A Study of the Interface Between Attitude Structure and Information Acquisition Using a Questionnaire-Based Information-Display Sheet

ABSTRACT - A new technique for measuring information acquisition in a survey questionnaire is applied to a study of the relation between the selection of brand-attribute cues and the multiattribute model of attitude structure. A relationship is found between attribute importance and the acquisition and impact of product-related information.


Morris B. Holbrook and Karl A. Maier (1978) ,"A Study of the Interface Between Attitude Structure and Information Acquisition Using a Questionnaire-Based Information-Display Sheet", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 93-98.

Advances in Consumer Research Volume 5, 1978      Pages 93-98


Morris B. Holbrook, Columbia University

Karl A. Maier, Ketchum, MacLeod & Grove, Inc.

[Morris B. Holbrook gratefully acknowledges the support of Columbia University's Faculty Research Fund.]


A new technique for measuring information acquisition in a survey questionnaire is applied to a study of the relation between the selection of brand-attribute cues and the multiattribute model of attitude structure. A relationship is found between attribute importance and the acquisition and impact of product-related information.


A growing body of consumer research has investigated the acquisition of information during the decision-making process. By far the most common technique for studying this search phenomenon uses information-display boards that array cues on m attributes of n brands in an m x n matrix (Bettman, 1976; Bettman and Jacoby, 1976; Chestnut, 1976; Jacoby, 1975, 1976; Jacoby, Chestnut, and Fisher, n.d.; Jacoby, Chestnut, Weigl, and Fisher, 1976; Jacoby, Szybillo, and Busato-Schach, 1977; Payne, 1976a, b). Typically, choice-simulating subjects select a brand-attribute combination of interest, remove a card from that cell of the matrix, read the information on the back of the card, and deposit it into a pile before choosing the next card. The resulting deck of used cards represents the extent and sequence of information sought about each brand on each attribute.

The potential relevance of such information-acquisition measures to the multiattribute models of attitude structure is obvious. Yet a comprehensive review of recent attitude research in marketing suggests that little effort has been made by consumer researchers (with or without the use of information-display boards) to study the interface between attitude structure and the informational inputs that presumably shape beliefs or other basic components of the multiattribute models (Holbrook, 1976).

Holbrook (1975, 1976, 1977) has argued that--according to Bayesian decision theory (Edwards, 1965; Green, 1964; Winter, 1975), Berlyne's (1960) theory of cognitive motivation, and the theory of perceived risk (Cox, 1967; Ross, 1975; Taylor, 1974)--information to aid consumer decision making should be sought primarily on those attributes that are subjectively perceived as most important. More broadly, it appears that the value of information should depend upon the importance, uncertainty, and confidence associated with the attributes to which it pertains. This conclusion is supported by the so-called Principle of Information-Processing Parsimony (Haines, 1974, p. 96), which suggests that limited cognitive capacity forces problem solvers to adopt heuristics restricting their attention to the minimum amount of data necessary for satisfying decisions (Bettman, 1974; Haines, 1974; Payne, 1976a; Wright, 1973, 1974). That principle is consistent with the lexicographic decision model's assumption that information on various attributes will be sought in order of their perceived importance (Payne, 1976a, b; Wright, 1975), but apparently this hypothesized relation between subjective attribute importance and search priority has been tested only occasionally, rather unsystematically, and with mixed results (Jacoby et al., n.d., 1976; Bettman, 1974; Myers, 1976; Nakanishi and Bettman, 1974; Tigert, 1966; Winter, 1975). In further accord with the Principle of Parsimony, numerous studies have shown that a very simple model of attitude structure, summing only the belief scores on those attributes considered the few most important, predicts affect at least as well as more complex forms of the multiattribute model (Bass and Wilkie, 1973; Holbrook, 1976; Holbrook and Hulbert, 1975). All these perspectives on attribute importance suggest that its role at the interface between attitude structure and information acquisition may be fundamental to our understanding of preference formation and brand choice.

If such propositions are to be tested on reasonably large and representative consumer samples of the type recently called for by Ferber (1977), it will be useful to develop measures of information acquisition that can be mass administered through survey-questionnaire techniques. To this end, the authors have designed an information-display sheet that is analogous to the information-display board except that the respondent acquires cues by removing gummed stickers placed over the cells of interest in an n x m brand-attribute matrix. Self-administration of this instrument makes it suitable for use in field-survey research.

Accordingly, the study reported below had three major purposes: (1) to evaluate the feasibility of the questionnaire-based sticker-removal task as a measure of information acquisition; (2) to test the hypothesis that attribute importance, as an indicant of information value, is a key determinant of the extent and sequence of information seeking on that attribute; (3) to test the hypothesis that preference is adequately predicted by a simple model summing the evaluative beliefs on those attributes considered important enough to be the object of search.


Subjects and Choice Task

One hundred students at Columbia University's Graduate School of Business participated in a questionnaire-based choice task in which each respondent sought information before selecting a phonograph record to receive as a raffle prize if he was one of five winners in the lottery for a $5.00 gift certificate. In order to approximate the conditions typical of a mail survey, respondents completed the questionnaires at their leisure on a self-administered basis.

The records presented in the choice task were fictitious vocal albums that differed on six attributes described by the questionnaire as follows:

1. SINGER'S STYLE: traditional--a well known pop artist who has been around for a number of years; or contemporary--a relative newcomer setting trends in the current idiom

2. TYPE OF SONGS: standards--familiar tunes that most listeners have heard before; or originals and recent hits--new songs composed by the performer or recently made popular by someone else

3. TYPE OF PRODUCTION: studio--recorded in a fancy studio, usually with complex background arrangements; or live--recorded in front of a live audience, typically with a small back-up band

4. JACKET: informative--album cover contains extensive liner notes with biographical data on the artist and full credits for producers, arrangers, musicians, etc.; or visual--album cover features attractive photos or other graphic artwork suggesting the mood conveyed by the recording inside

5. LABEL: major--one of the big entertainment-industry conglomerates; or independent--a small, privately run record company

6. PRICE: $5.59--a 20% discount from the list price of $6.98; or $4.19--a 40% discount from the list price of $6.98

Attribute-Specific Measures

Immediately following these descriptions, the six attributes were rated on 7-point scales representing: (1) the importance of each attribute to the respondent's evaluation of a pop vocal album (from "not at all important'' to "extremely important"); (2) the respondent's confidence in his ability to understand and interpret information concerning each attribute (from "not at all confident" to "extremely confident"); (3) the respondent's certainty that virtually any vocal recording on the market would be satisfactory with respect to each attribute (from "not at all certain of satisfaction" to "extremely certain of satisfaction"). In addition, the two characteristics cited in describing each attribute were rated for their desirability on the 7-point scales shown below:


The importance and confidence ratings were scored from 1 to 7. By contrast, the certainty ratings were scored in the direction of increasing uncertainty, from 7 to 1. Finally, in acknowledgement of the debate surrounding the proper coding of evaluative scores (Ahtola, 1975; Fishbein, 1976; Glassman and Fitzhenry, 1976; Lutz, 1976), two alternative codings of the desirability ratings were investigated: -3 to +3 and 1 to 7.

Test Objects

The dichotomous characteristics for each attribute were used to specify eight records according to the kind of fractional factorial design described by Green (1974) as suitable for conjoint-measurement studies. The rows of the resulting 8x6 matrix were randomized and labeled from "A" to "H." The entire matrix of attribute-specific cues describing each record album appears in Table 1.

The Information-Display Sheet

After providing the attribute-specific ratings discussed above, the respondent turned to an information-display sheet consisting of the matrix shown in Table 1 with each cell covered by a gummed sticker appropriately labeled from A1 to H6. Written instructions asked the respondent to remove enough stickers to give him the information needed to choose the album he wished to receive as his raffle prize if he won the drawing for a $5.00 gift certificate. The bottom of the page contained another matrix entitled "Order of Sticker Removal" with boxes labeled from "1st" to "48th." The respondent was instructed to place each sticker into the box designating its order of removal. He was told that he could obtain all the information needed to make his choice by peeling off as many stickers as he wanted, but that each piece of information acquired would result in the deduction of 5 cents from the $5.00 value of his potential prize.

After completing this task and making his choice, the respondent was further instructed to continue removing stickers until he had collected enough information to permit him to rank all eight recordings in order of preference. He was told that no cost would be imposed for these additional stickers and that they should therefore be placed in a second set of boxes labeled "1st" to "48th" and entitled "Order of Removal of Additional Stickers." When he had completed this second process of cue selection, the respondent ranked all eight records in order of preference.

In sum, this procedure obtained measures of both choice and preference order with associated indices of costly and costless information acquisition. Not surprisingly, the 5-cent charge for cues in the choice task was somewhat discouraging to information acquisition. On the average, respondents acquired only 9.02 cues in this task, as compared with an average of 25.46 in both the choice and ranking tasks combined. Full analyses of both the single and combined information-acquisition measures produced virtually identical results. Accordingly, only the data for the combined total of cues selected are discussed in the present report.

Operational Definitions

Extent of information acquisition. The extent of information acquisition (EIA) on each of the six attributes was defined operationally as the number of cues acquired across all eight records:


where ACQij is a zero-one dummy variable representing the removal of a sticker in either the choice or ranking task, i refers to the eight records (i = 1, . . . , 8), and j refers to the six attributes (j = 1, . . ., 6).

Sequence of information acquisition. The sequence of information acquisition (SIA) on each attribute was defined operationally as the average rank order of acquiring information on that dimension across the eight records:


where RANKACQij is either the rank order in which a cue was actually selected in the two tasks combined or the average rank order for those remaining cues not selected in either task.



Information value. Several alternative indices of information value were examined for their ability to predict the extent and sequence of information acquisition. In addition to the attribute-specific scores for importance (IMPj), confidence (CONj), and uncertainty (UNCj), multiplicative information-value indices were defined-as follows:

EQUATIONS  (3) and   (4)

Attitude structure. In accord with the multiplicity of available multiattribute models of attitude structure (Bass and Wilkie, 1973; Fishbein and Ajzen, 1975; Holbrook, 1976; Wilkie and Pessemier, 1973), several alternative attitude models were examined for their ability to predict preference order:

EQUATIONS  (5),   (6),  (7),  and  (8)

where k refers to the twelve characteristics that a given record does (k = 1, . . . , 6) and does not (k = 7, . . . , 12) possess; Bk = 1 if the record possesses characteristic k or Bk = -1 if it does not; Ek represents a respondent's evaluation of the desirability of characteristic k (scored from -3 to +3 or from 1 to 7); and ACQk is a zero-one dummy variable representing the acquisition of information on characteristic k.

In addition to the attitude models listed above, importance-weighted versions were also investigated. These were identical to the above forms except that BkCEkCACQk was replaced by BkCEkCIMPkCACQk. Note that, for all such models, the inclusion of the multiplicative ACQk term has the effect of setting belief scores equal to zero for those characteristics about which information was not acquired. Note also that the Partial Models assume that preference depends only upon the evaluations of the characteristics known to be present in a record whereas the Full Models assume that preference may be enhanced (reduced) by the absence of undesirable (desirable) characteristics. The Full Models thus incorporate logic analogous to that underlying arguments for the bipolar coding of belief and evaluative-aspect scores in the Fishbein model (Fishbein, 1976; Fishbein and Ajzen, 1975).


Given the operational measures defined above, two key hypotheses were tested:

H1: Information value predicts the extent and sequence of information acquisition

H2: Attitude-structure models based on the information acquired predict preference rank

There were no a priori hypotheses concerning the relative predictive efficacy of importance, confidence, uncertainty, the importance-confidence index, and the importance-uncertainty index as measures of information value or of the Partial and Full Additive and Averaging Models as representations of attitude structure. Rather, findings concerning the relative performance of these alternative indices and models were considered exploratory.

Statistical Procedures

Statistical routines were run separately for each respondent using (1) Pearson product-moment correlations across the six attributes to test hypothesis H1 and (2) Spearman rank-order correlations across the eight records to test hypothesis H2. The use of intraindividual analysis is considered essential in such correlational tests to avoid utility comparisons between individuals and to minimize yea-saying bias (Bass and Wilkie, 1973).


Few problems appeared to arise in the respondents' self-administration of the instrument described above. All but a handful of the questionnaires were returned in proper order (the exceptions were, of course, discarded), and where a sticker was occasionally damaged or defaced, respondents showed a reassuring tendency to write in the missing identification number by hand or to use a piece of Scotch tape to stick it firmly in place.



Results for hypothesis H1 are presented in Table 2. Attribute importance offered fairly good mean predictions of both extent and sequence of information acquisition (r = .587, zr = 15.45, [ < .0001; r = -.633, zr = -17.11, p < .0001). Because the distributions of these correlations were skewed, the medians were substantially stronger than the means (r = .722 vs. .587; r = -.752 vs. -.633). The predictive performance of attribute importance was significantly better than that of either confidence or uncertainty: r = .587 vs. .186 (t99 = 8.33, p < .0001); r = .587 vs. .055 (t99 = 7.38, p < .0001); r = -.633 vs. -.161 (t99 = 9.25, p < .0001); r = -.633 vs. -.052 (t99 = 8.18, p < .0001). Finally, the multiplicative indices of information value (importance-confidence and importance-uncertainty) failed to improve upon the predictions obtained by importance considered alone: r = .587 vs. .543 (t99 = 2.87, p < .01); r = .587 vs. .438 (t99 = 3.82, p < .001); r = -.633 vs. -.577 (t99 = 3.11, p < .01); r = -.633 vs. -.478 (t99 = 3.90, p < .001). In sum, among those variables tested, attribute importance (with median correlations above r = .70) was the simplest and best index of information value for predicting the extent and sequence of information seeking.

Table 3 presents the results for hypothesis H2. In no case did the coding of desirability (Ek) from 1 to 7, instead of from -3 to +3, have any statistically significant impact on the performance of the Partial or Full Additive or Averaging Models in predicting preference. Nor did the inclusion of importance weights provide any significant improvement in the predictive performance of these models (probably because the models included only those belief scores considered important enough to have prompted information seeking in the first place). Table 3 therefore contains only the rank-order correlational results for the simple Partial and Full Additive and Averaging Models with desirability (Ek) coded in the theoretically appropriate manner, from -3 to +3. It is clear that all four versions of the model supported hypothesis H2 by performing fairly well, with mean rank-order correlations ranging from -.542 to -.568 (Zr > 13.22, p < .0001). Again, the distributions of these correlations are skewed so that the medians are considerably stronger, ranging from -.670 to -.704. Since there is no statistically significant difference between the correlations obtained by any pair of these four models, it might be argued that the "best" is therefore the simplest--namely, the Partial Additive Model, which simply sums the desirability scores of those characteristics known (after information seeking) to be possessed by a particular record. Although these results concerning the relative performance of the Additive and Averaging Models were considered exploratory in the present research, this finding may reinforce doubts about the often-claimed superiority of the averaging formulation (Lutz, 1976; cf. Fishbein, 1976; Fishbein and Ajzen, 1975). Further work comparing additive and averaging versions, using the sticker-removal task, is in progress.





The study reported above is subject to several important limitations. First, though an effort was made to select a product (phonograph records) of considerable relevance to the student population, it may be that, in acquiring information to evaluate records, students behave differently from housewives evaluating other consumer products (e.g., toothpaste), thus limiting the generalizability of the findings. Moreover, the records used were identified only by letter rather than by title and name of artist. This procedure helped guard against confounding effects, but represented a further departure from the real market place. It might also be objected that the self-administration of questionnaires passed out by hand differs in some nonobvious way from the completion of those received through the mail. Given the relative anonymity of the respondents, however, this potential problem does not appear to have been too damaging in the present situation.

Perhaps most seriously, the study's design may be subject to the impact of task structure on acquisition strategy recently demonstrated by Bettman and Kakkar (1977). For example, the first attribute (singer's style) was widely regarded as the most important, thus suggesting the alternative hypothesis that the relationship between importance and acquisition may have resulted, in part, from a tendency for respondents to begin at the left of the matrix and work their way to the right. This interpretation is at least partly ruled out by the fact that the last attribute (price), on the far right, was also generally regarded as highly important. Nevertheless, future work with the sticker-removal task will attempt to eliminate such potential order effects.


Apart from these limitations, the study seems to have achieved its purpose of using a new technique, suitable for administration by survey questionnaire, to test two key hypotheses concerning the relation of information acquisition to the multiattribute model. In general, the findings support the key role played by attribute importance at the interface between information acquisition and attitude structure. Specifically, in accord with hypothesis H1, attribute importance appears to guide search toward the most important attribute-related cues. Further, in accord with hypothesis H2, these most important cues appear to determine preference rank. This support for hypotheses H1 and H2 suggests a determinant role of perceived attribute importance in directing search toward information that, in turn, shapes the beliefs incorporated into attitude structure.

Finally, the study demonstrated the usefulness of the questionnaire-based sticker-removal task for investigating information acquisition. If consumer researchers are to heed Ferber's (1977) recent call for studies using more representative samples, with a probable resultant loss in the feasibility of applying the techniques of laboratory experimentation, some such survey approach to studying information acquisition may well be necessary in the future.


O. T. Ahtola, "The Vector Model of Preferences: An Alternative to the Fishbein Model," Journal of Marketing Research, 12(February, 1975), 52-59.

F. M. Bass and W. L. Wilkie, "A Comparative Analysis of Attitudinal Predictions of Brand Preference," Journal of Marketing Research, 10(August, 1973), 262-269.

D. E. Berlyne, Conflict, Arousal, and Curiosity (New York: McGraw-Hill, 1960).

J. R. Bettman, "Toward a Statistics for Consumer Decision Net Models," Journal of Consumer Research, 1(June, 1974), 71-80.

J. R. Bettman, "Data Collection and Analysis Approaches for Studying Consumer Information Processing," Working Paper No. 41, Center for Marketing Studies, University of California, Los Angeles, 1976.

J. R. Bettman and J. Jacoby, "Patterns of Processing in Consumer Information Acquisition," Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976, 315-319.

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

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

D. F. Cox, ed., Risk Taking and Information Handling in Consumer Behavior (Cambridge, Mass.: Harvard University Graduate School of Business Administration, 1967).

W. Edwards, "Optimal Strategies for Seeking Information: Models for Statistics, Choice Reaction Times, and Human Information Processing," Journal of Mathematical Psychology, 2(1965), 312-329.

R. Ferber, "Research By Convenience," Journal of Consumer Research, 4(June, 1977), 57-58.

M. Fishbein, "Extending the Extended Model," Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976, 491-497.

M. Fishbein and I. Ajzen, Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research (Reading, Mass.: Addison-Wesley, 1975).

M. Glassman and N. Fitzhenry, "Fishbein's Subjective Norm: Theoretical Considerations and Empirical Evidence,'' Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976, 477-480.

P. E. Green, "Uncertainty, Information, and Marketing Decisions," in R. Cox, W. Alderson, and S. J. Shapiro, eds., Theory in Marketing (Homewood, Ill.: Irwin, 1964), 333-354.

P. E. Green, "On the Design of Choice Experiments Involving Multifactor Alternatives," Journal of Consumer Research, 1(September, 1974), 61-68.

G. H. Haines, Jr., "Process Models of Consumer Decision Making," in G. D. Hughes and M. L. Ray, eds., Buyer/Consumer Information Processing (Chapel Hill: University of North Carolina Press, 1974), 89-107.

M. B. Holbrook, "A Study of Communication in Advertising," unpublished doctoral dissertation, Columbia University, 1975.

M. B. Holbrook, "Attitude Structure and Communication Effects: The Neglected Interface," working paper, Columbia University, 1976.

M. B. Holbrook, "Beyond Attitude Structure: Toward the Informational Determinants of Attitude," working paper, Columbia University, 1977.

M. B. Holbrook and J. M. Hulbert, "Multi-Attribute Attitude Models: A Comparative Analysis," Proceedings, Association for Consumer Research, Fifth Annual Conference, 1975, 375-388.

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

J. Jacoby, "The Emerging Behavioral Process Technology in Consumer Decision-Making Research," Purdue Paper in Consumer Psychology No. 169, Purdue University, 1976.

J. Jacoby, R. W. Chestnut, and W. A. Fisher, "Simulating Nondurable Purchases: Individual Differences and Information Acquisition Behavior," unpublished paper, Purdue University, n.d.

J. Jacoby, R. W. Chestnut, K. C. Weigl, and W. A. Fisher, "Pre-Purchase Information Acquisition: Description of a Process Methodology, Research Paradigm, and Pilot Investigation,'' Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976, 306-314.

J. Jacoby, G. J. Szybillo, and J. Busato-Schach, "Information Acquisition Behavior in Brand Choice Situations," Journal of Consumer Research, 3(March, 1977), 209-216.

R. J. Lutz, "Conceptual and Operational Issues in the Extended Fishbein Model," Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976, 469-476.

J. G. Myers, "An Interactive Computer Approach to Product Positioning," paper presented at the Attitude Research Conference, Hilton Head, South Carolina, February, 1976.

M. Nakanishi and J. R. Bettman, "Attitude Models Revisited: An Individual Level Analysis," Journal of Consumer Research, 1(December, 1974), 16-21.

J. W. Payne, "Heuristic Search Processes in Decision Making," Proceedings, Association for Consumer Research, Sixth Annual Conference, 1976a, 321-327.

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

I. Ross, "Perceived Risk and Consumer Behavior: A Critical Review," Proceedings, Association for Consumer Research, Fifth Annual Conference, 1975, 1-19.

J. W. Taylor, "The Role of Risk in Consumer Behavior," Journal of Marketing, 38(April, 1974), 54-60.

D. J. Tigert, "An Investigation of Experimental Techniques to Determine the Relative Importance of Various Product Attributes in the Consumer Decision-Making Process," Institute Paper No. 126, Institute for Research in the Behavioral, Economic, and Management Sciences, Krannert Graduate School of Industrial Administration, Purdue University, 1966.

W. L. Wilkie and E. A. Pessemier, "Issues in Marketing's Use of Multi-Attribute Attitude Models," Journal of Marketing Research, 10(November, 1973), 428-441.

F. W. Winter, "Laboratory Measurement of Response to Consumer Information," Journal of Marketing Research, 12(November, 1975), 390-401.

P. L. Wright, "The Cognitive Processes Mediating Acceptance of Advertising," Journal of Marketing Research, 10(February, 1973), 53-62.

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

P. L. Wright, "Consumer Choice Strategies: Simplifying Vs. Optimizing," Journal of Marketing Research, 12(February, 1975), 60-67.



Morris B. Holbrook, Columbia University
Karl A. Maier, Ketchum, MacLeod &amp; Grove, Inc.


NA - Advances in Consumer Research Volume 05 | 1978

Share Proceeding

Featured papers

See More


K13. When Does Humor Increase Sharing?

John Yi, University of Arizona, USA
Caleb Warren, University of Arizona, USA

Read More


N7. Emotion Or Information? Effects Of Online Social Support On Customer Engagement

Chuang Wei, Tsinghua University
Maggie Wenjing Liu, Tsinghua University
Qichao Zhu, Tsinghua University

Read More


The Effect of Psychological Control on Temporal Discounting: Conceptual and Methodological Implications

Kelly Kiyeon Lee, Georgetown University, USA
Selin A. Malkoc, Ohio State University, USA
Derek Rucker, Northwestern 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.