A Study of the Concept of Affective Choice Mode For Consumer Decisions

Banwari Mittal, University of Miami
ABSTRACT - The concept of Affective Choice Mode is developed to reflect decision processes for highly involving products that do not lend themselves to extensive information processing. This construct is contrasted with the conventional "information processing mode." The two concepts are then set in a context marked by the consumer involvement in a brand-decision task and potential 'expressiveness' of products. Data from an experimental study based on advertising stimuli for hypothetical brands in four product categories are utilized to empirically test the proposed concepts and their interrelationships.
[ to cite ]:
Banwari Mittal (1994) ,"A Study of the Concept of Affective Choice Mode For Consumer Decisions", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 256-263.

Advances in Consumer Research Volume 21, 1994      Pages 256-263


Banwari Mittal, University of Miami

[on leave from Northern Kentucky University]


The concept of Affective Choice Mode is developed to reflect decision processes for highly involving products that do not lend themselves to extensive information processing. This construct is contrasted with the conventional "information processing mode." The two concepts are then set in a context marked by the consumer involvement in a brand-decision task and potential 'expressiveness' of products. Data from an experimental study based on advertising stimuli for hypothetical brands in four product categories are utilized to empirically test the proposed concepts and their interrelationships.

Consider a consumer buying an alarm clock radio. She inspects several models, noticing many differences among them. Some models have a snooze-alarm control feature, some don't. Some have a battery back-up, others don't. The models also vary on wake-to-music or to-alarm feature, top-mounted versus side controls, push-button versus rotary or sliding switches, lighted alarm-set indicator, automatic FM frequency control, the type of wood grain finish, and the price. She reviews her relative preference for these diverse features, and chooses the model that gives her the best combination of the desired features.

Now consider her buying a dress for an upcoming big social event. Scanning a rack full of dresses in a store, she pulls out a few that seemed nice. One of them particularly caught her eye; she tries it on, and thinks she looks great in it. She tries another one, which she thought made her look too conservative. A third one made her look too sexy. Somehow, the first one looked so right for her; a few more minutes of contemplation about what a great impression she would make donning that dress in the party, and she has made up her mind about that dress.

The above two episodes illustrate two alternative choice modes, termed here, respectively, the Information Processing Mode, and the Affective Choice Mode. These are explained below.


The Information Processing Mode. In his seminal book, An Information Processing Theory of Consumer Choice, Bettman (1979) describes the cognitive operations that a consumer performs in a brand decision context. The consumer is thought to acquire information about brand attributes, form evaluative criteria, judge the levels of these attributes in various brands, and employ some judgment rule or heuristic to combine these attribute-levels for overall brand evaluation. MacInnis and Price (1987) refer to the above as "discursive" or "descriptive" information processing, and point out that this has been the traditional focus of consumer information processing research.

The Affective Choice Mode. Zajonc (1980) suggested that the above view of consumer choice appears rather limited, and that not all consumer decisions follow such a path. He argued, "the assumption questioned here is that affect, such as that contained in preferences, is necessarily post cognitive, which implies that a feeling of a preference is generated upon the encoding of the specific properties of the object, after the evaluation of their utilities, and after the computation of the individual component utilities into a joint product that represents the overall preference" (Zajonc and Markus 1982, p. 125). Zajonc suggests that affect (i.e., liking) for an object or a brand can occur through other psychological processes which do not entail any elaborate cognitive operations. This alternative mode of preference formation may be termed Affective Choice Mode (ACM).

As one possibility, Zajonc suggests the "mere exposure" theory: when a consumer is repeatedly exposed to a stimulus, s/he develops a positive feeling for the object; this occurs due to sheer familiarity with the object, as one experiences a "glow of warmth, a sense of ownership, a feeling of intimacy." A related mechanism is habituation, which is "repeated use," rather than "repeated exposure". Yet another mechanism is "social reinforcement": teenagers and impressionable adults alike come to develop a preference for specific brands of conspicuous products (Banana Republic clothing) based not on any intrinsic qualities of the brand, but rather on the kind of following the brand has acquired, and on the brand's popularity among one's peers.

In each of the above "explanations" of preference formation, the important idea is that the component features of the object do not play a role as they do in the traditional, IP mode. Rather, it is a global, overall view of the object, a "holistic impression" if you will, that constitutes product appraisal. This appraisal can be, and often is, instantaneous, as in "first impressions," or as in "love at first sight." But it can also be the outcome of long term habituation, as in, say, "getting used to a colleague." In either case, the affect felt toward the object can be, and often is, independent of the encoding and processing of any component features or attributes of the object.

In such affective appraisals, the appraiser thinks not as much about the object per se, as about oneself in relation to the object. One asks, for example, how would I look in this hairdo, or does this watch suit my upscale conservative image? In the IP mode in contrast, as for example in buying a personal computer, the focus is on the object without much reference to self-image.

Affective appraisals also differ in how consumers explain their decisions and choices. Preferences formed in the IP mode are explained in terms of the "component" features on which these preferences are based. A personal computer is liked or not liked because of its memory capacity, speed, and connectivity, for example. In contrast, preferences based on cognition-independent affect are explained in overall global terms, e.g., "I liked this dress because its style is good." Or even more notably, they are explained by mere paraphrasing, i.e., not explained at all; for example, "I chose this dress because I liked it," or "I liked it because it is so 'cool'." Understandably, in common parlance, love at first sight is referred to as being "blind."

Low Involvement Is Not ACM. In describing the IP mode, Bettman (1979) argues, correctly, that these processes may be used extensively in some cases and only minimally in others. The contingent factor is Involvement, the extent to which a decision is risky or important. When the purchase is expensive and/or entails other social or psychological risks, consumer engages in the IP activity extensively. On the other hand, in most low price, low risk or otherwise trivial purchase decisions, the consumer processes sparse product information only cursorily (Mitchell 1981, Laurent and Kapferer 1985). Thus, levels of involvement correspond to the levels of the IP mode.

This may lead one to assume that ACM occurs in the low involvement decision task environment, and IPM in the high involvement task environment. This assumed low/high involvement and ACM/IPM correspondence is misguided. In our view, ACM and IPM are not opposites but rather simply different. The absence or low-level presence of IPM does not in and of itself imply the presence of ACM. In low levels of IPM, sparse, feature specific information may determine the final choice; for example, the brand of salt may be chosen simply by noting whether or not the brand is "iodized." ACM, on the other hand, entails or may entail as much deliberation as a high IPM-based choice. A dress, generally bought in the ACM mode, is by no means a low involvement decision. What matters is not the extent of deliberation, but rather the focus of deliberation: component features in the IP mode, and holistic, self-implicating object impressions in ACM. See Mittal (1988) for further discussion.

Product Expressiveness and ACM. The foregoing postulation is grounded in Park and Mittal (1985) and Park and Young's (1986) distinction concerning "cognitive" and "affective" involvement. According to these authors, cognitive involvement occurs when consumers "process attribute-based messages or engage in self-generated cognitive reasoning about message contents." Affective involvement occurs, in contrast, when a person identifies a new stimulus with an examplar (i.e., s/he encodes it as similar to something in memory with which an affect is already attached), and then automatically transfers that affect to the brand itself. A person reacts to the stimuli based on prior experiences, emotional experience the stimulus engenders, or social and personality images associated with the product's use, etc. For example, on seeing a particular dress, a consumer may be reminded of a favorite TV star who may have once donned a similar dress; this memory retrieval would then attach the retrieved affect or liking (or disliking) to the present stimulus. More commonly, such "extrinsic" ("extrinsic" to the product, that is) affect-laden exemplars are presented in the brand communications themselves, e.g., an ad featuring a football star in conjunction with the product. This distinction is also echoed in recent social-psychological literature on affective and cognitive types of attitudes (Millar and Millar 1990, Edwards 1990).

Not all products lend themselves to ACM appraisal. This may be because the affect-laden exemplar just does not seem relevant to the product category, e.g., featuring Cindy Crawford in an ad for table salt or for a personal computer. What matters most for such products is what the brand does, in physical terms, or how the brand performs objectively. Park and Young (1986) refer to such products as 'Utilitarian." The other category of products are called "value expressive." These are bought and consumed primarily for the emotional outcomes or the personality image that their use enables the user to express. As a slight departure from this utilitarian/expressive dichotomy, we recognize the presence of the utilitarian function in all products at least to some degree; e.g., a dress has an obvious functional, utilitarian aspect actually, apart from its expressive value. In addition to their varying functional utility, they have less or more expressiveness. Thus, we consider "expressiveness" as an independent characterization of objects (Mittal 1988). It is this "expressive aspect" of products, then, that leads to ACM: the more expressive the product, the more likely it would be appraised via ACM.

Hypotheses. The purpose of the empirical research reported below is to study the concept of affective choice mode and its relationship with related concepts. We expect involvement to be related positively to the IP mode; in contrast, we expect the "expressiveness" to be positively associated with ACM. Finally, even though, IPM and ACM do not conceptually imply the opposite of each other, in most empirical situations we would expect the two to be negatively related. Moreover, considering that affective judgments are more readily formed than cognitive judgments, we would model ACM to be negatively influencing IPM rather than the other way around. That is, the greater the extent to which ACM is utilized, the lesser the use of IPM. We turn next to an empirical study to test these expectations.


Student subjects from undergraduate business classes were exposed to two advertisements, one for each of the two hypothetical brands of a product category. Four product categories were chosen: shampoo, perfume, pencil, and greeting cardC selected due to their (a) pertinence to and familiarity among students, and (b) representation of a broad range of the two antecedent variablesC involvement and expressiveness. A priori, pencils were deemed to be low on both, greeting card to be moderate to high on both, shampoo to be moderate to high on involvement and low to moderate on expressiveness, and perfume to be high on both involvement and expressiveness. Also, perfume and greeting card were expected to be processed via ACM, shampoo via the IP mode, whereas both modes were expected to be low for pencils.

Each ad was a color print ad, converted, for purposes of large-group sessions, into a set of 6 to 8 slides with synchronized audio. The ad content for the two brands in each product category differed considerably, but in overall terms, both the shampoo ads and likewise both the pencil ads featured their utilitarian attributes (e.g., cleanses oily deposits for shampoo, and easy to sharpen, break resistent for pencils); in contrast, both the perfume and greeting card ads focussed on emotional/personality aspects (e.g., "When You want a certain presence," for perfume, and "fun to send and fun to receive" for cards). The ad content was designed not as a particular treatment, but rather to provide, collectively, adequate variation on the effects they were to generate.

Procedure. Subjects were randomly assigned to one of the four product categories: 21 for shampoo, 21 for pencil, 20 for perfume, and 20 for greeting card, each product run as a separate session. Perfume subjects were all females while the subjects for the other three products were both males and females. (Of the 35 female student subjects available, 20 were drawn randomly and assigned to the perfume condition. The remaining 15 female and all 47 male students were together assigned randomly to the other three products.) The study was introduced as a research sponsored by a local advertising firm interested in testing ads for two new brands. The subjects were told to make up their minds as to which of the two advertised brands they would like to buy. They were also informed that at the end they would be given a free sample of their preferred brand. After being shown the two ads twice (to provide adequate processing opportunity), subjects answered the research questionnaire.

Measures. Measures elicited for the four constructs (Involvement, Expressiveness, IPM, and ACM) are shown in Exhibit 1. Given that the experimental task was brand-decision, purchase involvement was the relevant concept, measured here by Mittal's (1989) scale. Expressiveness is measured by the perceived relevance of a set of consumption goals for the specified products. The information processing measures used here are drawn from prior literature (e.g., Laurent and Kapferer 1985). The ACM measures are based on our description of the concept in the foregoing. To supplement the structured measures, respondents were asked two open ended questions: "Why did you prefer this brand?" and "Why did you not prefer the other brand?"

Analysis and Results

Factor Analysis. Since the four constructs are measured by multiple items, the data were factor analyzed using principal components method. Five factors emerged with eigenvalues >1.0 and 75.8% total variance explained. As Table 1 shows one factor each emerged for involvement, IPM and ACM. Although there are some cross-factor minor loadings, the dominant pattern of item-factor loadings confirms that the measures hypothesized to operationalize a concept cohere. A notable exception is the emergence of two (rather than one) factor for the ten items of expressiveness. The first dimension (expressiveness-I) reflected the public-display aspect (Items 6 through 9 in Table 1); the second dimension (expressiveness-II) captured the aspect of one's inner enjoyment and self-concept congruence (Items 10 through 15 in Table 1). Subsequently, we found a parallel in the self-consciousness literature (e.g., see Scheier and Carver 1980), suggesting that a person's attentional focus can be on either one's private self or public self. In the following analysis, we treat expressiveness as a bi-dimensional construct.



Reliability. The coefficient alpha for internal reliabilities was: .938 for involvement, .885 for Expressiveness-I, .946 for Expressiveness-II, .767 for IPM, and .670 for ACM. The relatively lower reliability for ACM points to the need for further development for this scale.

Protocol Analysis. The two open ended questions ("Why did you prefer the brand you chose?" and "Why did you not prefer the brand you did not choose?") were coded independently by three judges. Each sentence was coded in one of the following two categories (see sample coding in Exhibit 2):

(1) IP Mode Related Thoughts. This included mention of an intrinsic feature of the brand, or of a consumer benefit whose linkage to intrinsic features is transparent, or reasons that concern a physical outcome from product use, and/or use of any cues for such benefits and outcomes.

(2) ACM Related Thoughts. Mention of personality-, personal images, personal enjoyment, or other psycho-social benefits.

The inter-judge agreement was 92% for pencils, 87% for greeting card, 81% for perfume, and 73% for shampoo. The differences were resolved through mutual discussions (with 7% unresolvables dropped). The frequency count in each category served as the measure of protocol-based IPM and ACM scores.

Product scores on the four constructs. Construct scores, computed by averaging the corresponding items in the structured questionnaire, are displayed in Table 2. As expected, pencils were the lowest on involvement and expressiveness, and also on IPM. Shampoo and perfume were moderate and comparable on involvement. Greeting cards scored the highest on involvement and equaled perfume on expressiveness-II (the personal enjoyment factor). On information processing, shampoo and perfume scored about equal, each scoring higher than the other two products. The ACM score for pencils was somewhat higher than one would expect, but it did not differ from the ACM score for shampoo. As expected, perfume and greeting cards both scored higher on ACM than did the other two products.

Also shown in Table 2 are protocol-based scores on ACM and IPM. Based on the reasons for the liked brand, perfume showed more ACM than did shampoo. Greeting card also showed more ACM than did pencils or shampoo. And, shampoos and pencils showed more IPM than did perfumes or greeting cards.

LISREL Estimates of the Model. The model in Figure was estimated by a maximum likelihood, LISREL-IV procedure (Joreskog and Sorbom 1984). The chi-square statistic shows Model 1 to have a poor overall fit (chi-square=9.10, df=3, p=.028). Inspection of the modification indices showed the need to free the causal path from expressiveness-I to IPM. Because expressive products do not lend themselves to being assessed via detailed cognitive operations, this link would seem logical, and if significant, would be expected to be negative in valence. The revised model, Model 1a showed good overall fit (chi-square=.83 (d.f.=2, 'p'=.66). Bentler and Bonnet's (1980) incremental fit index also shows model 1 to be poor but model 1a to have good fit.

Structural Path Estimates. In Models 1 and 1a, all paths are significant except the expressiveness-I to ACM path which is positive in valence (as expected) but falls below statistical significance. Expressiveness-II was positively and significantly related to ACM. Involvement had a positive and significant effect on IPM. Finally, the significance of ACM-to-IPM path (in Models 1 and 1A) provides support for the hypothesized negative association between these alternative choice modes.

Model Estimates Using Protocol Data. The lower panel in Figure presents similar estimates, but with the protocol-based measures for IPM and ACM. As can be seen, Model 1 had good overall fit, as had Model 1a. However, Model 1a was not superior to Model 1 as the chi-square difference between the two was not statistically significant for one degree-of-freedom difference. In terms of the hypothesized paths, again expressiveness-II was but expressiveness-I was not significantly related to ACM. However, involvement-to-IPM link was not statistically significant (discussion later). The support for the ACM-to-IPM path was robust. Finally, in Model 1a, the released link of expressiveness-I to IPM was significant and also had a negative sign which is consistent with prior expectations.







Discussion. Some aspects of our research design bear closely on our results. Because the experimental setting virtually ensured a high level of attention to the ads, subjects are likely to have processed information simply because the information was presented to them.

The expected relationship of ACM with other concepts was observed. The magnitude of beta for the ACM to IPM link was robust in both but stronger in the protocol data (-.55 versus -.28) Likewise, the expressiveness-II to ACM linkage is stronger for the protocol- than for the structured-measures based data (beta =.49 versus .25). The failure to obtain significant linkage of expressiveness-I to ACM in both estimates is an oddity. A plausible explanation is that the appropriateness of brands for social prestige and public display (which is what expressiveness-I reflects) is judged in the real world from day-to-day observations of brand-users' life-styles in one's social environment, and the experimental stimuli did not provide for such opportunity.

Finally, the involvement-to-IPM linkage needs a comment. This linkage is significant in direct measure data but not in the protocol data. In direct measures, respondents' real world decision experience (high IP for high involvement product) may have favorably influenced the consistency in the reported involvement and the reported IPM. In the protocols on the other hand, high IPM may have been reported in high as well as low involvement products simply because the information was presented. Thus, even a low-involving product like pencil scores high on protocol based IPM (see Table 2); this would appear to be responsible for the nonsignificance of the InvolvementC>IPM link in the protocol based analysis.

On the other hand, the failure of this path in this particular analysis may imply that the hypothesis is actually false. As Table 2 shows, the protocol scores on IPM are quite low for the relatively high involvement products perfume and greeting card. Indeed, a major theme of this paper has been that some high involvement products (the ones that are expressive) are appraised not via IPM, but via ACM. The IPM relationship may have been supported in the past due to the inclusion of primarily utilitarian products.

As already argued, IPM and ACM are negatively related but they are not mirror images of each other. Rather, they are distinct concepts. This is because both can exist at low levels (as for a low involvement product, like light bulb) or at high levels (as for someone buying a car if the performance features as well as stylistic aspects of the car are important).



In arguing that when products are highly expressive, the brand is likely to be appraised via ACM, and that information processing is less likely to occur (this being the major theme of the present paper), information processing was defined in a specific way. It was defined as a cognitive algebra performed on a product's intrinsic attributes. Were it to be instead defined more broadly as attention to and encoding of the brand as a whole, then any mental operation performed on the brand (e.g., contemplation of the brand in relation to self, observation of personalities of other brand users, etc.) would qualify as information processing. The processing differences between products that are and are not expressive could not then be studied. Thus, although the definition of information processing is somewhat arbitrarily limited, that limited conception has guided much of the consumer decision literature. It is of value to study the boundaries of that limited conception, and to explore choice processes that transcend it. The investigation of ACM in the present paper has been an attempt in this direction.


Bettman, James R., 1979. An information processing theory of consumer choice, MA: Addison-Wesley Publishing Company, Inc.

Edwards Kari (1990), "The Interplay of Affect and Cognition in Attitude Formation and Change," Journal of Personaity and Social Psychology, Vol. 59 (2), 202-216.

Hirschman, E.C., 1980. "Attributes of attributes and layers of meaning," in Advances in Consumer Research, Vol 7, J.C. Olson (ed.), Ann Arbor, MI: Assoc. for Cons. res.

Hirschman, E.C. and M.B. Holbrook, 1982. "Hedonic consumption: emerging concepts, methods and propositions," Journal of Marketing, 46, 92-101.

Joreskog, Karl G. and Dag Sorbom, 1984. LISREL VI: Analysis of Linear Structural Relationships by the Method of Maximum Likelihood, Mooresville, IN: Scientific Software Inc.

Kroeber-Riel, W., 1979. "Activation research: psychological approaches in consumer research," Journal of Consumer Research, 5 (March), 240-250.

Laurent, Gilles and Jean-Noel Kapferer, 1985. "Measuring Consumer Involvement Profiles," Journal of Marketing Research, 22, 41-53.

Lutz, K.A. and R.J. Lutz, 1979. "Effects of Interactive Imagery on Learning: Application to advertising," Journal of Applied Psychology, 62, 493-98.

MacInnis, Deborah J. and Linda Price (1987). "The Role of Imagery in Information Processing: Review and Extensions," Journal of Consumer Research, 13 (March), 473-91.

Millar, Murray G. and Karen U. Millar (1990), "Attitude Change as a Function of Attitude Type and Argument Type," Journal of Personaity and Social Psychology, Vol. 59 (2), 217-228.

Mitchell, A.A., 1981. "The Dimensions of Advertising Involvement," ed. K. B. Monroe, Advances in Consumer Research, Ann Arbor, MI: Assoc. for Cons. Res., 25-35.

Mittal, Banwari (1989), "Measuring Purchase-Decision Involvement," Psychology and Marketing, 6, 147-162.

Mittal, Banwari (1988), "The Role of Affective Choice Mode in the Consumer Purchase of Expressive Products," Journal of Economic Psychology, 9, 499-524.

Park C. Whan and Banwari Mittal (1985), "A Theory of Involvement in Consumer Behavior: Problems and Issues," in Jagdish N. Sheth (ed.), Research in Consumer Behavior, Vol. 1, Greenwich, CT: Jai Press, Inc. pp. 201-231.

Park, C. Whan and S. Mark Young, 1986. Consumer response to television commercials: The impact of involvement and background music on brand attitude formation. Journal of Marketing Research, (Feb.) 11-24.

Raaij, W.F. van, 1984. "Affective and Cognitive Reactions to Advertising," Report No. 84-111, Cambridge, MA: Marketing Science Institute.

Scheier, M.F. and C.S. Carver, 1981. "Private and public aspects of self," L. Wheeler (ed.), Review of Personality and Social Psychology, 2, Beverly Hills, CA: Sage.

Solomon, Michael R. (1983), "The Role of Products as Social Stimuli: A Symbolic Interactionism Perspective," Journal of Consumer Research, 10 (Dec.), 319-329.

Zajonc, Robert B., 1980. "feeling and thinking: preferences need no inferences," American Psychologist, 35 (February), 151-75.

Zajonc, Robert B. and Hazel Markus, 1982. "Affective and Cognitive Factors in Preferences," Journal of Consumer Research, 9 (Septem), 123-31.