Behavioral Measurement of the Relative Importance of Attribute-Related Information Cues: the Case of Cold Breakfast Cereals

John A. Quelch, University of Western Ontario
ABSTRACT - An information display was used to investigate the relative importance of a set of attribute-related information cues in influencing trial and repeat brand choice behavior for a frequently purchased product category, adult-oriented (i.e., non-presweetened) cold breakfast cereals. Information on subjective attribute dimensions (including physical samples of the brands on offer) was included in the display as well as information on objective attribute dimensions.
[ to cite ]:
John A. Quelch (1979) ,"Behavioral Measurement of the Relative Importance of Attribute-Related Information Cues: the Case of Cold Breakfast Cereals", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 263-268.

Advances in Consumer Research Volume 6, 1979      Pages 263-268

BEHAVIORAL MEASUREMENT OF THE RELATIVE IMPORTANCE OF ATTRIBUTE-RELATED INFORMATION CUES: THE CASE OF COLD BREAKFAST CEREALS

John A. Quelch, University of Western Ontario

[The author acknowledges the support of the Marketing Science Institute in funding this study.]

ABSTRACT -

An information display was used to investigate the relative importance of a set of attribute-related information cues in influencing trial and repeat brand choice behavior for a frequently purchased product category, adult-oriented (i.e., non-presweetened) cold breakfast cereals. Information on subjective attribute dimensions (including physical samples of the brands on offer) was included in the display as well as information on objective attribute dimensions.

INTRODUCTION

Myers and Alpert (1968 and 1977) have concluded that an attribute dimension may be important to a consumer but may not influence the brand decision-making process in a determinant sense if all brand alternatives are perceived by the consumer to be similar on that dimension. Thc distinction between importance and determinance emerged from efforts to improve the predictive power of the multiattribute model. For example, Wilkie and Weinreich (1972) reported that predictive power diminished beyond the inclusion of the two or three attributes with the highest determinance scores. However, while attribute importance weights and determinance scores may vary, the attitudinal nature of the data inputs to the model prevent the researcher from distinguishing those attributes sufficiently important or determinant to be used in decision-making from those which are not.

The pilot study reported here obtained behavioral measures of relative attribute importance through the use of an information display matrix of brand-attribute information. Characteristically, participants in information display studies elicit as few or as many items of information as they wish prior to selecting a brand. The researcher can, therefore, distinguish those attribute dimensions which are considered sufficiently important to be consulted before a brand selection is made from those which are not. Prior evidence (Jacoby, Chestnut, Weigl and Fisher, 1976) indicates that few consumers acquire all the information available to them before making their brand selections. In this study, determinant attributes are treated first as those used by a consumer in the decision-making process and secondly as those which result in a reduction in the number of brand alternatives under consideration in the process.

METHOD

Each subject was required to select a brand of adult-oriented cold breakfast cereal from among an array of six brands on offer in an information display. To assist in their decision-making, subjects could elicit information about the attribute profile of each brand.

The Sample

A non-random sample of 250 female homemakers was recruited at a shopping mall in Massachusetts. Potential subjects were screened for recent purchase and consumption of cold breakfast cereal, in addition, they had to designate an adult-oriented brand when asked to name a favorite. Qualifying subjects were invited to make appointments to participate in a study investigating a new market research technique. Each consumer was told that she would receive ten dollars in cash and coupons for her cooperation. Of 341 prospective subjects, fourteen failed to qualify on the screening criteria. The remaining non-participants either declined to participate or failed to keep their appointments.

To avoid bias against inclusion of working women in the sample population, screening and task administration were conducted during evening as well as day-time hours and during one Saturday as well as three weekdays.

Subsequent analysis of the sample indicated that lesser educated, lower income, and elderly consumers were under-represented in terms of both local and national census figures. Aggregate results cannot, therefore, be considered nationally or locally representative.

Display Structure and Task Administration

The brands on offer were listed horizontally at the same eye level. However, because people characteristically read from left to right, it was thought that a bias towards selection of information about brands on the left side of the information display might occur. Since randomization of the order in which brands were listed for each subject would have created logistical difficulties, the potential problem was investigated through the use of two information displays in different rooms at the mall location. One display presented the brands in alphabetical order, the other presented them in reverse alphabetical order. Subsequent analysis indicated no significant differences in the results derived from each information display in terms of the mean numbers of information units elicited for each brand. This test may, however, have been inadequate for the third and fourth brands in the sequence, whose positions were simply reversed between information displays, particularly if both primacy and recency effects were operative.

Since people commonly read from top to bottom, the listing of attribute dimensions on the vertical dimension presented a similar problem of order effects. In this study, attribute dimensions were not listed on the information displays. Instead, subjects were presented on a rotation basis with cards which listed the available attribute dimensions in different sequences.

Six project personnel, who had been briefed as a group and had been involved in a feasibility test of the design, administered the information display task, recording the information elicited by each subject as she proceeded. Insofar as work schedules and minimization of subject waiting time permitted, personnel were rotated between the rooms containing the two information displays to limit the impact of any inter-viewer-related biases.

Specification of Treatment Groups and Attributes

The specification of attribute dimensions is no easier a task in information display studies than in investigations of relative attribute importance using other methodologies. Subjects were assigned to one of two treatment groups which varied in the number of attribute dimensions available for elicitation. Subjects in the first treatment group (124 subjects) were presented with four information cues - ingredient information, nutrition information, physical appearance, and price. Subjects in the second treatment group (126 subjects) were presented with an additional information cue -the opportunity to elicit physical samples of the brands on offer in the exercise.

The nutrition information and ingredient information cues replicated the information presented on the boxes of the cereal brands on offer in the exercise. The format and specificity of information on these dimensions did not vary across brands. Price information represented an average of store prices in the area from which the sample was drawn, standardized for a twelve ounce package. This package size was specified on each of the price information cards.

The physical appearance dimension included a verbal description of the size, color, and texture of the cereal pieces, and a life-size drawing which enabled subjects to tell whether a particular brand was in the form of flakes, nuggets, or puffs. Photographs were not used to avoid brand identities being revealed. It is possible that the number of subjects who elicited the physical appearance dimension may have been unrealistically reduced as a result.

Physical samples of the brands on offer were available in bowls, covers on which were removed if subjects elicited the "physical sample to look at and/or to taste" attribute dimension. The number of subjects in the second treatment group who elicited physical sample information may have been unrealistically limited. Some subjects may have considered the sampling procedure unhygienic, others may have been disinclined to sample the brands in a semi-public environment. Still others may simply not have been hungry. In addition, subjects were not able to sample the brands with milk or sugar.

With the exception of price, these information cues each provided subjects with information about several attributes. For example, the nutrition information dimension provided subjects with the calorie content, protein content, and vitamin and mineral content of each brand. The rationale for not decomposing these dimensions into their constituent attributes stemmed from one of the main study objectives. This was to investigate possible variations in the relative importance of attribute dimensions among each other, conditional upon whether the purchase occasion represents a trial or a repeat purchase. In a repeat purchase situation, the consumer has had an opportunity to experience the product or brand in question. In the case of cold breakfast cereals, it was hypothesized that the importance of physical sample information to the repeat buyer would be such that the importance of the other attribute dimensions on which performance information was available to both trial and repeat buyers would be significantly reduced in the repeat purchase situation. The information available to the two treatment groups was therefore intended to represent those attribute dimensions on which performance information is available to the consumer making a trial and a repeat purchase respectively. The attribute dimensions included in each set were not evoked from subjects in this study but were based upon the results of proprietary consumer research conducted by a major cereal manufacturer. It should be noted that the simulation did not consider the impact of differential information recall, personal influences, and other distortions common to naturally occurring purchase situations.

Recognizing that actual physical samples of brands on offer in the display would constitute a multiattribute information cue, it was decided that the attribute dimensions common to both treatment groups should not be decomposed into their constituent attributes. To have done so would, furthermore, have resulted in a preponderance of nutrition or ingredient related attributes, potentially affording them an unnatural prominence and enhancing the possibility of consumer learning biasing the information elicited. In addition, Arch, Bettman, and Kakkar (1978) have indicated that in real purchase situations a consumer may acquire information on several attributes simultaneously by examining the nutrition label. Decomposition of the nutrition information (and other) dimensions might have had the effect of increasing acquisition time and thereby reducing the total quantity of information acquired. The biases which presentation formats impose upon the external validity of the information acquired in display studies merit further investigation. The search time for each information unit is equalized. The absolute and relative length of time required to search information on different attributes in the real world may not be accurately replicated. Since information search time may influence the quantity of information and the nature of attribute information acquired, the accuracy of the results of information display studies may justifiably be questioned on this basis. In particular, the display of information available may actually stimulate demand for information which, in a real purchase situation, might not have influenced choice behavior.

In contrast to the Jacoby, Chestnut, Weigl, and Fisher (1976) study, brand name was not included as an information cue. If brand name had been available and had been the only cue elicited by a subject, it would not have been possible to determine whether non-selection of additional attribute dimensions was due to their relative unimportance or to the fact that the information which they could have provided was subsumed in the brand name dimension. Similarly, selection of additional attribute dimensions might have been due to the complexity of the information associated with them, or to the difficulty of storing this information in long-term memory. Additional attribute dimensions consulted beyond brand name could, therefore, have been not the most important but those on which consumers found it relatively more difficult to store information. In addition, if brand name had been available as an information dimension, some consumers might have sought more information on those brands with which they were less familiar, and vice versa. For these reasons and because brand name was not included as an information dimension in this study, the information elicited by subjects may not represent the information which consumers with purchase experience in the cereal category would actually acquire when making a brand purchase in a supermarket.

Specification of Brands

Six brands of cold breakfast cereal were on offer in the brand selection exercise. The brands were real, but were identified by code letters (L through Q) rather than brand name. To discourage subjects from selecting attribute information in order to identify their favorite brands, they were told that the brands on offer were not currently available in New England. Clearly, the attribute dimensions most useful in brand identification may not necessarily be those most important to the consumer.

One constraint upon the choice of brands was the need to avoid identifiable brands, since physical samples were available as an information cue to subjects in the second treatment group. Otherwise, subjects might have believed that they had identified particular brands and might have made their brand selections on the basis of brand name impression rather than the attribute information available from the information display.

After consultation with marketing research executives of a major cereal manufacturer, six brands were selected to represent the adult-oriented segment of the cold breakfast cereal market. Representativeness of brands in terms of product form, price, nutritional and ingredient content, was essential if measures of the relative importance of these attribute dimensions were to be generalizable rather than specific to the brand set. Industry studies classified two of the six brands (Kellogg's Corn Flakes and Rice Chex) as all-family, two (Total and Special K) as fortified, and there was, in addition, one bran (Post Raisin Bran), and one granola (Quaker Natural). In terms of volume, the six brands represented about twenty percent of the cold breakfast cereal market.

Procedural Constraints

Ordinarily, in information display studies, subjects have elicited information on a cell-by-cell basis. Since this study focused on investigating the relative importance of attribute information, subjects were constrained to acquiring information attribute-by-attribute across brands, and they were not permitted to return to an attribute dimension a second time to acquire additional information.

This procedure is compatible with a lexicographic model of information acquisition which assumes that consumers examine several brands on the most important attribute and, in the event of a tie, examine the remaining brands on the second most important attribute and so on until a preferred brand emerges. The lexicographic model suggests that the relative importance of attributes be delineated in advance of the selection process (Wright, 1972) and may, therefore, be appropriate for the study of frequently purchased product categories (such as breakfast cereals) in which most subjects have prior purchase experience.

Information display studies have indicated that more consumers use an attribute-by-attribute approach to information acquisition than a brand-by-brand approach (Bettman and Kakkar, 1977; Jacoby, Chestnut, Weigl, and Fisher, 1976; Reilly, Holman, and Evered, 1976). In an eye movement study, Russo and Rosen (1975) also found a greater propensity to information acquisition attribute-by-attribute. While the literature supports the attribute approach, Bettman and Jacoby (1976) have pointed out that, because product information is organized by brand on the supermarket shelf, some consumers who prefer to use an attribute-by-attribute approach may, in fact, use a brand-by-brand approach to information acquisition. The common procedure used in this study may have constrained some consumers to follow an information acquisition strategy with which they were unfamiliar or uncomfortable. In these cases, information processing time may have been unrealistically increased or the quantity of information acquired may have been unrealistically curtailed.

Subjects had to consult the information cards for all six brands on the first attribute dimension which they selected. The purpose of this constraint was to encourage subjects from the outset to look at the information display as a whole rather than to concentrate on the left-hand side, for example. In addition, one research objective was to ascertain the percentage of consumers whose favorite brand of cereal was one of the six brands on offer in the exercise who, in fact, chose the favorite brand [These results are reported in John A. Quelch, "Explaining and Predicting Brand Choice Behavior: An Application of Information Display Research", Working Paper No. 181, School of Business Administration, The University of Western Ontario. In fact, of ninety-six subjects who had the opportunity to select their favorite brands, thirty-four did so, a proportion significantly (at the 0.05 level) higher than the one-sixth or sixteen subjects which would have been expected by chance.]. To do this, it was necessary to ensure that all subjects were exposed to at least one item of information about all six brands.

This constraint may have affected the brand selection process of those consumers who had in mind to acquire a fixed quantity of information or to spend a fixed period of time on the task prior to making a brand selection. Some subjects might have consulted fewer brands on the first attribute dimension. Alternatively, the first attribute dimension selected might not have been the most important, but, rather, one on which the performance of six brands was perceived as being reviewable in a shorter period of time.

Beyond the first attribute dimension, the constraint did not apply. Subjects could choose the brand or brands for which they elicited information on each additional attribute dimension. They were not, therefore, required to take potentially redundant attribute information on brands which they might have eliminated from further consideration on the basis of information already acquired.

Once a subject selected an attribute dimension, she was asked to remove the cards for those of the Six brands on which she wished to see the information before reviewing any of these cards. While this constraint may have increased information acquisition time, its purpose was twofold. First, it counteracted the advantage which brands on the left side of the information display might have had if a subject, proceeding from left to right across a row of attribute information, had been permitted to review each card in turn. Secondly, the validity of investigations into whether selection of particular attribute dimensions resulted in differential reduction in the number of brands considered on the next dimension elicited would have been impaired if subjects had been able to review information cards one by one - the reason being that the number of cards actually removed might have been fewer than the number of cards which the subject originally intended to take based on brand information elicited on the previous attribute dimension.

Subjects in this study were not required to turn information cards face down once they had looked at them. In a study focusing on relative attribute importance, it would have been difficult to determine whether back referencing was an indication of the importance of the attribute dimension, consumer forgetfulness, or the relative difficulty of retaining the information. Accordingly, subjects in this study were permitted to freely refer back to information already acquired.

Compensation

To introduce a sense of motivational realism to the task similar to that present in naturally occurring purchase situations, and to discourage elicitation of information on the basis of perceived social acceptability, subjects were told that they would receive two fifteen cents-off coupons towards future purchases of the brand which they selected.

It was believed that the compensation level could influence the perceived importance of the task and, therefore, the quantity and nature of information elicited. A free box of each subject's selected brand was rejected because it removed any incentive to elicit price information, other than perhaps to discover the highest-priced brand. Measurements of the relative importance of the attribute dimensions included in the exercise would have been rendered unrealistic.

It was believed that a fifteen cents-off coupon was within the normal range of coupon values for cold breakfast cereals. No more than two coupons could be offered since brand choice decisions are, in reality, discrete, and consumers purchase on average between one and two boxes of cold breakfast cereals at any one time. To have offered more than two coupons might have caused subjects to become unrealistically risk averse in their information acquisition, attribute elicitation, and brand selection strategies.

RESULTS

In this section, the quantity of information acquired by subjects in terms of number of attribute dimensions consulted and number of information units elicited is first reported together with an examination of the relationship between quantity of information available and quantity of information acquired across treatment groups. Secondly, three measures of relative attribute importance are applied to the data.

Acquisition of Information

No subject in either treatment group elicited all the information units available. Indeed, 23 percent of subjects were willing to base their brand selection decisions on only six information units, the minimum number which they were permitted to take.

Across the whole sample, each subject consulted an average of 2.26 attribute dimensions. A larger percentage (40 percent) of subjects selected information on two attribute dimensions than on any other number of attributes. The results indicated that the majority of consumers were content to use only one or two attribute dimensions as the basis for making brand selections.

Comparing the results for subjects in both treatment groups, the quantity of information acquired was found to be unrelated to the quantity of information available. The mean numbers of information units acquired were respectively, 9.79 for subjects in the first treatment group (with 24 units available), and 9.34 for subjects in the second group (with 30 units available). The corresponding mean numbers of attributes consulted were 2.25 and 2.27. In neither case was the difference between the two means statistically significant.

It is possible that the incremental six units of information were insufficient to make meaningful any comparison between the two treatment groups in terms of the quantity of information available, particularly since the additional attribute dimension available to subjects in the second treatment group (physical sample information) was, in fact, an expanded version of the physical appearance dimension and may have been regarded by subjects as substitute rather than independent information.

The numbers of information units acquired by subjects in this study were similar to those reported by Jacoby, Chestnut, Weigl, and Fisher (1976) for an information display study in which subjects were also required to select a brand of cold breakfast cereal. The two studies differed in the number and nature of the brands and attributes included, in the nature of the task, and in the level of compensation offered to participants. Interestingly, however, half of the Jacoby et al. sample acquired eleven or fewer information units, compared to 74 percent of subjects in this study. The tendency of subjects in this study to acquire fewer information units may be explained by the fact that the information dimensions were not as decomposed as in the Jacoby et al. study. For example, nutrition information was presented as a single attribute dimension in this study, whereas it was decomposed by Jacoby et al. into the constituent nutrients. An information unit in this study tended, therefore, to represent a larger "chunk" of information than an information unit in the Jacoby et al. study.

Behavioral Measures of Relative Importance

Three different, though not independent, measures of the relative importance of information on attribute dimensions were derived from the data collected in the brand selection exercise:

Selection or non-selection of an attribute dimension. This distinction enables those attribute dimensions to be identified which were perceived by each subject to be sufficiently determinant to warrant investigation. Other attribute dimensions, not consulted, may also have been perceived to be determinant (i.e., both important and differentiated across brands) but not to the level necessary to justify their inclusion in the decision-making process. Perceived determinance is a necessary but not a sufficient condition for an attribute to be consulted. The aggregate frequency with which each of the five information dimensions was elicited by members of the two treatment groups is indicated in the following tabulation:

TABLE

The results, assuming sample homogeneity, indicate that nutrition information and ingredient information were most frequently perceived as determinant attribute dimensions in the trial purchase situation, while physical sample information was most often perceived as determinant in the repeat purchase situation.

Given that the total number of elicitation occasions did not increase significantly between treatment groups, consultation of the physical sample dimension clearly occurred at the expense of the other four. The results enable the comparative changes in importance of these attribute dimensions to be assessed. For example, the share of total elicitation occasions held by the physical appearance dimension declined substantially between treatment groups, while the share held by the price dimension remained constant. The differential changes in elicitation occasion shares probably reflect, in part, the relative degree to which information on each of the four common attribute dimensions was perceived to be obtainable from information on the physical sample dimension. For example, physical samples of breakfast cereals may provide some consumers with information about their principal ingredients but do not apparently provide sufficient information on the price dimension to cause a decrease in the frequency with which this attribute is consulted.

Order of Attribute Dimension Selection. Table 1 shows the frequency with which each information dimension was elicited by subjects in the two treatment groups for each of five possible selection sequence positions. A similar measure of relative attribute importance has been used in an information display study reported by Holbrook and Maier (1978). Conclusions drawn from the frequency of elicitation data can be amplified if the rank order data are examined. For example, while the overall frequency of elicitation of ingredient information fell by 27 percent between the two treatment groups, the frequency with which this attribute dimension was elicited first fell by 42 percent. In contrast, the overall frequency of elicitation of nutrition information fell by 23 percent, but the frequency with which it was elicited first fell by only 18 percent. Nutrition information appeared, therefore, to be less susceptible to displacement than ingredient information, given the availability of the physical samples.

The credence attached to this measure of relative importance rests upon the assumption that consumers will first elicit information on the attribute which is perceived a priori to be the most important with a view to minimizing the duration of the decision-making process. In this regard, three problem areas should be noted. First, for the consumer whose decision-making process involves a distinction between narrowing the range of brand options to a set of acceptable alternatives, and then choosing from among the brands in this set, the first attribute dimension selected may not be the most important in contributing to the final decision - particularly in the case of a consumer who uses different attribute dimensions at different stages of the decision-making process. Secondly, the order in which attribute information is acquired throws no light on the intended use of the information - in particular, whether the subject wishes to treat the information in a threshold or maximizing sense. Third, in the absence of prior purchase experience and prior definition of purchase goals, subjects may choose attribute dimensions which they believe to be important but which prove in fact to be less important in facilitating a brand choice decision than other attribute dimensions elicited later in the rank order. In brief, a priori and post hoc assessments of relative importance may be divergent.

To investigate this third problem area, each subject in the study was asked upon completion of the exercise to allocate one hundred points among those attribute dimensions which she consulted to reflect the relative importance of each in facilitating the brand selection process. For each subject, a Spearman nonparametric rank order correlation coefficient was computed between the rank order in which the attribute dimensions were selected, and the rank order in which points were allocated to each of the same dimensions. The mean coefficient across all subjects was 0.71. The reliability of the points allocation measure may have been distorted by socially acceptable response patterns (for example, allocating a disproportionate percentage of points to nutrition information), and by cognitive consistency pressures (prompting subjects to reflect the rank order of elicitation in their points allocations). With these reservations, the correlation statistic does suggest that, among consumers accustomed to making brand choice decisions within a particular product category, the rank order of elicitation may be an appropriate surrogate for relative importance.

Determinance Measure. In addition to distinguishing those attribute dimensions sufficiently determinant to warrant consultation from those which were not consulted, a method of identifying the relative determinance of those attribute dimensions which were con-suited was developed.

The determinance of an attribute dimension was operationally defined as the degree to which its elicitation caused a reduction in the number of brands under consideration (measured in terms of the number of brands on which the next attribute dimension was elicited). A feasibility test had indicated that most subjects could be expected to "pyramid" towards their final brand selections in this manner. Since subjects were constrained to receive information on all six brands for the first attribute dimension, a mean determinance score could he calculated for each attribute dimension by treatment group on the basis of the average number of brands on which the second attribute dimension was elicited. Thus, the lower the mean score, the more deterministic the attribute dimension in question. In those cases where the first attribute dimension selected was the only one selected prior to a brand choice being made, a determinance score of zero was recorded. The following tabulation presents mean determinance scores by information dimension and by treatment group:

TABLE

The results indicate that for subjects in the second treatment group (representing the repeat purchase condition), physical sample was the most determinant attribute dimension, in part reflecting the fact that 42 percent of subjects who selected this attribute dimension first did not seek any further information. Subjects who selected this attribute dimension first selected information on the second attribute dimension for an average of only 1.472 brands, compared to 2.357 brands when nutrition information was selected first (t statistic for difference between means significant at 0.05 level).

Two reservations regarding this determinance measure should be noted. First, the degree to which consultations of attribute dimensions resulted in reductions in the number of brands under consideration when the attribute dimensions were consulted other than first in order of elicitation were not considered. Differences in the nature of information previously consulted would have made it impossible to compare subjects on a common basis. Nevertheless, the determinance measure was deficient in that it was based on only a portion of the data available. Secondly, the determinance measure assumes that the pattern of processing (in particular, the number of brands under consideration) is solely a function of previously acquired attribute information. Predisposition regarding search time and number of information units to be acquired, given the nature of the decision, probably vary across individuals. Such individual differences have been considered randomly distributed throughout the sample.

TABLE 1

INFORMATION CUE ELICITATION OCCASIONS BY SELECTION SEQUENCE POSITION AND BY TREATMENT GROUP

SUMMARY AND IMPLICATIONS

Although information display exercises are behavioral in nature, artificialities clearly exist in the manner in which the information is presented and in the procedural constraints which are imposed upon subjects in order to facilitate subsequent analysis. The impact of these artificialities on external validity requires further investigation. In addition, the information display approach offers no advance over other techniques in the area of specification of the attribute dimensions to be included.

However, the use of information displays to measure relative attribute importance offers two principal advantages over traditional attitudinal measures. First, the behavioral approach permits motivational realism to be incorporated into the data collection process, so that the propensity of subjects to provide what they perceive to be socially acceptable response patterns may be reduced.

Secondly, by permitting each subject to determine how many information units are to be acquired prior to a brand selection decision, information display exercises provide data which enable the researcher to distinguish those attribute dimensions perceived by subjects to be sufficiently important to warrant investigation from those not so perceived. The distinction cannot be drawn among a set of attribute dimensions which are rank ordered directly or according to attitudinal importance weights. In addition, information display exercises permit differences in the relative importance of attribute dimensions under varying conditions of information availability to be identified.

The relevance to policy makers of whether attribute information available to consumers is acquired has been emphasized by Jacoby, Chestnut, and Silberman (1977), and by Scammon (1977). The question is also of significance to marketers who must ensure that their products are superior on those attribute dimensions which target consumers consider sufficiently important to warrant inter-brand comparisons in their decision-making processes.

REFERENCES

David C. Arch, James R. Bettman, and Pradeep Kakkar (1978), "Subjects' Information Processing in Information Display Board Studies", in H. Keith Hunt, ed., Advances in Consumer Research, Vol. 5, Association for Consumer Research, 555-560.

James Bettman, and Jacob Jacoby (1976), "Patterns of Processing in Consumer Information Acquisition', in Beverlee B. Anderson, ed., Advances in Consumer Research, Vol. 3, Association for Consumer Research, 315-320.

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

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 H. Keith Hunt, ed., Advances in Consumer Research, Vol. 5, Association for Consumer Research, 93-98.

Jacob Jacoby, R. W. Chestnut, and William Silberman (1977), "Consumer Use and Comprehension of Nutrition Information", Journal of Consumer Research, 4, 119-128.

Jacob Jacoby, R. W. Chestnut, Karl C. Weigl, and William A. Fisher (1976), "Pre-Purchase Information Acquisition: Description of a Process Methodology, Research Paradigm, and Pilot Investigation", in Beverlee B. Anderson, ed., Advances in Consumer Research, Vol. 3, Association for Consumer Research, 306-314.

James H. Myers, and Mark I. Alpert (1968), "Determinant Buying Attitudes: Meaning and Measurement", Journal of Marketing, 32, 13-20.

James H. Myers, and Mark I. Alpert (1977), "Semantic Confusion in Attitude Research: Salience vs. Importance vs. Determinance", in William D. Perrault Jr., ed., Advances in Consumer Research, Vol. 4, Association for Consumer Research.

M. Reilly, R. Holman and R. Evered (1976), "Individual Differences in Information Processing: An Exploratory Report", in Proceedings, 8th Annual Conference, American Institute of Decision Sciences.

Edward Russo, and Larry D. Rosen (1975), "An Eye Fixation Analysis of Multi-Alternative Choice", Memory and Cognition, 3, 267-276.

Debra L. Scammon (1977), "Information Load and Consumers'', Journal of Consumer Research, 4, 148-155.

William L. Wilkie, and Rolf P. Weinreich (1972), "Effects of the Number and Types of Attributes Included in an Attitude Model: More is Not Better", in M. Venkatesan, ed., Proceedings, Third Annual Conference, Association for Consumer Research, 325-340.

Peter L. Wright (1972), "Consumer Judgement Strategies - Beyond the Compensatory Assumption", in M. Venkatesan, ed., Proceedings, Third Annual Conference, Association for Consumer Research, 316-324.

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