Consumer Information Processing For Different Information Structures and Formats

ABSTRACT - Consumer information processing behavior is influenced by the structure and format of the available product information (alternatives X attributes) in the choice task. Four information structure modes (paired, by alternative, by matrix, and sequential) and six information formats (unique vs. common attributes, attribute variability, major vs. minor attributes, absolute vs. relative attributes, presentation with or without brand name, and unspecified) provide 24 possible information presentation modes. In only 12 of these modes, research has been conducted in order to assess consumer information processing behavior. This research is reviewed in this paper.


W. Fred van Raaij (1977) ,"Consumer Information Processing For Different Information Structures and Formats", in NA - Advances in Consumer Research Volume 04, eds. William D. Perreault, Jr., Atlanta, GA : Association for Consumer Research, Pages: 176-184.

Advances in Consumer Research Volume 4, 1977   Pages 176-184


W. Fred van Raaij, University of Illinois

[This research has been supported by a grant from Tilburg University, The Netherlands, and the experiments have been conducted while the author was Assistant Professor of Economic Psychology at Tilburg University. The author is Visiting Assistant Professor (1976-1977) at the Department of Business Administration, University of Illinois at Urbana-Champaign.]


Consumer information processing behavior is influenced by the structure and format of the available product information (alternatives X attributes) in the choice task. Four information structure modes (paired, by alternative, by matrix, and sequential) and six information formats (unique vs. common attributes, attribute variability, major vs. minor attributes, absolute vs. relative attributes, presentation with or without brand name, and unspecified) provide 24 possible information presentation modes. In only 12 of these modes, research has been conducted in order to assess consumer information processing behavior. This research is reviewed in this paper.

Consumers tend to use simplifying heuristics in making a choice. These heuristics often lead to a suboptimal choice from the consumer viewpoint. A task for consumer education is not only the provision of relevant product information but also the study how this information is processed in the brand/variant choice.

Two process tracing methods are compared: Eye-movement recording and analysis of information card sequence in an information display matrix (IDM).


In attitude and utility theory, several models exist for predicting consumer choice. In these models, it is assumed that a linear or nonlinear relationship exists between the evaluation of the product attributes and the attitude toward the product, or the overall utility of the product. These models may perform their predictive role very well but it is questionable whether these models describe consumer information processing behavior (descriptive validity). In this paper, we review consumer information processing behavior by direct monitoring of the choice process.

Our objectives are:

1. To assess what kind of simplifying heuristics are employed in information processing;

2. To assess which attributes are important in the choice process and not only (normatively) rated as "important";

3. To investigate "satisfying" behavior instead of "maximizing" utility;

4. To find behavioral measures or indexes in the choice process, in lieu of verbal reports or protocols of the decision process as provided by the consumer after the choice task;

5. To assess the discrepancies of what people say and what people do (attitude-behavior discrepancy);

6. To develop new models of consumer choice, whenever observed patterns in the processing sequence give rise to this.

These objectives are only partly covered in this paper but are more extensively treated in a forthcoming publication (Van Raaij, 1977).

Verbal protocol elicitation has certain disadvantages. Self-censorship and reporting bias of the consumer tend to produce a too rational choice process. Thoughts related to final solutions are overreported and thoughts related to "discarded" choice procedures tend to be underreported. As has been verified in other research contexts, subjects form their own hypotheses about what is relevant for the researcher. This phenomenon is called "evaluation apprehension" (Rosenberg, 1969).

We will use behavioral measures, such as the sequence of information "bits" as acquired from an information display matrix (IDM) or the product package; eye-movement recording and direct observation of the choice process. An information display matrix is a rectangular display of product/brand alternatives and their attribute values, in most cases the alternatives represented as columns, and the attributes as rows (See Jacoby, 1975). An early observation is that consumers tend to use only a small fraction of the available information, reject alternatives on the basis of negative evidence (without knowing other probably compensating attribute values) but they are nevertheless satisfied and feel certain having selected the best possible alternative.


Consumer information processing is determined by a host of factors grouped below:

1. Prior knowledge, familiarity, and satisfaction with the product;

2. Interest or involvement in the choice;

3. Personality factors, including cognitive style;

4. Perceived financial, social, or personal risk;

5. Environmental or situational factors: Time pressure, distraction, crowding in the store. Wright (1974a) investigates the influence of time pressure and distraction;

6. Structure and format of the product information.

In this paper we concentrate on the influence of structure and format of the information on the choice process.

Consumers tend to adapt to the task environment. Adaptation means the choice of a behavioral alternative that is in accordance with one's needs and aspirations and is expected to lead to an optimum among the choice alternatives, time, effort, and budget. For simple choice tasks, a standard alternative (brand loyalty) may be the optimum choice. Wright (1974b) calls this affect referral. But for choice tasks involving large perceived risk, such as high price, the consumer will be more interested in getting the "best" alternative. This leads to more information collection and evaluation. In this case, the economic model of the consumer as a utility maximizer is more applicable, as represented by the linear-compensatory attitude and utility models. In choice tasks of intermediate risk and involvement, restricted information acquisition and processing are likely to occur. Simplifying heuristics and shortcuts in information processing function here in order to reduce the choice task to manageable proportions.

From research in cognitive psychology, we know that human information-processing and memory capacities are limited. Without extended processing and memory capacity (paper and pencil, calculator), the consumer has to simplify the choice task. We do not expect that consumers spontaneously engage in very much data transformation or rearrangement in reviewing product or brand information for a choice (Wright, 1974b). The information structure and format will influence the choice strategy.

In a recent paper, Dawes (1975) argues that cognitive psychologists developed models of the subjects' information-processing behavior that are primarily models of the task structure. These models represent the subjects' behavior only because the subjects have acquired abilities that allow them to behave appropriately (not necessarily optimally) in the task situation. Understanding the task requirements yields an understanding of the subject who performs more or less adequately$ Instead of classifying all kinds of consumer characteristics (personality, attitude), we have to direct our research attention to the conditions of information structure, format, and the environmental factors determining consumer choice behavior. Individual differences will still be observed for the same information-presentation conditions but significant differences may be expected for the "same" brand/alternative information under different modes of information presentation.

From a consumer viewpoint, it is reasonable to request those product information structures and formats that facilitate the choice task and diminish the chance of a suboptimal choice. This does not mean that certain brands (retailer vs. national brands) will always be preferred or that the consumer is induced to select certain brands. But is means that information acquisition, evaluation, and processing are done more easily and that the outcome of the choice process will be more satisfactory to the consumer.


The basic inputs for the choice process are not the attribute values as perceived by the consumer or scored on a favorableness scale but are the objective attribute values such as in informative labeling. However, individual differences may still exist in the internal representation of these objective attribute values.


The first component of the way in which the information is presented is the information structure mode. This structure does not tell what kind of information or which format the information is presented in, but how the information is presented. We distinguish four basic information structures:

A. Pairs of alternatives with two or more attributes, and a paired comparison. The attribute values are organized per alternative.


B. Structure by alternative: Three or more alternatives with two or more attributes are presented. The attribute values are organized per alternative. Information structuring by alternative is the most common form of information structure. The attribute values of an alternative are presented together and the alternatives are separated in space or time: packages on the shelf containing the product information, advertisements (except comparative advertising), and consumer information labels attached to the prod-


C. Structure by matrix of alternatives and attributes with three or more alternatives and two or more attributes. An example is the information display matrix (IDM).


Information structuring by matrix is found in unit price lists, product quality ratings, and consumer association comparative product test tables. The in- formation can be read by the column or by the row, processing all attribute values of one alternative, all alternatives on one attribute, or a combination of both.

D. Sequential availability of the alternatives with the corresponding attribute values.


If n choice alternatives exist, a constantly changing set of r choice alternatives is present (r<< n). In many consumer choice situations, the product information comes sequentially. 0lander (1975) mentions the prospective shoe buyer who has to decide to accept one of the available offers in the store or to go to another store. A house buyer meets the same situation; in his case, the vacant house may even been sold to another buyer. It is cumbersome, or even impossible, to compare all alternatives, but nevertheless a choice has to be made.


The second component of the way in which the information is presented is the information format. In most attitude and utility research, the information format is unspecified, assuming that all alternatives have the same set of relevant or salient attributes. In experimental research, some specific information formats are studied in order to assess their influence on information-processing behavior and the outcome of the choice task. These formats are as follows:

1. Unique vs. common attributes: Some attributes are common for all alternatives and some attributes are unique for a subset of alternatives.

2. Major vs. minor attributes: For alternatives equated in total value, a slightly higher score on a major attribute may have more influence on the choice process than a much superior score on a minor attribute.

3. Variability of the attribute values: Alternatives may be the same "on the average" over the relevant attributes but may differ in the range of attribute values.

4. Absolute vs. relative attribute values: Attribute values may be provided as absolute numbers or in a relative (comparative) format.

5. With or without brand name: Some laboratory experiments are conducted comparing task conditions with and without the brand name available.

6. Unspecified format: None of the above mentioned specific information formats may be present in the choice task.


The four information structure modes and the six information format categories are combined in Table 1. From this table we learn that experimental research is done for only 12 of the 24 possible structure X format combinations. The cells in Table 1 where a code is entered indicate an information structure X format combination investigated in experimental research. We shall review the 12 indicated information structure X format combinations.




Before reviewing specific studies, it is useful to look at information-processing behavior in more detail. The consumer is confronted with the product information in a certain structure (A, B, C, or D) and format (1, ..., 6). The new information must be brought into agreement with prior information, attitudes, and opinions and is transformed to an easier structure and format for subsequent processing. This is called the "information restructuring phase," resulting in the internal representation of the information. The next phase is the processing of the restructured information, which under some task conditions is done in two steps: the "computation'' of a summary measure for each alternative and the comparison of the summary measures. The outcome of the choice process is either to choose one of the available alternatives or to postpone the choice. In most experiments, the subject is forced to make a choice.

Figure 1 on next page shows that several phases precede the actual processing phase. In each of these earlier phases, the information is transformed. These transformation are already a form of information processing, adapting and distorting the objective information.


Slovic and MacPhillamy (1974) investigated the attribute (cue) utilization in comparative judgment. Subjects compared pairs of students with respect to potential college GPA. Both students in each pair had scores on one common attribute (English skills) and one unique attribute (quantitative aptitude or need achievement). The experimental subjects tended to give a higher weight to the common attribute than to the unique attribute. Even cautioning the subjects not to increase the weight of the common attribute did not reduce the effect, nor did correct answer feedback with a reward for accuracy. This lack of effect of cautioning is not very encouraging for consumer educators.

An explanation for this effect is that the use of the common attribute provides a direct and unambiguous means of comparison between the two alternatives. To use unique information, one must deal with questions of relative attribute importance or "trade offs" between the unique attributes. The common dimension effect is a bias in human information processing. It indicates that unique product attributes (features, extras) may have less influence in the consumer choice process than the common attributes. Marketing experience, however, says that a new and innovative product attribute may attract buyers. It may lead to the rearrangement of beliefs and evaluations and to a shift in consumer preference. Does this experience conflict with the outcome of Slovic and MacPhillamy's experiment? They did not use a new or binary (absent vs. present) unique attribute but only an attribute on which the value of other alternatives was not provided. It is premature to generalize that a co-,non attribute will be used more in other information structure modes (B, C, and D). In Slovic and MacPhillamy's experiment, the common attribute provided an easy way to choose one alternative from a pair. If the scores on the common attribute differentiate less, or if combinatory indexes of the common attributes do not differentiate between the alternatives the unique attribute will turn the scale.

Slovic and MacPhillamy's experiment presents a case of incomplete information. Some attribute values were not known, although it is reasonable to assume that the alternative had some value on that attribute: Each student has some degree of quantitative aptitude or need achievement. An obvious interpolation for these missing values is the average value, computed from the available attribute values in the choice set (information structure modes B, C, or D). The attribute value variability decreases; the attribute becomes less discriminating among the alternatives; and is used less in the choice process. Therefore, a unique gets a lower weight.

In the marketing example, a new (innovative) attribute is introduced, and it is clear to the consumer that the competing brands do not possess this new attribute. The other products have a zero value on this new attribute, increasing the attribute variability. The new attribute becomes more discriminating among the alternatives and is more influential in the choice process. However, the other products may also possess new and unique attributes. A process of comparing unique attributes will be more difficult than comparing common attributes, and it may be expected that the common attributes are more influential in the choice process.

Jacoby, Speller, and Kohn (1974a) and Jacoby, Speller, and Kohn Berning (1974b) employed information structure modes B and C in their experiments (cells B1 and C1 of Table 1). In one condition, half the subjects received values from the same attributes in the same order for each brand, while the other subjects received the identical information presented in a different (scrambled) order for each brand. In the former condition, the attributes are clearly perceived as common attributes, while in the latter condition, some attributes may have been perceived as being unique. However, the hypothesis that subjects tend to give a higher weight to the common attributes is not tested in these experiments.




A greater variability of attribute values facilitates the differentiation of the choice alternatives on that attribute. Subjects are motivated to differential judgments, and a greater variability of attribute values increases the possibility of differentiating. We may expect that an attribute with a larger variance in attribute values over a choice set is used more in the choice process.

Related to attribute variability is the experimental comparison of small and large attribute variability, while keeping the average value of the attributes equal. The averaging model from information integration theory predicts that the judgment of a set of alternatives containing extremely negative and extremely positive attribute values averages with the judgment of a set of alternatives having moderately negative and positive attribute values.

In one of our experiments (Van Raaij, 1977), two conditions were created for a set of alternatives (camping sites): extreme and moderate attribute values, conditions EA and MA, respectively. Fifty subjects processed the camping site information which was presented in an information display matrix (IDM), information structure mode C, by turning over information cards containing the attribute value of an alternative (See, Jacoby, 1975). Results of this experiment pertain not only to the proportion of information acquired but also to the information-processing patterns. The processing pattern can be derived from the sequence of information cards of the IDM, as seen by the subject. In the EA condition (large attribute variability), more choice by alternative occurred, especially in the second half of the choice process. In the MA condition (small attribute variability), more choice by attribute occurred, especially in the first half of the choice process.

Extremely high and low attribute values cause the consumer to consider the alternatives in depth in order to avoid a low and unacceptable attribute value (conjunctive rule) or to obtain an alternative with an outstanding attribute value (disjunctive rule). This kind of choice behavior increases the choice accuracy (according to a normatively "best choice" rule), resulting in more choice satisfaction and certainty. Moderately high or low attribute values elicit satisfying behavior, since the differences between the alternatives are less dramatic, but choice satisfaction and certainty are lower (See Van Raaij, 1977).

Another explanation is that in the EA condition the consumer "calculates" a summary index for each alternative and compares these summary indexes of the alternatives. The "calculation" of the summary indexes is done in the short-term memory, and the summary indexes are stored in the long-term memory, from which they can be retrieved into the operational memory. The summary indexes are an intermediate step in the choice process in order to facilitate information handling.

Table 2 shows the proportions of processing by alternative (characterized by type 2 transitions in the IDM), or by attribute (characterized by type 3 transitions in the IDM). Four transition types are distinguished in the consumer information processing sequence:

type 1 transition: same alternative/same attribute (the same information is repeated).

type 2 transition: same alternative/different attribute.

type 3 transition: different alternative/same attribute.

type 4 transition: different alternative/different attribute.

The theoretical expectations of the proportions of type 2 and 3 transitions in a 10 X 6 matrix are, respectively: 5/60=.08 and 9/60=.15 for a random sequence of information processing. The transition-type analysis has been developed by Jacoby, Chestnut, Weigl, and Fisher (1976).




Slovic (1975) asked subjects to choose an alternative from a pair of alternatives that they had previously equated in value. Within each pair, one alternative was superior on a major attribute but so inferior on a minor attribute that this disadvantage cancelled the advantage on the major attribute. As choice alternatives were used baseball players and baseball teams. Most subjects solved this conflict of choosing among equally valued alternatives by consistently selecting the alternative that was superior on the major attribute, disregarding the inferiority on the minor attribute. Major and minor attributes were the more or less important attributes for each subject individually.

Slovic used not only baseball players and teams but in later experiments also choice alternatives such as TV commercials, auto tires, gift packages, college and secretarial applicants, routes to work, and amounts of cigarettes. "Reliance on easily justifiable aspects to the neglect of other important factors could lead one to reject alternatives whose overall utilities are superior to those of the chosen alternative" (Slovic, 1975). The choice simplification heuristic of relying on the major attribute may lead to nonoptimal choice of the consumer. A presidential candidate C who is slightly better on a major issue will be preferred over another presidential candidate F who is superior on a minor issue.

Tversky's (1972) elimination-by-aspects model and the lexicographic model both have the characteristic that the attributes are used sequentially in their order of importance. These heuristics may lead to a quick, but nonoptimal, choice. Wright (1974b) advocates the propagation of this choice heuristic by advertisers, using Tversky's example of a commercial advertisement for a computer training course. From a consumer viewpoint, however, it is better to educate the consumer to avoid this choice strategy, which may lead to the selection of an inferior product.

Slovic (1975) used information structure mode A, and it is worthwhile to investigate whether consumers, who operate mainly under the other information structure modes (B, C, and D), use the same simplifying heuristic of overestimating the importance of a major attribute.


Important in consumer decision making and especially in informative labeling of products is the absolute or relative format of the brand/variant information. The hypothesis is that a relative format facilitates the use of the information by the consumer. An indication of the absolute quantity of nicotine or tar in cigarettes requires the consumer's prior knowledge of what are high or low quantities. Relative information does not require this cognitive transformation. Another example is unit pricing: Unit prices are easier to compare than traditional prices. Gatewood and Perloff (1972) found that the additional information of price per ounce of net weight produced a significant increase in the accuracy of choices over conditions without unit prices or with a computational device to aid in price calculations. Also, time to make a choice was reduced. Russo (1975a) investigated the posting of shelf tags with unit prices in the supermarket. The unit price information decreased consumer expenditure by 1%. This is the information structure mode B (by alternative). When unit prices were displayed on an organized list, consumer savings increased to 3%. This is the information structure mode C (by matrix). The matrix structure mode C4 caused a 5% increase in market share of the cheaper (store) brands. Russo concluded that "the financial benefits to both consumers and retailers justify the posting of unit price information on a widespread basis." Russo (1975b) also investigated consumer information integration problems and the optimal amount of information and its display format.

Gatewood and Perloff (1972) and Russo (1975a) did not study information-processing behavior, only the outcomes of information-processing behavior, which they investigated in a supermarket setting.

A similar comparison is made by Chestnut (1976). He compared verbal and numerical information experimentally in the information-processing behavior of the choice among brands and variants of 75-watt light bulbs. The numerical information format proved to be easier in the information-processing task itself, but the verbal information was more effective in the long run. Verbal information needed a first transformation to become comparable across alternatives. This extended processing (stimulus elaboration) caused a better long-term memory of the information (16 minute period). Chestnut presented the product information structured by alternative or by attribute (B4,C4). The verbal information format resulted in extended processing and longer retention. On the other hand, a numerical information format facilitated information processing and resulted in a shorter decision time. But this facilitation was restricted to the short-term memory; long-term memory retention prospered by extended information processing (stimulus elaboration). Thus, the facilitation of consumer information processing by a numerical format has an adverse effect on information retention.


Several experiments have been conducted comparing information-processing behavior in experimental conditions with and without brand name available. In two experiments, the information was structured by alternative (B5), and the subjects' processing behavior was recorded by direct observation or by eye-movement recording (Van Raaij, 1977). In another experiment the information was structured by matrix (C5) in an IDM with information cards (Van Raaij, 1977). The product class in these three experiments was ground coffee, and 13 brands with four attributes with or without brand name were presented. It was found that the brand name is used as an "information chunk" for identifying the packages and facilitating the remembering of the brand's attribute values. Just as in the condition with large attribute variability (C3), the consumer tends to form a summary index for each brand and to compare indexes. In a condition with the brand name available, less information is acquired on the alternatives, but the choice accuracy is higher. The acceptance or choice of an alternative is based on positive evidence, when the brand name is available. With the brand name, the "bundle of attributes" becomes a recognizable and known object, both in the positive (acceptance) sense and in the negative (rejection) sense.

Jacoby, Szybillo, and Busato-Schach (1976) conducted a similar experiment in 1973, using psychology undergraduates as subjects selecting a "brand" of toothpaste. They created two experimental conditions: One with and one without brand and manufacturer's name. Brand and manufacturer's name were considered as two of the product attributes. They found that subjects tend to place substantial behavioral importance on price and particularly on brand name information. "Not only were consumers more satisfied with their purchase decision when brand name information was available, but they also tended to select fewer information dimensions (attributes) under this condition, thereby suggesting that brand name does indeed serve an information chunking function in consumer decision making." This experiment has been replicated in Western Germany (Mannheim) in 1975 (RaffTe, Hefner, Sch÷ler, Grabicke, and Jacoby, 1976). In the German study, other attributes proved to be most important, such as consumer association test rating (Stiftung Warentest), and special active ingredients. Price came in the third place and brand name in the eighth place. In the "brand name available" condition, more information is acquired, a result contrary to the American study. German consumers tended to acquire more information, while fewer attributes were presented (14 and 12) than in the American study (18 and 16). Several explanations for these observed differences are brought out by Raffee, Hefner, Sch÷ler, Grabicke, and Jacoby (1976) and Jacoby, Hefner, Sch÷ler, Grabicke, and Raffee (1976). Cross-cultural differences between German and American consumers, experimental factors, or a greater influence of the German Stiftung Warentest compared with the American Consumers' Union may be the cause. It suggests that intensified dissemination of consumer test reports may lead to increased utilization of this material by consumers.


In many consumer choice situations, the product information will come sequentially. +lander (1975) investigated a choice process in which a part of the choice alternatives is not accessible simultaneously and there was a cost involved (money, time, effort) in getting access to these alternatives. The hypothesis was tested that in a nonsimultaneous choice situation, satisfying behavior is an adequate and used consumer heuristic. Satisfying behavior means that the first alternative is chosen that meets certain requirements (minimum attribute values or a combination of attribute values above a certain threshold). The threshold may exist before the alternatives are seen and is adapted dependent on the observed values in the sequential processing (anchoring of a threshold and adaptation based on new information). A maximizing strategy (searching the maximum combination of attribute values) is virtually impossible, since the set of choice alternatives changes continually.


Early research by Jacoby, Speller, and Kohn (1974a) and Jacoby, Speller, and Kohn Berning (1974b) on the effect of information load on the consumer information processing patterns is done with an IDM information structure mode (C). The effects of information load (size of the IDM) or task complexity will not be reviewed in this paper.


It is premature to generalize the effects of information structure and format on consumer information-processing behavior based on the 12 task conditions reviewed above. We can, however, make some general conclusions:

1. Consumers tend to utilize simplification heuristics in situations of information overload, primarily in the information structure modes B, C, and D. Examples of these heuristics are: Summary indexes per alternative, limited information acquisition, and satisfying behavior.

2. Consumers tend to bias information processing by relying on common attributes or attributes with a larger variability, or by overestimating major attributes in a paired comparison choice task (A). The biases are generally in conflict with an optimal choice from the consumer viewpoint.

3. Simplifying the information display by IDM, or a relative information format, leads to easier processing but to less retention of the information (learning) in the long run.

4. Not simplifying the information display into an information structure by matrix (C), or a relative information format leads to a less accurate choice, since less information is acquired and used. But the required information is remembered better as a consequence of the information transformation task (stimulus elaboration).

The results of the experiments in the information structure modes B, C, and D may be partly attributed to the information overload in these conditions, resulting in less complete information acquisition and less optimal processing behavior. The results of Paul Slovic's experiments in the information structure mode A (pairs) cannot be explained as a consequence of information overload but are a pure effect of the information format. The simplifying heuristics found here show the limited "rationality" of the human mind, even in dealing with "simple" nonoverloaded paired choice tasks.

Under conditions of information overload, we expect a higher frequency of the utilization of simplifying heuristics in order to cope with the task complexity. 3acob Jacoby employs the information structure C (by matrix) in his experiments on consumer information processing. He does not specify the information format (C6), although more recent studies by his co-workers include a specific information format (Chestnut, 1976). The results of the information overload experiments by Jacoby are

(1) limited utilization of the available information, even corrected for the fact that the exposed brand name conveys some information, and

(2) a high degree of ignorant satisfaction with a (suboptimal) choice, and choice certainty.


In the remainder of this paper, we compare two approaches in the study of consumer decision processes: Eye-movement recording and the IDM approach in the comparison and choice among 13 brands of ground coffee with four attributes (price, quality, caffeine content, usage instructions). The experimental design contains three price levels, two quality levels, two levels of caffeine content, and usage instructions present or not. The overall "value" or "utility" of the 13 brands is nearly equal for the subjects. The 13 brands are all available in The Netherlands, except one Belgian brand. The packages contain 250 gram ground coffee.

Twenty housewives participated in the eye-movement study in December, 1975, and in the IDM study in April,1976. The subjects are recruited from a residential area in Tilburg-West (The Netherlands), and were rewarded by their preferred brand of coffee and an additional amount of money up to F 6,00 ($2.25). In total, they participated in four experimental conditions.

In both experiments, a condition with brand name available (BC) and a condition without brand name available (UC) were created. In the eye-movement experiment, the BC condition had real product packages containing the four attribute values, and the UC condition had coffee packaged in brown boxes not containing a brand name. The information in the IDM experiment is provided on information cards in a matrix display. The columns of the matrix correspond to the alternatives. In the BC condition, the brand name is provided, and in the UC condition, a letter code is used instead of the brand name. Note, that the eye-movement study employs an information structure mode B (by alternative) and that the IDM study employs an information structure mode C (by matrix).


The amount of total information in both experiments is 13 X 4 = 52 attribute values. Some information usage variables are presented in Table 3 for the two experiments. We see that in the eye-movement experiment more information is collected than in the IDM experiment. All information usage variables are higher in the eye-movement study: The average number of information pieces used, the average number of alternatives considered, the average number of information pieces of the most important attribute. Over 50% of the information pieces is seen twice or more in the eye-movement study, while this percentage is 0 in the IDM study.

Information structure by alternative (B) facilitates the acquisition of information as available on the real product packages. The subjects have to scan more information before they find the searched information. Much information is acquired accidentally. In the IDM experiments, a more goal-directed information search is possible, since the information is prestructured in a matrix design.

We also observe that in the UC conditions (without the brand name) more information is collected than in the BC condition (with brand name). The brand name in the BC conditions functions as an information chunk and an "identity mark" for the alternatives, so that less information needs to be acquired more than once. The information is more easily remembered in the BC condition and less redundant information is collected by the consumers.





Three-way analyses of variance were computed with a two-level fixed factor IS, information structure mode with levels B and C; and a two-level fixed factor BN, brand name with levels UC and BC. The factor "subjects" is crossed with the other factors. The main effect of IS is tested with MS(IS X subjects within group); the main effect of BN is tested with MS(BN X subjects within group); and the interaction IS X BN is tested with MS(IS X BN X subjects within group). All F ratios have 1 degree of freedom for the numerator and 19 degrees of freedom for the denumerator.

Four analyses of variance were performed and in Table 4 the F ratios are presented. All main effects are significant, seven of the eight on a .01-level (***). The main effect IS stands for both the difference in information structure mode (B or C) and the time interval between December, 1975, and April, 1976. Differences between information structure mode explain about 25% of the total variance, while differences between the UC and BC conditions explain not more than 5% of the total variance, as measured by w2 measures.


Table 5 provides the proportions of type 2 and type 3 transitions in the two experiments. Type 2 transitions dominate, even in the IDM experiment where the information is structured in a matrix, facilitating information processing by alternative as well as by attribute. Note that the theoretical expectation of type 2 transitions is only 3/51=.06, assuming a random processing pattern. The theoretical expectation of type 3 proportions is 12/51=.23. The observed proportions are not significantly different from the expected value for type 3.



From Table 5 we see that type 2 transitions occur more in the second half of the choice process, and type 3 transitions relatively more in the first half of the choice process. In the first half of the choice process, relatively more comparison of alternatives on one or more attributes (type 3) takes place, especially in the IDM design. The alternatives are evaluated in more detail in the second half of the choice process (type 2). But, processing by alternative is still more common procedure in all conditions, even in the IDM experiment where processing by attribute is facilitated.

Differences between matched pairs of proportions are tested. In the eye-movement recording experiment no significant differences are obtained. Between the UC and BC conditions in the IDM experiment, the differences between type 2, type 3, first-half type 2, and second-half type 2 are significant at p <.10. Between the first and the second half of the choice process in the IDM experiment, the differences between type 2 in the UC condition, type 2 in the BC condition, and type 3 in the UC condition are significant at p <.10.

Condition UC in the IDM experiment represents, in fact, the most abstract experimental condition: Information structure by matrix without the brand name. Relatively more processing by attribute (type 3) and less by alternative (type 2) takes place. Only in this abstract choice task is the emphasis not predominantly on the alternatives but on a direct comparison of attribute values. This effect is stronger in another but similar experiment, described in Van Raaij (1977).


In both experiments, the subjects rated the importance of the four product (coffee) attributes on a 5-point scale (very important - very unimportant). This is a verbal, attitudinal measure of attribute importance. From the actual information-processing behavior, a measure of attribute usage can be computed, i.e., the proportion of information pieces used belonging to a certain attribute. In Figure 2, the mean rank-order correlations between attribute usage and rated attribute importance are provided. The mean is the average rank-order correlation over the 20 subjects. B and C represent the attribute importance ratings for the experimental designs B and C, the eye-movement study and the IDM study, respectively. UC and BC represent the conditions without and with brand name, respectively.



The higher mean rank correlations are found between the B(UC) and B(BC), C(UC) and C(BC), and B and C. The correlations r(C(UC),C(BC))=.86 and r(B(UC),B(BC))=.54 indicate a high correspondence in attribute usage between the two conditions of an experiment UC and BC. r(B,C)= .59 indicates a high correspondence between attribute importance ratings in December, 1975, and April, 1976, a stability measure of the verbal measure. The other rank correlations are lower. These are correlations between a processing measure and a rating scale measure of attribute importance. These correlations are low, indicating a low correspondence between what subjects say and what subjects do. With a rating scale we generally measure the normative importance of an attribute and not the use of an attribute in the choice process. "Quality" is normatively rated as very important in the choice process but "caffeine content" is used mainly in order to differentiate among the products. The use of a product attribute in the choice task depends also upon the variability of the attribute values on that attribute. The higher the variability, the more differentiation is possible with help of that attribute. From these results it can be concluded that the correlations between the behavioral measures or between the verbal (rating scale) measures, are generally higher than the correlations between a behavioral and a verbal measure.


Research on the influence of information structure and format, and information load on consumer information processing and choice started a few years ago. Concepts and methods from cognitive psychology are adopted and applied in the multi-attribute choice situation. This development is not only healthy for our knowledge on consumer choice behavior but also for a more descriptive approach in the study of human decision making.

Consumer research may contribute to the design of the structure and format of informative product labeling, the layout of comparative testing tables, and the design of store displays. The objective is the ameliorization of the information structure and format in order to facilitate information acquisition, evaluation, and choice. For consumer education, it is a starting point for a program to educate the consumer to avoid suboptimal choice. The application of consumer information processing research to advertising, as proposed by Wright (1974b), has to be avoided, because it reinforces suboptimal choice and simplifying heuristics as employed by the consumer.

Behavioral measures may prove to be superior to scale ratings in assessing attribute importance. The relationship between attribute variability and attribute usage is not determined, however.


Robert W. Chestnut, "The Impact of Energy-Efficiency Ratings: Selective vs. Elaborative Encoding," Purdue Papers in Consumer Psychology, No. 160, 1976.

Robyn M. Dawes, "The Mind, the Model, and the Task," in H.L. Castellon and F. Restle (Eds.), Proceedings of the Seventh Annual Indiana Theoretical and Cognitive Psychology Conference, 1975. 119-129.

Robert D. Gatewood and Robert Perloff, "An Experimental Investigation of three Methods of Providing Weight and Price Information to Consumers," Journal of Applied Psychology, 57(February, 1973), 81-85.

Jacob Jacoby, "Perspectives on a Consumer Information Processing Research Program," Communication Research, 2(July, 1975), 203-215.

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

Jacob Jacoby, Donald E. Speller, and Carol A. Kohn, "Brand Choice Behavior as a Function of Information Load," Journal of Marketing Research, 11(February, 1974a), 63-69.

Jacob Jacoby, Donald E. Speller, and Carol Kohn Berning, "Brand Choice Behavior as a Function of Information Load: Replication and Extension," Journal of Consumer Research, 1(June, 1974b), 33-42.

Jacob Jacoby, George J. Szybillo, and Jacqueline Busato-Schach, "Information Acquisition Behavior in Brand Choice Situations," Journal of Consumer Research, 1976. in press.

Jacob Jacoby, Margarete Hefner, Manfred Sch÷ler, Klaus Grabicke, and Hans Raffle, "Information Acquisition Behavior in Brand Choice Situations: A Cross-cultural Extension," Purdue Papers in Consumer Psychology, No. 162, 1976.

Folke +lander, "Search Behavior in Nonsimultaneous Choice Situations: Satisficing or Maximizing?" in D. Wendt and C. Vlek (Eds.), Utility, Probability, and Human Decision Making, (Dordrecht, The Netherlands: D. Reidel Publishing Company, 1975), 297-320.

Hans Raffle, Margarete Hefner, Manfred Sch÷ler, Klaus Grabicke, und Jacob Jacoby, "Das Informationsbeschaffungs-verhalten bei der Markenwahl - Eine Experimentelle Untersuchung," UniversitSt Mannheim, Sonderforschungs-bereich 24, 1976.

Milton J. Rosenberg, "The Conditions and Consequences of Evaluation Apprehension," in R. Rosenthal and R.L. Rosnow (Eds.), Artifact in Behavioral Research, (New York: Academic Press, 1969), 279-349.

J. Edward Russo, "The Value of Unit Price Information," (a), "An Information-Processing Analysis of Point-of-Purchase Decisions," (b), Working Paper CHIP, No. 51, 1975. Center for Human Information Processing, University of California at San Diego.

J. Edward Russo, Gene Krieser, and Sally Miyashita, "An Effective Display of Unit Price Information," Journal of Marketing, 39(April, 1975), 11-19.

Paul Slovic, "Prom Shakespeare to Simon: Speculations -and some Evidence - about Man's Ability to Process Information," ORI Research Bulletin, 12(2, 1972).

Paul Slovic, "Choice between Equally Valued Alternatives," Journal of Experimental Psychology: Human Perception and Performance, 1 (July, 1975), 280-287.

Paul Slovic, Baruch Fischhoff, and Sarah Lichtenstein, "Behavioral Decision Theory," Annual Review of Psychology, 28(1977), in press.

Paul Slovic and Sarah Lichtenstein, "Comparison of Bayesian and Regression Approaches to the Study of Information Processing in Judgment," Organizational Behavior and Human Performance, 6(November, 1971), 649-744.

Paul Slovic and Douglas MacPhillamy, "Dimensional Commensurability and Cue Utilization in Comparative Judgment,'' Organizational Behavior and Human Performance, 11(April, 1974), 172-194.

Amos Tversky, "Elimination by Aspects: A Theory of Choice," Psychological Review, 79(July, 1972), 281-299.

W. Fred van Raaij, Consumer Choice Behavior: An Information-Processing Approach, (voorschoten, The Netherlands: VAM, 1977).

Peter L. Wright, "Research Orientations for Analyzing Consumer Judgment Processes," in S. Ward and P. Wright (Eds.), Advances in Consumer Research, Vol. 1, 1973, 268-279.

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

Peter L. Wright, "An Adaptive Consumer's View of Attitudes and other Choice Mechanisms, as viewed by an Equally Adaptive Advertiser," Research Paper, No. 219, August, 1974b. Graduate School of Business, Stanford University.



W. Fred van Raaij, University of Illinois


NA - Advances in Consumer Research Volume 04 | 1977

Share Proceeding

Featured papers

See More


The Trusted Influencer: How They Do It and How Brands Can Benefit

Gillian Brooks, Oxford University, UK
Mikolaj Piskorski, IMD

Read More


Too Much of a Good Thing? Consumer Response to Changes in Brand Essence

Tarje Gaustad, Kristiania University College
Bendik Samuelsen, BI Norwegian Business School
Luk Warlop, Norwegian School of Management, Norway
Gavan Fitzsimons, Duke University, USA

Read More


A Model of Consumer Self-Regulation Failure

Keith Wilcox, Columbia University, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.