A Functional Analysis of the Role of Overall Evaluation of Alternatives in Choice Processes

James R. Bettman, University of California, Los Angeles
ABSTRACT - A functional analysis of the use of evaluation processes during choice is developed, arguing that overall evaluations are likely to be performed under some conditions, but not others. The conditions which may lead to presence or absence of evaluation processes are analyzed.
[ to cite ]:
James R. Bettman (1982) ,"A Functional Analysis of the Role of Overall Evaluation of Alternatives in Choice Processes", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 87-93.

Advances in Consumer Research Volume 9, 1982      Pages 87-93

A FUNCTIONAL ANALYSIS OF THE ROLE OF OVERALL EVALUATION OF ALTERNATIVES IN CHOICE PROCESSES

James R. Bettman, University of California, Los Angeles

ABSTRACT -

A functional analysis of the use of evaluation processes during choice is developed, arguing that overall evaluations are likely to be performed under some conditions, but not others. The conditions which may lead to presence or absence of evaluation processes are analyzed.

INTRODUCTION

Several views exist regarding the role played by evaluations of alternatives in the consumer choice process. An evaluation of an alternative, in this discussion, will be defined as some overall judgment of the goodness or badness of the alternative. In many consumer choice models (e.g., Howard and Sheth 1969, Engel, Blackwell and Kollat 1978, Hansen 1972), evaluation processes (i.e., forming an attitude toward each alternative) appear to be viewed as a necessary part of choice. That is, an overall evaluation of each considered alternative is assumed to already exist or to be formed before a choice is made. Zajonc (1980) also argues that evaluation always occurs before choice. However, he argues that evaluation is predominant, can occur in the absence of cognitive inferences, and that the evaluative and cognitive systems are relatively independent. Finally, Bettman (1979) also claims that evaluation and choice can proceed independently, but argues that choice processes need not involve any overall evaluation of alternatives.

The latter viewpoint does not imply that attitudes and evaluations are not involved in choice processes. Rather, Bettman's notions imply only that formation of overall evaluations and the process of choice are conceptually independent. Each process can occur in the absence of the other. The purpose of the present paper is to expand upon this argument, and to claim that a uniform view of the role of evaluations in choice is incorrect; rather, a contingent view is more appealing. Hence, a functional analysis of the use of evaluation processes in decision making is presented. It is argued that evaluations are likely to be performed in certain circumstances and not in others, and the conditions which may lead to presence or absence of evaluation processes are analyzed.

In the following sections, the three viewpoints on the role of evaluations in choice noted above are discussed in more detail. Then a functional analysis of the use of evaluations in choice is presented, discussing first the conditions under which evaluations might not be performed and then the circumstances favoring the utilization of evaluations. Finally, discussions of methods for testing the hypothesized effects of such sets of conditions and the ecological frequency with which these conditions may arise are presented.

THREE VIEWS ON THE ROLE OF OVERALL EVALUATIONS OF ALTERNATIVES IN CHOICE

Overall Evaluation as an Integral Part of Choice

Most consumer choice models (e.g., Howard and Sheth 1969, Engel 9 Blackwell and Kollat 1978, Hansen 1972) assume that formation of an overall evaluation of each alternative is an integral and necessary component of choice processes. There is a heavy emphasis on the role of attitude in these theories, perhaps stemming from the great degree of reliance on linear compensatory models in research on choice. Such models, of course, provide a direct evaluation of each alternative. Hence, the use of linear models implies that formation of an evaluation is the essence of the choice process. Even if usage of linear models is not directly assumed, the allied concept of alternative evaluation appears to have been accepted in most theories of consumer choice. Thus, in this view forming overall evaluations of alternatives and the choice process are inextricably linked.

Some might argue that this depiction of current theories of consumer choice should be qualified. For example, in many of these theories, the differences between complex and routine or habitual choices are noted. In particular, the argument is made that for routine choices, problem recognition may lead directly to intention and then to choice. However, upon closer examination, it is not clear that evaluation is bypassed. Such habitual choices are i often directed by previously formed evaluations, the process Wright (1975) terms affect referral. The most detailed analysis of different types of choices is probably contained in Howard (1977), who distinguishes among routinized response behavior, limited problem solving and extensive problem solving. In the latter two cases, attitude (and hence evaluation) plays a central role (see pp. 47 and 101). In the case of routinized behavior, Howard (1977, pp. 23-30) argues that impersonal attitude is involved, which he defines as the consumer's evaluation ! of characteristics such as price and availability, which define the conditions of purchase. In addition, prior evaluations of brands are used during the process. Thus, evaluations (perhaps previously formed) still play a central and integral role in choice. A second potential qualification, the case of low involvement processes, is discussed in more detail below.

The Predominance of Evaluations

Zajonc (1980), in a provocative and controversial analysis, argues that affect and cognition are physiologically and psychologically independent systems. Although the focus of Zajonc's analysis is affect, and hence is broader than the more limited question of whether or not overall evaluations are formed, he appears to argue that evaluations always occur, generally occur prior to cognitions, and can in fact occur without accompanying cognitions. (The distinction between the presence or absence of affect and the presence or absence of overall evaluations in choice is discussed further below.) In particular, Zajonc notes that the features of stimuli most useful for cognitive tasks such as recognition or categorization, features he calls discriminanda, are unlikely to correspond to the features used in forming evaluations, which he calls preferenda. Thus, although Zajonc appears to believe that evaluation and cognition are independent systems, he also argues that evaluations always accompany choice processes.

A Contingent View of the Role of Overall Evaluations

Bettman (1979) also claims that evaluation and choice are conceptually independent, noting (pp. 180-185) that many choice heuristics (e.g., conjunctive, lexicographic, elimination by aspects and additive differences) do not yield overall evaluations of each alternative. The argument is made that choice And formation of overall evaluations are in general two different tasks that need not be interrelated. One can make a choice with or without overall evaluations of alternatives, or form overall evaluations of alternatives with or without making a choice. Given this view, one must attempt to specify the conditions under which evaluations will be used in making a choice and the conditions when choice will be made without forming overall evaluations of the alternatives considered (for excellent related discussions, see Wright 1976a, 1976b).

Prior to examining conditions under which overall evaluations may or may not accompany choice processes, the scope of the analysis must be delimited. In the following, the major focus is on cases where no strongly held prior evaluation exists and hence there is some (perhaps limited) active processing of available information. The situation of routinized decision processes where there is a strongly held prior evaluation (affect referral) is briefly considered following the main analysis. Zajonc (1980) and Wright (1976a) also make such a partitioning in their discussions. Hence such cases as first choices in a product class, choices where there are very long interpurchase times, and so on are the focus in the analysis below.

In addition, the analysis only considers the use or nonuse of overall evaluations during choice. This is a different issue than the presence or absence of affect during choice. For example, one might make a choice using a conjunctive heuristic, which does not yield an overall evaluation of each alternative. However, affect may be involved in determining the cut-off levels for each attribute, deciding which levels of the attribute are acceptable and which are not. Thus, the analysis presented below does not concern the more general topic of affect, but the more limited issue of whether or not overall evaluations of alternatives are formed.

Before considering the conditions associated with usage or non-usage of evaluations, a brief introduction to the philosophy underlying the functional analysis is in order. The approach taken is functional in the sense that it is argued that evaluations are made when they serve a function in making a choice, when the choice is made easier or somehow 'better' if evaluations are formed. The main arguments are based on the notion that evaluations will generally be formed when choice would be too difficult or much more effortful to do wit:lout using evaluations; when evaluations can be carried out; and when consumers feel their decisions might be more optimal if evaluations are made. Conversely, evaluations would not be made when they were not needed to make a choice, or when they could not be carried out easily. The specific conditions underlying these global notions are discussed below.

WHEN OVERALL EVALUATIONS MAY NOT ACCOMPANY CHOICE--THE CASE FOR COLD COGNITIONS

It is perhaps easiest to understand the factors which might lead to formation of evaluations during choice by considering the opposite case, the case where evaluations might not be undertaken. As noted above, there are many choice heuristics which do not yield direct overall evaluations of alternatives, but which could lead to choice of an alternative. Some brief and highly simplified scenarios pointing out such possibilities are considered, and then the conditions fostering choice without evaluations are detailed.

Scenarios for Choice Without Overall Evaluations

Consider a consumer who has just moved to a new state, and who is examining a bank's brochure about the various possible savings accounts available at that bank. The brochure presents a matrix display of information on the various accounts and their features (e.g., interest rates, minimum amount required, time the amount must be left in the account, and so on). The consumer is in a hurry, and quickly scans the data, looking at one account at a time, eliminating all accounts requiring a minimum deposit of more than $2,500 or with interest below 7%. This leaves only one account, which the consumer chooses. This choice could be described as a conjunctive process. Such a process does not yield an overall evaluation of each alternative, but simply accepts or rejects it. Hence, one could choose without explicitly forming evaluations of each alternative.

Next, imagine a consumer shopping in a grocery store who has just acquired a dog as a pet, and wishes to purchase a flea collar. The consumer is not very involved in the choice of a particular collar and knows very little about them. When the consumer finds the display of collars, there are only two alternatives. Each has the percentage of various ingredients, the length of time it can be used, and the price on the front of the package. The consumer quickly notes that the prices are the same, and chooses the brand with the longer usage life, ignoring the chemical information. This choice, essentially lexicographic, was made by considering information on one attribute at a time. Each alternative was not processed as an overall entity. Hence, one might argue that there would tend to be no overall evaluation of each alternative in such a case.

These brief scenarios provide instances which illustrate the potential for choice without an accompanying evaluation process. In the following, some of the factors underlying these examples are discussed.

Conditions Underlying Choice without Overall Evaluations

Now several types of factors which might lead to choice without overall evaluations are considered. The list is not intended to be complete; individual difference factors are not considered in any detail, for example. However, several major sets of factors are examined. These sets of factors include the mode of presentation and type of information available; the level of importance or involvement characterizing the choice; characteristics of the set of alternatives; characteristics of the decision situation; and the level of knowledge the consumer possesses about the product category. The necessity and sufficiency of these various factors, alone and in combination, for producing choice without evaluations are discussed later in the paper.

Mode of Presentation and Type of Information. The presence of the following conditions would encourage choice processes without overall evaluations of alternatives: i) all information considered is externally available at the point of choice; ii) the information on various brands and attributes is presented or available simultaneously; iii) the format in which the information is presented is compatible with strategies other than those based on processing one brand at a time; iv) the information available can be easily compared within attributes, across brands; and v) the information available is "cognitive" rather than "affective" in nature. The rationale for these conditions is now considered.

If conditions i, ii, iii, and iv are met, the consumer is facing an information environment which allows for a great deal of flexibility in processing the available information. If all the information considered is available at the point of choice, whether the point of choice is in a store, at home, or elsewhere (e.g., on packages, displays, in brochures, in a table in Consumer Reports), then there is a relatively complete external memory. If that information is also available simultaneously (i.e., the consumer has all the information available at once, rather than receiving it in bits and pieces), then there is no particular need to form and remember a summary evaluation of each alternative based on information obtained elsewhere or obtained earlier in the decision process. Rather, it is relatively easy for the consumer to make a choice based only on some processing of the information currently available in external memory. In addition, if the format in which the information is presented is organized by attribute, or in a matrix display, or in some other form allowing easy attribute-based comparisons (e.g., a display with very few brands, each with the same information on the package), then various heuristics based upon attribute processing which do not yield global evaluations can be used (Bettman 1979, Wright 1976a).

Conditions iv and v imply that certain types of attributes would be predominant in the information considered during the choice. One useful typology of attributes has been proposed by Nelson and others (Nelson 1970, 1974, Darby and Karni 1973). Attributes of alternatives are of three types: search (can be ascertained by inspection, prior to use--e.g., price, size); experience (can be ascertained after use--e.g., taste); and credence (cannot be ascertained even after normal usage without consulting an expert--e.g., vitamin content). The type of information characterized in conditions iv and v might often be objective search attribute information, but this need not be the only case. For example, Consumer Reports often provides ratings of brands on experience attributes. Such ratings provide a common scaling across brands, ensuring condition iv. If the same experience attribute is described in different terms across different brands, then the consumer would find it harder to process by attribute, and might tend more to look at each brand as a separate entity, leading to a greater chance for overall evaluation of each brand.

Finally, a brief explanation of condition v is in order. Some products are marketed using information which is very "cognitive," whereas others use information which is much more "feeling" or "affective" in tone. For example, bath soaps often stress how one feels after using them, rather than weight of the bar, usage life, or other features. To the extent that such "affective" attributes are the major ones stressed for a product category, development of overall evaluations might be fostered. A related distinction might be information which is expressed basically in semantic form versus that which is basically visual. It is possible that the visual information may more readily lead to evaluative responses in some individuals. (for a very interesting related discussion, see Holbrook and Hirschman 1981). In summary, conditions i-v specify an information environment which makes it very easy to make a choice without forming an evaluation of each alternative, and which fosters very "cognitive" processing.

Importance and Involvement. The conditions above allow the consumer to process without forming overall evaluations. However, there has been no consideration thus far of those motivational conditions which might lead consumers to process in such a fashion. That is, one could argue that the environment described above would also make it easier for consumers to form evaluations if they so desired. Hence, the role of the importance of the choice to the consumer or the degree of the consumer's involvement in the choice must be considered.

The simplest notion about the role of involvement would be that consumers tend to process without forming evaluations for unimportant, low involvement decisions. This is related to the prediction of the standard low involvement hierarchy (Ray 1974), which argues that cognition leads directly to behavior, which then leads to attitude change.

Despite some conceptual and empirical support, however, there is still controversy over the proposed hierarchy. Petty and Cacioppo (1981) argue that the sequence is still cognition, evaluation, and behavior in processing persuasive communications under low involvement, but that the focus of the cognitions switches from issue-relevant arguments in high involvement to non-issue-relevant features, such as properties of the source, in low involvement. Also, Leavitt, Greenwald, and Obermiller (1981) postulate the existence of a low involvement path characterized by preattentive processing and change in evaluations. Finally, Hansen (1981) argues that individuals may use evaluative right brain hemisphere processes in low involvement.

The resolution of this debate is presently unclear. For the moment, the hypothesis that consumers will tend more to process without making evaluations for uninvolving, unimportant choices will be retained. Combining this discussion with that in the previous section, the choice situations which should lead to the highest incidence of choice without evaluations are those characterized by a well-structured processing environment and uninvolving decisions.

Characteristics of the Set of Alternatives. Features of the particular alternatives considered may affect whether or not evaluations are performed. Two such characteristics are the number of alternatives and their complexity. There is some evidence that either a large or small number of alternatives can lead to usage of choice strategies which may not involve an evaluation. In choice settings with a large number of alternatives, consumers often use cut-off strategies to eliminate alternatives (Wright 1976b, Wright and Barbour 1977, Payne 1976). Such strategies include the conjunctive and elimination by aspects heuristics. Neither of these heuristics requires an explicit evaluation of the alternatives. On the other hand, as in one of the scenarios described above, there can also be choice without evaluation with a small set of alternatives. Even if the information is displayed by brand, the consumer can sometimes process by using attribute-based comparisons if the number of alternatives is small. Such attribute-based processing often can lead to choice without evaluation, as noted above.

Complexity of the set of alternatives also has an impact on the use of evaluations. The more complex the alternatives, other factors being equal, the more the consumer tends to rely on simplifying heuristics, and simplifying strategies often involve no direct evaluation of alternatives. Both the number of alternatives and their complexity thus effect the consumer's ability to form evaluations.

Characteristics of the Decision Situation. Two characteristics of the choice task which can lead to decreased use of evaluations are time pressure and distraction. Both Wright (1974, 1976a, 1976b) and Einhorn, Kleinmuntz, and Kleinmuntz (1979) note that decisions made under time pressure may involve greater use of non-compensatory choice strategies, which do not involve alternative evaluations. Greater usage of unidimensional comparisons and cut-offs is observed. Greater use of cut-offs has also been observed as the time horizons involved in a choice are shortened (Wright and Weitz 1977). Finally, distraction may also lead to decreased usage of evaluations (Wright 1974). Again, both distraction and time pressure lessen the consumer's ability to form evaluations (or do any kind of processing) in a given choice setting.

Knowledge of the Product Category. In general, one would expect that the consumer would have to possess a reasonable understanding of the product category to enable an evaluation to be formed. The consumer would need to understand the nature of the attributes, potential tradeoffs, and so on. In other words, the consumer would need to understand the product concept and choice criteria (Howard 1977), or possess a memory schema for the product category (Olson 1978). Hence, low to moderate levels of product class knowledge may lead to choice without evaluations. There is some indirect support for this notion, as several studies have found that consumers with less experience use attribute-based processing to a greater extent (Bettman 1979, Bettman and Park 1980). However, if knowledge is extremely low, the consumer may find processing any available information too difficult, and may use evaluations based on very simplistic criteria (Bettman and Park 1980). Finally, Johnson and Russo (1981b), contrary to the reasoning above, found no differences in use of overall evaluations as a function of knowledge. However, the product category used was automobiles, which may be somewhat familiar to most individuals. Thus, the role of level of knowledge in the usage of evaluations requires empirical clarification.

To summarize, several conditions which might lead to choice without forming overall evaluations have been postulated. These conditions specify a choice setting characterized by information which is conveniently structured, easily compared, and "cognitive" in nature; low involvement; and factors which might reduce the consumer's ability to develop evaluations (number and complexity of alternatives, time pressure and distraction, lack of knowledge). Hence, when the consumer lacks the ability or motivation to form evaluations, "colder," more cognitive information processing may occur.

WHEN OVERALL EVALUATIONS MAY ACCOMPANY CHOICE--HEATING UP THE CHOICE PROCESS

Scenarios for Choice with Overall Evaluations

Imagine a consumer who is about to make his or her first purchase in a product category such as bath soap, toothpaste, coffee, or beer. Such categories are characterized by a relative lack of information on the package or elsewhere in the choice setting; rather, much information comes in advertisements for the various brands. Hence, the consumer often receives information sequentially, one brand at a time. In addition, the information which is received is often very "affective" or "feeling" in tone, and is not easily comparable across brands, even for the same attribute. For example, various soaps may attempt to characterize how refreshed or invigorated one feels, or beers may stress various qualities of their taste. Such information can be difficult to process using attribute-based comparisons. For example, how one compares "gusto" and "high country" taste for two brands of beer is not obvious. In such choice situations, overall evaluations of each alternative can play a very useful role, as detailed below.

Conditions Underlying Choice with Overall Evaluations

Mode of Presentation and Type of Information. The presence of the following conditions would encourage choice processes with overall evaluations of alternatives: i) some of the information considered and impacting the choice is not externally available at the point of choice; ii) information on various brands is presented sequentially; iii) the format in which the information is presented leads to processing one brand at a time; iv) the information available cannot be easily compared within attributes, across brands; and v) the information available is "affective" or "feeling" in tone.

If conditions i to v are met, the consumer faces a choice environment which severely constrains the types of processing which can be readily carried out. In effect, the sequential, brand-based environment changes the consumer's task to one of single brand evaluation rather than multiple brand choice (Wright 1976a, 1976b). In many product categories, much of the information and persuasive claims are available in advertising messages, but not on the package or elsewhere at the point of choice. That is external memory may be relatively incomplete. If the consumer is to use such information from advertising, it must be stored in and later retrieved from memory. If, in addition, the consumer sees information sequentially, typically one brand at a time, comparisons using attribute-based processing are very difficult. The consumer would have to store information on various brands in memory, retrieve this data and make comparisons and calculations with little external memory aid. This might also be true if the format of the information were by brand, even if the information were available simultaneously. Such processing would be quite difficult.

What type of processing strategy or approach would ease this memorial and computational burden? Perhaps the easiest strategy in such a situation is to form a summary evaluation of each brand when information on that brand is presented, and then to update those evaluations as new information is received. Such a strategy is feasible, given the sequential, brand-based presentation, and reduces the burden on memory. Only the summary evaluation must be remembered, not the more detailed data which underlie that evaluation. Thus, evaluations serve an important function in such a constrained processing environment by allowing the consumer to make a choice without undue burdens on memory or computational skills.

Such formation and updating of evaluations may seem to contradict the claim above that only cases where no strongly-held prior evaluation exists would be considered. However, the intent was to consider cases where there is no prior evaluation from a previous decision episode. The decision episode examined, however, may extend over time, particularly in the case of sequential information presentation.

There is evidence and reasoning to support the analysis above. Johnson and Russo (1981) show that global evaluations are remembered better than the underlying brand-attribute ratings, and cite other studies consistent with the hypothesis that the results of internal processing are often remembered better than the original data which were processed (e.g., Russo and Wisher 1976, Dosher and Russo 1976, Johnson and Raye 1981). Hence, the consumer is apparently better able to recall evaluations which have been formed than the data underlying those evaluations. Thus, conditions which require the consumer to use information in memory rather than merely processing externally available information are likely to result in retrieval of evaluatively laden memories. Tversky (1969) argues that sequential presentation of information will encourage brand processing, and Bettman and Kakkar (1977) show that information presented in a brand format (which is by nature sequential) tends to be processed by brand. finally, Herstein (1981 p. 852), presents some strong evidence for the notions discussed above. Herstein collected protocols from 20 subjects during a voting choice task. A total of 42 overall evaluation statements occurred in these protocols, of which 40 of 42, or 95%, occurred in a condition with a format encouraging processing of one candidate at a time.

The analyses above involve conditions i, ii, and iii. Conditions iv and v also would lead the consumer to form and update summary evaluations. If information cannot be easily compared within attributes, then brand-based processes would be encouraged, where the task again becomes analysis and probably evaluation of single alternatives, rather than simultaneous comparison of several options. In additions if the information available is very "affective" or evaluative in tone, stressing experience or credence attributes, evaluations may be encouraged. As noted in the brief scenario above, the information in advertising for many products is of this sort, describing in highly "feeling" terms the taste or experience of a brand. Such information would also be difficult to compare within attributes across brands, making the tendency for forming evaluations of each brand even higher. Hence, for brands with many experience and credence attributes which are depicted in "affective" terms, where information is presented sequentially by brand, and where the consumer does not have the aid of a relatively complete external memory, it is very functional to form overall evaluations and update them.

Importance and Involvement. In addition to the impact of the conditions described above, the importance of the choice can affect the use of evaluations. Einhorn, Kleinmuntz, and Kleinmuntz (1979) argue that more important decisions will receive more consideration and deliberation, and more deliberation will lead to greater incidence of overall evaluation. The standard high involvement hierarchy (Ray 1974) argues that cognitions lead to evaluations which then lead to behavior. In addition, important or involving choices implicate the self, which may lead to greater involvement of evaluation (Zajonc 1980). Self-relevant information also appears to be better remembered (Rogers, Kuiper, and Kirker 1977), which provides additional indirect evidence that evaluations may be well remembered. That is, evaluations are more self-relevant than the basic data from which they are derived, and hence may be better remembered.

Despite these arguments, there is some indirect contradictory evidence. One might argue that to the extent consumers are involved in a choice or feel it is important, they will want to optimize that choice as much as possible. Wright (1975) found that consumers perceived linear compensatory strategies, which lead to evaluations, as less likely optimizers than conjunctive strategies, which do not lead to direct evaluations.

Characteristics of the Set of Alternatives. Evaluations will generally be easier to carry out, and hence more likely, if the alternatives are few in number and low in complexity (Wright 1976b).

TABLE 1

SUMMARY OF CONDITIONS ENCOURAGING CHOICE WITH AND WITHOUT OVERALL EVALUATIONS

Characteristics of the Decision Situation. Choice tasks which are characterized by ample time for processing (lack of time pressure) and few distractions will be more conducive to the formation of evaluations (Einhorn, Kleinmuntz, and Kleinmuntz 1979, Wright 1974, 1976a, 1976b). The arguments are the reverse of those made for the no evaluation case and are not repeated here. In addition, tasks where some kind of quantitative, graded, or ordered judgment is required will tend to evoke greater usage of overall evaluations (Einhorn, Kleinmuntz, and Kleinmuntz 1979; Johnson and Russo 1981b). Finally, Zajonc (1980) claims that social interaction is dominated by affect. Hence, choices made with others may be characterized by a good deal of information arising from that social interaction which is affective or evaluative in tone. such information, as noted above, may tend to foster the use of overall evaluations.

Knowledge of the Product Category. As noted in the earlier discussion, one might argue that the consumer must have some basic understanding of the product category to be able to form evaluations based on available information. However, it is also possible that consumers may form impressionistic evaluative judgments based on no information or very little information.

In summary, conditions which might encourage the use of overall evaluations include a choice setting with sequential, brand-organized, "affective" information; high involvement; and other factors which affect either the consumer's motivation (need for a quantitative judgment; a choice involving social interaction) or ability (few simple alternatives, no time pressure or distraction) to 7 form evaluations. Thus, two contrasting types of situations have been described, either encouraging the use or non-use of overall evaluations. A summary outline of the characteristics of each is presented in Table 1.

How the Conditions May Interact

In the above discussions, the various sets of conditions have been treated more or less independently. However, one might wish to consider how these sets of conditions might interact. That is, what might happen if information presentation is sequential and there is low involvement? The following is an admittedly speculative discussion of such interactions.

For a choice to involve overall evaluations, it may be sufficient if any of the five constraining mode of presentation and type of information conditions described in the discussion underlying choice with evaluations are met. These factors appear to be so strong that their presence may lead to use of overall evaluations of each alternative regardless of the levels of other factors such as involvement, task characteristics, or knowledge. Certainly if several of the five conditions hold, there are very strong pressures for use of evaluations. It is also possible that demands for a graded judgment and choices involving social interaction would consistently evoke evaluations, but these features may not be as strong as mode of presentation and information type.

Level of involvement alone is probably not sufficient to insure evaluations. Other factors probably need to be present, such as few alternatives, knowledge, lack of time pressure, and so on. If several of these other conditions are present, then one might find use of overall evaluations under high involvement even when concrete "cognitive" information is presented simultaneously, say in matrix format. Finally, factors such as knowledge, time pressure and number of alternatives need other conditions such as involvement or information mode to be met before evaluations would occur.

A choice would probably not involve overall evaluations only if all the simplifying presentation mode and type of information conditions specified in the discussion of when evaluations may not accompany choice are met and if some other conditions such as low involvement, lack of knowledge, or time pressure hold. These speculative characterizations obviously have implications for the frequency with which evaluations might occur in actual choices. However, discussion of this point is deferred for the moment so that the case where prior evaluations are present can be briefly considered.

WHEN PRIOR EVALUATIONS ARE PRESENT

If the consumer has made previous choices in a product category or in some other fashion has developed evaluations of the alternatives, a particularly simple mode of decision making is to simply use those evaluations to make the current choice, without processing additional information. Such a procedure, termed affect referral by Wright (1976a), may be especially useful for habitual and routinized choices. Wright (1976a, pp. 120-121) provides a good discussion of conditions which may evoke affect referral, including existence of prior evaluations, low involvement, time pressure or distraction, and formats incompatible with other simple strategies.

IMPLICATIONS

Testing the Proposed Conditions

Testing the proposed conditions is not an easy task, as detection of the presence or absence of overall evaluations is not likely to be straightforward. However, several methods seem possible. One would first manipulate some set of factors discussed above (e.g., format, presentation mode, time pressure), and have the individual make a choice. The dependent measures might then include:

a) use of response times to memory probes asking about global evaluations of alternatives (Johnson and Russo 1981a). If an evaluation had been formed during choice, response times to the probe should be reliably faster than if one had not been formed; b) use of think-aloud protocols, which could then be coded for the frequency of use of evaluative statements (Herstein 1981, Bettman and Park 1980), or use of recall protocols (Johnson and Russo 1881b); c) use of electromyogram measures of muscular activity (Cacioppo and Petty 1981). Cacioppo and Petty note that it may be possible to test the temporal sequencing of cognitive and evaluative reactions to stimuli using this technique. Thus, although tests of the notions above may be difficult, there are several potential methods for approaching the problem.

Frequency of Evaluations in Actual Choice Tasks

Having described the types of conditions which may lead to use or non-use of evaluations, the question which naturally arises is, "How often will overall evaluations be used in making choices?" Based on the characterizations of these conditions, it appears that global evaluations may very often accompany choice. Many choices will be made using affect referral, i.e., most choices are routine or habitual. For choices which are not routine, sequential presentation, incomplete external memory, presentation by brand, and some "affective" information are likely to be very common. Put another way, the well-structured information environment postulated to foster choice without evaluations may occur relatively infrequently (perhaps most prevalently in the choice tasks given subjects in experiments on choice).

Thus, one would expect to see overall evaluations of alternatives accompany choice processes quite frequently. Therefore, positions such as those held by Zajonc (1980) and several consumer choice theorists, that evaluations are a necessary component of choice, may agree empirically with the present contingency position. However, there is an important conceptual distinction, in that the functional approach outlined above argues that evaluations need not accompany choices even if they very often do so.

The implication of this line of reasoning is that the study of evaluation processes and attitudes should play an important role in cognitive information processing approaches to choice, and that careful consideration must always be given to the structure of the environment within which choices are made. "Cold" and "hot" theories of choice both have a role in understanding decision processes; rigorous analysis of choice tasks and the decision environment can help elucidate those roles more precisely.

REFERENCES

Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Readings Massachusetts: Addison-Wesley.

Bettman, James R., and Kakkar, Pradeep, (1977) "Effects of Information Presentation Format on Consumer Information Acquisition Strategies," Journal of Consumer Research, 3, 233-240.

Bettman, James R., and Park, C. Whan (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis," Journal of Consumer Research, 7, 234-248.

Cacioppo, John T., and Petty, Richard E. (1981), "Electromyograms as Measures of Extent and Affectivity of Information Processing," American Psychologist, 36, 441-456.

Darby, Michael R., and Karni, Edi (1973), "Free Competition and the Optimal Amount of Fraud," Journal of Law and Economics, 16, 67-88.

Dosher, Barbara Anne, and Russo, J. Edward (1976), "Memory for Internally Generated Stimuli," Journal of Experimental Psychology: Human Learning and Memory, 2,

Einhorn, Hillel J., Kleinmuntz, Don N., and Kleinmuntz, Benjamin (1979), "Linear Regression and Process Tracing Models of Judgment," Psychological Review, 86, 465-485.

Engel, James F., Blackwell, Roger O., and Kollat, David T. (1978), Consumer Behavior, Third Edition, Hinsdale, Illinois: Dryden.

Hansen, Flemming (1972), Consumer Choice Behavior: A Cognitive Theory, New York: Free Press.

Hansen, Flemming (1981), "Hemispheral Lateralization: Implications for Understanding Consumer Behavior," Journal of Consumer Research, 8, 23-36.

Herstein, John A. (1981), "Keeping the Voter's Limits in Mind: A Cognitive Process Analysis of Decision Making in Voting," Journal of Personality and Social Psychology, 40, 843-861.

Holbrook, Morris B., and Hirschman, Elizabeth C. (1981), "Consumer Fantasies, Feelings, and Fun: The Experiential Aspects of Consumption," unpublished working paper.

Howard, John A. (1977), Consumer Behavior: Application of Theory, New York: McGraw-Hill

Howard, John A., and Sheth, Jagdish N. (1969), The Theory of Buyer Behavior, New York: Wiley.

Johnson, Eric, and Russo, J. Edward (1981a), "Consumer Knowledge of Familiar Products," Unpublished manuscript.

Johnson, Eric (1981b), "Product Familiarity and Learning New Information," in Advances in Consumer Research, Vol. 8, ed. Kent B. Monroe, Ann Arbor: Association for Consumer Research, 151-155.

Johnson, Marcia K., and Raye, Carol L. (1981), "Reality Monitoring," Psychological Review, 88, 67-85.

Leavitt, Clark, Greenwald, Anthony G., and Obermiller, Carl N. (1981), "What Is Low Involvement Low In?" in Advances in Consumer Research, Vol. 8, ed. Kent B. Monroe, Ann Arbor: Association for Consumer Research, 15-19.

Nelson, Phillip (1970), "Information and Consumer Behavior," Journal of Political Economy, 78, 311-329.

Nelson, Phillip (1974), "Advertising as Information," Journal of Political Economy, 81, 729-754.

Olson, Jerry C. (1978), "Theories of Information Encoding and Storage: Implications for Consumer Research," in The Effect of Information on Consumer and Market Behavior, ed. Andrew A. Mitchell, Chicago: American Marketing Association, 49-60.

Payne, John W. (1976), "Task Complexity and Contingent Processing in Decision Making: An Information Search and Protocol Analysis," Organizational Behavior and Human Performance, 16, 366-387.

Petty, Richard E., and Cacioppo, John T., (1981), "Issue Involvement as a Moderator of the Effects on Attitude of Advertising Content and Context," in Advances in Consumer Research, Vol. 8, ed. Kent B. Monroe, Ann Arbor: Association for Consumer Research, 20-24.

Ray, Michael L. (1974), "Consumer Initial Processing: Definitions, Issues and Applications," in Buyer/Consumer Information Processing, ed. J. David Hughes and Michael L. Ray, Chapel Hill: University of North Carolina Press, 145-156.

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.

Russo, J. Edward, and Wisher, Robert A. (1976), "Reprocessing as a Recognition Cue," Memory and Cognition, 4, 683-689.

Tversky, Amos (1969), "Intransitivity of Preferences," Psychological Review, 76, 31-48.

Wright, Peter L. (1974), "The Harassed Decision Maker: Time Pressures, Distractions, and the Use of Evidence," Journal of Applied Psychology, 59, 555-561.

Wright, Peter L. (1975), "Consumer Choice Strategies: Simplifying vs. Optimizing," Journal of Marketing Research, 11, 60-67.

Wright, Peter L. (1976a), "An Adaptive Consumer's View of Attitudes and Choice Mechanisms, As Viewed by an Equally Adaptive Advertiser," in Attitude Research at Bay, ed. William D. Wells, Chicago: American Marketing Association, 113-131.

Wright, Peter L. (1976b), "Conditional Consumer Choice Processes and Advertising Strategy: An Introduction to the Principle of Control via Advocacy and the "Mod Squad" for Advertising Strategists," in Attitude Research Moves Ahead, ed. Yoram Wind, Chicago: American Marketing Association, 101-106.

Wright, Peter L. and Barbour, Frederic (1977), "Phased Decision Strategies: Sequels to an Initial Screening," in North Holland/TIMS Studies in the Management Sciences Volume 6: Multiple Criteria Decision Making, ed. Martin K. Starr and Milan Zeleny, Amsterdam: North Holland, 91-109.

Wright, Peter L. and Weitz, Barton, "Time Horizon Effects on Product Evaluation Strategies," Journal of Marketing Research, 14, 429-443.

Zajonc, R.B. (1980), "Feeling and Thinking: Preferences Need No Inferences," American Psychologist, 35, 151-175.

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