Patterns of Processing in Consumer Information Acquisition

ABSTRACT - Patterns of information acquisition from a brands x information dimension matrix were studied for sixty subjects. Three major patterns were found for those subjects seeking information: processing by attributes, by brands, and a hybrid form, feedback processing. Classification procedures for these patterns were developed. Relating these patterns to other variables was explored.


James R. Bettman and Jacob Jacoby (1976) ,"Patterns of Processing in Consumer Information Acquisition", in NA - Advances in Consumer Research Volume 03, eds. Beverlee B. Anderson, Cincinnati, OH : Association for Consumer Research, Pages: 315-320.

Advances in Consumer Research Volume 3, 1976      Pages 315-320


James R. Bettman, University of California, Los Angeles

Jacob Jacoby, Purdue University

[Preparation of this report was facilitated, in part, by a grant from the National Science Foundation (GS-43687) to the junior author. A more complete working paper version of this article is available from either author.]


Patterns of information acquisition from a brands x information dimension matrix were studied for sixty subjects. Three major patterns were found for those subjects seeking information: processing by attributes, by brands, and a hybrid form, feedback processing. Classification procedures for these patterns were developed. Relating these patterns to other variables was explored.

Despite recent interest in consumer information processing (Hughes and Ray, 1974), perceptual and memory phenomena have not been handled well in most models. These include: perceptual encoding, information acquisition, storage, and retrieval. These areas have been virtually ignored in consumer information processing research. For example, consumer decision net models (Bettman, 1970) use an extremely simplistic memory structure. Memory is modeled as a fixed, given data array. Encoding is ignored, as is acquisition. The data are there to be utilized, represented as elements in a matrix. Retrieval is assumed to be automatic and error-free. As Calder (1975) points out, how information is stored in memory can effect the type of processing rule used. Even if the rules used in making choices are of primary concern, perceptual and memory issues must be examined. This paper focuses on patterns of information acquisition, only briefly considering storage and retrieval.

Study of acquisition patterns is necessary for a theory of consumer information processing. It also has pragmatic value for study of public policy issues related to provision of information to consumers. Tasks such as reading labels on packages in a supermarket or looking at a table in a Consumer Reports article are tasks of information acquisition. As Russo, Krieser, and Miyashita (1975) show, how information is presented in such a task can greatly affect whether the information is used. Knowing how consumers acquire information from data displays can aid in developing more effective displays (Russo, 1975).

Some recent research in both cognitive psychology and consumer decision making has begun to attack issues of information acquisition (Berning and Jacoby, 1974; Jacoby, Szybillo, and Busato-Schach, 1974; Lussier and Olshavsky, 1974; Payne, 1976; Russo and Rosen, 1975; Simon and Barenfeld, 1969; Svenson, 1974)

A common focus is on the order and patterns of information acquisition during the choice process. Some stable findings emerge, although often presented without formal analysis. First, there are individual differences in order of information search and acquisition. Some subjects proceed by examining one brand at a time. That is, they choose a brand and gather information on several attributes of that brand. Then they choose a second brand and gather information on several attributes (not necessarily the same as those for the first brand) and so on. This strategy may be called Choice by Processing Brands (CPB). [The terms 'Choice' and 'Processing' are used, although acquisition is the primary focus, because subjects are asked to choose an alternative after their information search. Hence, acquisition and choice processes are confounded. This is discussed further below.] A second group acquires information by choosing an attribute, determining values for each of several brands on that attribute, choosing a second attribute and determining values for several brands and so on. This may be called a Choice by Processing Attributes (CPA) strategy. Finally, some subjects engage in hybrid strategies. These strategies seem to be adaptive, governed by immediate feedback from the information found during the search. Alternating short sequences of brand and attribute processing are often observed (e.g., a subject may process brands 2, 3, and 4 for attribute 1, become interested in brand 4 and next process attributes 2 and 5 for it, then examine attribute 5 for brands 2 and 3, and so on.) This may be termed a Choice by Feedback Processing (CFP) strategy.

A second set of findings is that some subjects use the basic strategies above in a uniform manner; they do not change their strategy during the course of the search. Other subjects use phased strategies (Wright, 1974). where the type of processing varies. A common type of phased strategy is one where the first phase is information input, and the second phase is a set of paired comparisons among specific alternatives (Svenson, 1974; Lussier and Olshavsky, 1974; Russo and Rosen, 1975; Payne, 1976).

This study examined information acquisition in a choice of breakfast cereal brands. Two methodological issues must be dealt with: first, the task must be designed so that the data on the sequence of acquisition are available; second, measures must be developed that allow classification of patterns of acquisition. The efforts described here represent advances over previous work.


The details of the procedures employed in this study are provided in Jacoby, Chestnut, Weigl, and Fisher (1976). Suffice it to say that it involves the presentation of information via a brand x information dimension matrix display and the sequential acquisition of information from this array by 60 subjects involved in a simulated breakfast cereal purchasing situation. By examining the pile of information cards acquired by each subject, a record of the information acquisition process as a sequence of brand-attribute pairs could be developed.

This task is essentially the same as that used by Payne (1976) in his apartment choice study. Payne also collected verbal protocols from his subjects. Russo and Rosen (1975) used eye movements, and Lussier and Olshavsky (1974) and Svenson (1974) used only verbal protocols. The present methodology yields a behavioral record of the complete external information search (internal retrieval from long term memory was, of course, not monitored), as does the eye movement data. Protocol data has value in potentially illuminating some aspects of internal search and rationale for the external search, although the record of the external search obtained may be incomplete. The combination of an information display board and protocols would obtain fairly complete information, but be cumbersome to use with many Subjects.

The main differences between the method of the present study and those used in past studies are the larger sample size (60 compared to 12 [Russo and Rosen, 1975]; 6 and 12 [Payne, 1976]; 6 [Svenson, 1974]; and 27 [Lussier and Olshavsky, 1974]); and the use of actual brands as choice objects and actual information values. The previous studies have used hypothetical alternatives.


The major datum available for classifying information acquisition patterns is the sequence of information cards (for each subject who took cards), which yields a sequence of brand-attribute pairs. The only previous attempt at developing an index to measure the structure of this input acquisition sequence was by Payne (1976), based upon examining adjacent pairs of brands and of attributes in the sequence. If n cards are chosen in the task, there are n-1 pairs of brands and of attributes. Each pair can consist of two identical brands (attributes) or two different ones. Let SB equal the number of identical brand pairs divided by n-l, and let SA be defined similarly for attributes. Then Payne's index is defined by (SB-SA)/(SB+SA). For a subject following a pure Processing Brands (CPB) strategy, this index would be +1. For a pure Processing Attributes (CPA) strategy, it would be -1. Payne also developed an index of shifts in the sequence. A shift occurs whenever neither a brand nor an attribute is common for two adjacent cards in the sequence. The proportion of shifts is approximately given by 1-SA-SB.

Payne's index for classifying subjects does not take into account that values of the SA and SB measures have different ranges for different numbers of cards chosen and total number of attributes and brands considered. With 10 cards chosen and 4 brands in total considered, the maximum value possible for SB is (10-4)/9 or .667, for example. In this study, normalization was performed by dividing SB and SA by the maximum possible for each subject, given the number of cards chosen and numbers of brands and attributes considered. These normalized measures are denoted by SBI (Same Brand Index) and SAI (Same Attribute Index). Both range from 0 to 1. This focuses on a structural criterion, rather than using amount or specific content of the information chosen.

These indices were used to determine the basic form of strategy used. In addition, however, the phased nature of the strategies was examined. The major phased strategy was where a paired comparison between brands occurred near the end of the sequence. The rule used to identify paired comparisons was that used by Russo and Rosen (1975) (their weak criterion): that brand alternations of the form X-Y-X appear in the sequence. The rule to determine whether a phased strategy (end comparison) was being used was whether or not a subject's sequence contained at least one paired comparison in the last six entries.

The classification procedure was exploratory. One of the authors (JB) examined each subject's sequence in detail. An attempt was then made to model the pattern classification rule used in making his judgments. For this particular author, this naturally resulted in a discrimination net, given in Figure 1. Although clustering algorithms based on the indices discussed above and other possible indices could clearly be applied, it was felt that at this stage of the research using more flexible human pattern recognition capabilities was desirable.

This classification procedure is based mainly upon the indices discussed above. First, subjects who chose no cards are classified as Choice by Brand Name (CBN). Second, subjects who used only one attribute or one brand are singled out as special cases of the CPA or CPB strategies, CPA-1 and CPB-1. Then, if the Same Brand Index (SBI) is high and the Same Attribute Index (SAI) is low, Choice by Processing Brands (CPB) is selected as the appropriate pattern. If SAI is high and SBI low, then Choice by Processing Attributes (CPA) is the classification. If both SAI and SBI are moderately high, Choice by Feedback Processing (CFP) is indicated. Note that for CFP to be as it was characterized above (alternating short same brand and attribute sequences), the number of shifts should not be a great deal higher than for CPA or CPB. Phased strategies with an end comparison (EC) are checked for in each of these instances.

This classification scheme is one of the only "formal" procedures presented in the literature. Most studies simply report what subjects appear to do without measuring processing by brands versus processing by attributes, described above. In his second experiment, Payne used his index to classify subjects, but he only used two groups, comparable to the attribute processing and brand processing groups of this study. This does not recognize feedback processing as a separate strategy, but confounds it with the other two. This is unfortunate, since the feedback processing group seems to have interesting and distinct properties, as shown below. The procedure used in the present study adds further precision in classifying subjects.


The results of applying the discrimination net to the data from the sixty subjects are as follows, using the abbreviations in the key for Figure 1: CBN, 12; CPB, 11; CPB-1, 4; CPB-EC, 3; CPA, 6; CPA-l, 5; CPA-EC, 11; CFP, 6; CFP-EC, 1; and Other, 1. There is relatively high use of Choice by Brand Name. Approximately equal numbers of CPB and CPA strategies were utilized, with more phased strategies for CPA users. Use of the CFP strategies is more limited. [The depiction of Choice by Feedback Processing utilized above was supported by the shift index. Values of the index were .154 for the 18 CPB subjects; .194 for the 22 CPA subjects, and .197 for the 7 CFP subjects (p < .66 in a one way analysis of variance). The average values of Payne's classification index for the three groups were .780 (CPB), -.772 (CPA), and .034 (CFP). The nearly zero value for the feedback processing group shows the confounding effects of Payne's classification scheme.] These classification results present a more formal view of individual differences in acquisition processing than past research, and provide insights into the empirical extent of different types of processing lacking in the previous studies. However, since only one product class was studied, the results may be dependent on this specific class.

In an attempt to characterize each strategy type more closely, three main types were examined for relationships with other variables. The various versions of CPA, CPB and CFP were collapsed together, to form groups of 18(CPB), 22(CPA), and 7(CFP). The analyses reported below were carried out for these 47 subjects. Choice by Brand Name subjects were not included in the analyses because they did not have most of the variables utilized available. Formal analyses of relationships between process type and other measures have not been reported in previous studies.



One way analyses of variance were performed for each of several variables using the three groups as factor levels. The mean values for each group and a-level probabilities for these ANOVAS on each variable follow: Number of Cards Chosen-CPB, 12.39; CPA, 13.09; CFP, 22.43 (p < .19); Amount of Time Taken (Seconds)-CPB, 201.89; CPA, 178.05; CFP, 313.86 (p < .15); Number of Brands Examined-CPB, 2.89; CPA, 4.86, CFP, 5.71 (p < .05); Number of Attributes Examined-CPB, 6.06; CPA, 3.64; CFP, 6.29 (p < .04); Perceived Quality Differences (1--very large, 5=hardly any)-CPB, 3.50; CPA, 3.41, CFP, 2.57 (p < .18). These results show that subjects using a Choice by Processing Brands strategy tend to examine fewer brands; those using a Choice by Processing Attributes strategy tend to use fewer attributes. These subjects tend to concentrate in depth on a few of the aspects they are using to organize their search (i.e., brands or attributes). Choice by Feedback Processing subjects may be using a more detailed acquisition process. They tend to use more time, take more cards, look at more brands and attributes, and see greater quality differences among brands in the product class. These findings are of course preliminary, since the data set is small, the significance levels only suggestive, and they are selected from a larger set of findings, but they provide hypotheses for future research. They are also supported by a contingency table relating processing strategy and a brand loyalty index (one if one brand bought more than 50% of the time, two otherwise). For CPB, 14 of 18 are brand loyal, for CPA 10 of 21, and for CFP, 1 of 7. This table yields a X2 significant at p < .02 (although expected cell sizes for the CFP subjects are small).

A consistent pattern seems to emerge: subjects using CFP strategies seem to engage in the most explicit processing, CPA subjects in an intermediate amount, and CPB subjects in a lesser amount. The CPB strategy does not explicitly include across brand comparisons (although internal processing may be high), the CPA strategy includes such comparisons for each attribute, and the CFP strategy seems to be a more highly external and evaluatively guided search process. To more fully understand the actual effort allocated in each case, however, evidence on internal as well as external processing would be necessary.


Properties of the Task

Some biases inherent in the task used must be discussed, since they may affect the results. First, since there is no limitation on the number of cards a subject can choose, there is no necessary incentive for the subject to "optimize" his search process, or organize his or her search. There is no explicit time pressure involved, and subjects may in fact be resigned to spending time 'in the lab'. Hence, search may be less directed than in a normal shopping environment. A more serious problem may be the availability of brand name to the subjects. This availability does coincide with the real world; however, most studies of cognitive strategies attempt to create situations where extra-experimental knowledge and differences in long-term memory functioning are not factors in performance. In this task, however, differential familiarity with brand names could affect the results. Also, "check up" strategies may be employed, where subjects may choose cards to verify information they already have stored in long-term memory (this may be true of actual store decisions also). These factors could bias the form of the process toward Choice by Brand Name or Choice by Processing Brands.

Another potentially biasing factor, mentioned above, is that the sequence of cards chosen confounds acquisition, evaluation, and choice processes, since subjects were required to choose a brand. However, this confounding seems to be inevitably characteristic of actual consumer decision making and experiments on choice, although Russo and Rosen (1975) present evidence that some of the effects can be unconfounded. In the present study, there may be a good deal of evaluative processing done internally, without ever being reflected in the card sequence explicitly. [This explanation is bolstered by the fact that very few cards chosen were repeats (7.6% of all cards chosen). This implies that evaluation processes relied on retrieval from memory, while acquisition was the main process underlying the observable card sequences.]

Finally, the experimental setup makes it equally easy, in terms of the effort needed to acquire the information, to process by brands or by attributes or to use a feedback strategy. However, this property is not true of the real world, except for perhaps tasks such as reading a table in Consumer Reports. For a choice of brands from a supermarket shelf, information is organized by brand more than by attribute. Thus perceptual encoding by brand is facilitated; this may in fact hinder evaluative brand comparison processes.

Implications for Consumer Information Processing Models

The results show that over half of the subjects process by attributes (CPA) or by a combination of brands and attributes (CFP). However, decision nets (Bettman, 1970) typically assume processing by brands (CPB). An alternative model, the elimination by aspects model by Tversky (1972), suggests that processing by attributes is used. The findings of this study suggest individual differences, but support the Tversky position. The discussion of acquisition effort above makes this conclusion more tentative. The task of this study would be more conducive to attribute processing than many real world brand choice situations. In actual brand choices, more processing by brands strategies may be used, because of the way supermarket shelves are organized. These real world information displays add to the difficulty of the choice task.

The distinction between brand and attribute processing may be affected by more than the task environment, Payne (1976) distinguishes between two possible ways of representing an attribute-brand combination in a list structure in memory: brand (attribute, value) or attribute (brand, value). The first representation, where the "tag" for the information is the brand, would facilitate brand processing. The second would make attribute processing easier. This issue points to a problem in coordinating the phases of problem solving in brand choice. Perceptual encoding and memory storage may be more efficient if done by brand, given typical information displays. However, as discussed below, brand comparisons may be much more effective if done by attribute. In fact there is some evidence that coding by object is more effective than encoding by attribute for accuracy of encoding (Haber, 1964; Lappin, 1967). See Kahneman (1973, especially Chapter 6) for further discussion.

Finally, the large proportion of Choice by Brand Name subjects, 20% (possibly product-class dependent), supports the findings of Jacoby, Szybillo, and Busato-Schach (1974) regarding the importance of brand name as a chunk--an organized, familiar unit of data that summarizes a great deal of information (Simon, 1974). A brand name itself comes to summarize configurations of attributes rather than serving as a pointer in memory for examining a detailed attribute list.

Implications for Designing Consumer Information Environments

In a study of unit pricing, Russo, Krieser, and Miyashita (1975) found that the typical unit price display (tags below each size of each brand on the shelf facing) makes severe information processing demands upon the consumer. They argue that a consolidated display listing unit prices for all brands and sizes is much more easily processed, and show that more consumers actually used unit price information when it was displayed in the store in such a manner. Russo (1975) extends this analysis to displays involving both price and quality information.

Theoretically, the argument underlying these findings is that comparisons of brands along a given attribute are relatively easy to make, because initially weightings across attributes (possibly incommensurate) are not required. One can determine which alternative is "best" on attribute 1 and which is "best" on attribute 2 without weighting attributes 1 and 2. Since some alternatives will be discarded just by examinations within an attribute, a smaller set of alternatives is left for making trade-off decisions. In addition, some attributes may be ignored if all brands are equivalent for that attribute. However, processing by brands involves immediate weighting (explicit or implicit) of attributes to develop an overall evaluation for each brand. Thus, processing by attributes can be argued to be cognitively easier than processing by brands in making brand comparisons for typical consumer information environments (Russo, 1975; Russo and Rosen, 1975; Tversky, 1969). In the present study, more people used attribute processing. Use of brand processing may still be relatively high because of the brand name and the information it provides. Also, Russo's arguments apply to the brand comparison phase of brand choice. For perceptual encoding and storage of information about the alternatives, brand processing may be easier, as argued above. If this conjecture is true then there may be a conflict between memory storage form (by brand) and ability to make comparisons and evaluations (easiest if organized by attribute). The consolidated display should eliminate much of the need for retrieval from memory, so that facilitation of processing by attributes would still be the most effective strategy. Memory is partially decoupled from evaluation, eliminating some of the potential conflict between storage and processing.

Thus, information should be presented in a fashion which facilitates processing by attributes. This is contrary to most store displays, which are arranged by brand. This argues for a centralized summary display to facilitate attribute processing. There is one additional factor, however. Given the evidence for the limited size of short term memory (Miller, 1956; Simon, 1974), anything which helps the consumer form chunks of information would be potentially helpful. Russo (1975) argues for more use of summary information in comprehensive point of purchase displays, and for use of small numbers (2 to 3) of such summary attributes.

Thus, in general the notion is to combine into a single display features which facilitate making brand comparisons and within attribute processing and which also summarizes information for the consumer. [Brand comparisons may not be the only element of interest. Evaluation of an entire product class relative to some standard, say for nutrition, may also be desired. However, a design which facilitates the ability of consumers to process the information presented should also suffice for this purpose. In fact, processing by attribute emphasizes the value of the entire class for each attribute.] This is an instance of a more general strategy for design of information environments. This strategy is to first determine the information processing abilities and strategies of the population of interest, e.g., the form in which information is stored in memory, what kinds of information integration rules are used, the capabilities of short term memory. These parameters then are used to determine how the task of gathering information can be structured so as to be maximally congruent with the known capacities of the population. The current consumer information environment, as typified by supermarket shelf displays, seems almost maximally incongruent with what is being learned about human information processing capabilities (Russo, Krieser, and Miyashita, 1975). The typical display is set up for brand processing, the most difficult cognitive processing for comparing and evaluating brands, and summary information is typically not provided. Thus, concentration on organization of information and mode of presentation is crucial for aiding consumers through public policy (cf. Jacoby, 1974), and must augment the current focus on what specific types of information should be provided.


Carol A. Berning and Jacob Jacoby, "Patterns of Information Acquisition in New Product Purchases," Journal of Consumer Research, 1(September 1974),18-22.

James R. Bettman, "Information Processing Models of Consumer Behavior," Journal of Marketing Research, 7(August 1970).

Bobby J. Calder, "The Cognitive Foundations of Attitudes: Some Implications for Multi-Attribute Models," in M. J. Schlinger, ed., Advances in Consumer Research, Volume II. (Chicago, Ill.: Association for Consumer Research, 1975,241-7).

Ralph N. Haber, "Effects of Coding Strategy on Perceptual Memory," Journal of Experimental Psychology, 68(November 1964),357-62.

G. David Hughes and Michael L. Ray, Buyer/Consumer Information Processing (Chapel Hill, North Carolina: University of North Carolina Press, 1974).

Jacob Jacoby, "Consumer Reaction to Information Displays: Packaging and Advertising," in Salvatore F. Divita, ed., Advertising and the Public Interest (Chicago: American Marketing Association, 1974,101-18).

Jacob Jacoby, Robert W. Chestnut, Karl C. Weigl, and William Fisher, "Prepurchase Information Acquisition: Description of a Process Methodology, Research Paradigm, and Pilot Investigation," in B. B. Anderson, ed., Advances in Consumer Research, Volume III (Chicago: Association for Consumer Research, 1976), forthcoming.

Jacob Jacoby, George J. Szybillo, and Jacqueline Busato-Schach, "Information Acquisition Behavior in Brand Choice Situations," Purdue Papers in Consumer Psychology, No. 140, 1974.

Daniel Kahneman, Attention and Effort (Englewood Cliffs, N.J.: Prentice-Hall, 1973).

Joseph S. Lappin, "Attention in the Identification of Stimuli in Complex Visual Displays," Journal of Experimental Psychology, 75(November 1967),321-28.

Denis A. Lussier and Richard W. Olshavsky, "An Information Processing Approach to Individual Brand Choice Behavior," paper presented at the ORSA/TIMS Joint National Meeting, San Juan, Puerto Rico (October 1974).

George A. Miller, "The Magical Number Seven, Plus or Minus Two," Psychological Review, 63(March 1956),81-97.

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

J. Edward Russo, "An Information Processing Analysis of Point-of-Purchase Decisions," Proceedings of the American Marketing Association Fall Conference (1975), forthcoming.

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

J. Edward Russo and Larry D. Rosen, "An Eye Fixation Analysis of Multi-Alternative Choice," Memory and Cognition, 3(May 1975),267-76.

Herbert A. Simon, "How Big is a Chunk?" Science, 183(February 8, 1974),482-88.

Herbert A. Simon and Michael Barenfeld, "Information-Processing Analysis of Perceptual Processes in Problem Solving," Psychological Review, 76(September 1969), 473-83.

Ola Svenson, "Coded Think Aloud Protocols Obtained When Making A Choice to Purchase One of Seven Hypothetically Offered Houses: Some Examples," unpublished paper, (University of Stockholm, 1974).

Amos Tversky, "Elimination by Aspects: A Theory of Choice," Psychological Review, 79(July 1972),281-99.

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

Peter Wright, "The Use of Phased, Noncompensatory Strategies in Decisions Between Multi-Attribute Products," Research paper 223, Graduate School of Business, (Stanford University, 1974).



James R. Bettman, University of California, Los Angeles
Jacob Jacoby, Purdue University


NA - Advances in Consumer Research Volume 03 | 1976

Share Proceeding

Featured papers

See More


How Matte Product Surface Enhances Perceived Durability

Taehoon Park, University of South Carolina, USA
Junghan Kim, Singapore Management University, Singapore

Read More


Product Complexity as a Barrier to Consumer Financial Decision-Making

Timothy Dunn, University of Colorado, USA
Philip M. Fernbach, University of Colorado, USA
Ji Hoon Jhang, Oklahoma State University, USA
John Lynch, University of Colorado, USA

Read More


Nostalgiacising: A Performative Theory of Nostalgic Consumption

Ela Veresiu, York University, Canada
Ana Babic Rosario, University of Denver
Thomas Derek Robinson, City University of London, UK

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.