980). We also plan to use persuasion techniques with familiar consumer products to directly test some of the applied possibilities suggested in this paper.


Abelson, R. P. (1986). Beliefs are like possessions. Journal for the Theory of Social Behaviour, 16, 223-250.

Ajzen, I., and Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Belk, R. W. (1988). Possessions and the extended self. Journal of Consumer Research, 15, 139-168.

Cacioppo, J. T., and Petty, R. E. (1981). Social psychological procedures for cognitive response assessment: The thought-listing technique. In T. V. Merluzzi, C. R. Glass and M. Genest (Eds.), Cognitive assessment. New York: Guilford Press.

DeBono, K. G. (1989). On the processing of functionally-relevant consumer information: Another look at source factors. Advances in Consumer Research, 16, 312-317.

Herek, G. M. (1986). The instrumentality of attitudes: Toward a neofunctional theory. Journal of Social Issues, 42, 99-114.

Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly, 24, 163-204.

Shavitt, S. (1989). Operationalizing functional theories of attitude. In A. R. Pratkanis, S. J. Breckler and A. G. Greenwald (Eds.), Attitude structure and function (pp. 311-338). Hillsdale, NJ: Erlbaum.

Shavitt, S., and Fazio, R. H. (1987, May). Attitude functions in the attitude-behavior relationship. Paper presented at the annual meeting of the Midwestern Psychological Association, Chicago, IL.

Smith, M. B., Bruner, J. S., and White, R. W. (1956). Opinions and personality. New York: Wiley.

Zanna, M. P., and Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar-Tal and A. Kruglanski (Eds.), The social psychology of knowledge. New York: Cambridge University Press.