Frequency Information As a Dimension of Consumer Knowledge

ABSTRACT - An often ignored aspect of knowledge - knowledge about event frequency -- is discussed in terms of consumer decision making. A brief description of frequency theory is provided and some speculations about how frequency knowledge might affect judgment and choice are offered. Finally, a preliminary experiment is presented which demonstrates that frequency knowledge can dominate more important information when consumers form price images of retail stores.

Citation:

Joseph W. Alba and Howard Marmorstein (1986) ,"Frequency Information As a Dimension of Consumer Knowledge", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 446-449.

Advances in Consumer Research Volume 13, 1986      Pages 446-449

FREQUENCY INFORMATION AS A DIMENSION OF CONSUMER KNOWLEDGE

Joseph W. Alba, University of Florida

Howard Marmorstein, University of Florida

ABSTRACT -

An often ignored aspect of knowledge - knowledge about event frequency -- is discussed in terms of consumer decision making. A brief description of frequency theory is provided and some speculations about how frequency knowledge might affect judgment and choice are offered. Finally, a preliminary experiment is presented which demonstrates that frequency knowledge can dominate more important information when consumers form price images of retail stores.

INTRODUCTION

Until recently, researchers had been more concerned with the effects of consumer knowledge on consumer behavior than on the nature of consumer knowledge itself. Though such an orientation directly addresses the questions that are most relevant to real-world behavior, it carries with it the inherent danger of fostering simplistic conceptualizations and measures of consumer knowledge (e.g., product familiarity, product ownership). Not surprisingly, attempts to explore the nature of consumer knowledge have revealed it to be complex and multidimensional (Alba and Hutchinson 1985; Kanwar, Olson and Sims 1981). Moreover, recent research that has incorporated a multidimensional perspective has shown that the effects of consumer knowledge on consumer behavior are a function of how that knowledge is operationalized (Brucks 1985; Brucks, Mitchell and Staelin 1984; Punj and Staelin 1983).

An analogous situation exists with regard to the learning of new information. That is, traditional measures of consumer learning constitute a fairly narrow set. Though numerous variations exist, most measures consist of some form of aided or unaided recall. The focus of these recall tests often is on memory for the substantive facts presented in a promotional message or information display board. In many cases, these measures are the most appropriate and informative ones available. Judgment and choice often are based on the specific faces one can recall about alternative brands, and recall and recognition of brand-attribute facts represent reasonable measures of what a consumer has learned about them. However, problems lie in the exhaustiveness and sensitivity of these measures.

In this paper we wish to make the point that there are aspects of knowledge that contain little semantic detail about the characteristics of a brand or the content of a message, but nonetheless contain information relevant to consumers' decisions. We will discuss one such aspect, namely, the knowledge consumers have about brand and attribute frequency.

Frequency Knowledge

The notion that people are attentive to the frequency with which events occur in the world is not new. There now exists a considerable amount of research demonstrating that people are highly sensitive to the frequency with which even very mundane and innocuous events occur, both in real-world and laboratory settings. Moreover, this research has identified some peculiar characteristics of frequency knowledge which suggest that the learning of frequency information is an automatic process. Within the context of the Hasher and Zacks (1979, 1984) theory of automaticity, this implies that the process of encoding frequency information (a) operates continually, (b) cannot be improved by practice, (c) cannot be willfully inhibited, (d) does not require conscious awareness, and (e) drains minimal amounts of cognitive resources. As Hasher and Zacks report, empirical support for each of these assertions has been found. It has also been shown that, tue to its unique mental representation, once frequency information is encoded it may be less susceptible to interference than other types of knowledge (Hintzman, Nozawa and Irmscher 1982). These traits suggest that frequency knowledge may represent a particularly important aspect of memory when information is presented under conditions characterized by low involvement or high information load.

Frequency knowledge is important also from a measurement perspective. As noted, people may learn and remember information concerning the frequency of events without being able to recall the exact nature of those events (cf. Pitz 1976). Inasmuch as one major goal of marketing research is to assess what consumers know about products, a significant underestimation of consumer knowledge may obtain if a researcher measures only recall for specific brand and attribute information. Sounding a similar theme, Singh and Rothschild (1983) have argued that simple recall measures understate memory for an advertisement; they call instead for greater use of recognition measures. It is our contention that tests of frequency memory may tap an additional and unique aspect of knowledge.

What Might Consumers Count?

Assuming that people are sensitive to event frequency, the next issue concerns the nature of an event. That is, what can and what do people count? Below, we speculate about events that are pertinent to consumer behavior.

Individual Objects. The majority of research demonstrating frequency knowledge has dealt with the frequency of single sterile events such as individual words in a longer list of words. Typically, subjects are shown a word list, in which some words are repeated various numbers of times, and are asked to estimate the number of times each critical word appeared. However, extensions of this paradigm to less sterile situations have proven successful. For example, people's sensitivity to the true relative frequencies of real-world occupations and lethal events has been shown to be surprisingly accurate (Lichtenstein, Slovic, Fischhoff, Layman, and Combs 1978).

In a marketing context, there are numerous events that, if counted, could influence perceptions of a product, even if the details of the produce cannot be retained. For example, consumers may be sensitive to the relative frequencies of different brands they observe. Such knowledge might lead to beliefs about the relative popularity of different market offerings. These beliefs about market share, in turn, might lead to further beliefs concerning quality (cf. Duncan and Olshavsky 1982). And, for consumers who have a high need for conformity, frequency knowledge might lead directly to purchase.

Similarly, consumers may have an implicit understanding of the frequency with which advertisements for different brands appear. Such knowledge could lead to inferences about market share, company size, and company stability. Moreover, if research can establish the accuracy of such knowledge, frequency estimation could be used as a direct measure of advertising exposure and impact.

Frequency counting, however, may not be limited to mere objects or events such as brands and advertisements. It may also apply to attributes associated with those objects or events. For example, a consumer's price image of a store might be influenced by the consumer's perception of the number of sale items and/or sale periods promoted by that store. The consumer's rankings of competing stores along a price dimension might then be determined by inter-store comparisons of the aforementioned frequencies. Such impressions may be formed without effort or conscious counting and may exist independently of the details of each store's prices.

This is not to say that details cannot be counted. Attributes are objects and, as such, may be subject to counting . Consider a recent study by Malmi and Samson (1983). They presented to subJects a range of numbers and asked for an estimate of the mean value. Results showed remarkable accuracy regardless of the skewness and node of the distribution of the numbers. Further investigation suggested that subjects had constructed in memory a frequency distribution and had derived the mean by estimating the balance point. Generalizing this finding to knowledge of prices, it may be the case that, without conscious intention to do so, consumers learn the price distribution of a brand across stores or over time; the price distribution of different products within a store; or the price distribution of all brands within a product category. Distribution information could then lead to beliefs about the central tendency of a category and the deviation of particular brands.

Semantic Categories. In addition to a sensitivity to the frequency and distribution of individual items within a category, a small amount of evidence suggests that people directly tally the number of times a category has been encountered when they have been exposed only to exemplars of that category (Alba, Chromiak, Hasher, and Attig 1980; Barsalou and Ross, forthcoming). For example, after being presented with particular names of animals, trees, etc., people can state with accuracy the number of times each category was encountered without appealing to memory for the individual instances. Though we would expect this result to generalize fairly well to product categories, a more interesting but unexplored question concerns the case of attribute categories. For example, the abstract category of "comfort" in the context of an automobile can be instantiated by such attributes as bucket seats, plush carpeting, leg room, and so on. When presented with such features, to consumers automatically encode and tally them as comfort features? The answer may depend on the degree of association between the features and their category. To the extent that consumers do tally such information, they may automatically form impressions of how a product rates on its major dimensions. And, such knowledge may endure while specific information decays.

Evaluative Categories. With respect to decision making, one of the most important types of frequency counts a consumer can make involves evaluative categories. Here the issue is whether consumers tally the number of desirable and undesirable attributes associated with a product, irrespective of the exact meaning of those attributes. Research on this question is sparse, but at least one set of studies suggests that consumers may be sensitive to such information (Alba 1985). More importantly, there is a recent but growing body of evidence demonstrating that, once encoded, such information can drive judgment and choice even in the presence of more relevant and important information.

Consider, for example, research on message persuasiveness. Chaiken (1980) has demonstrated that when subjects are highly involved and the message arguments are relevant, persuasion is positively related to the number of arguments favoring a particular position (cf. Asam and Bucklin 1973). This finding by itself is not unexpected, inasmuch as message recipients should be more persuaded by many good arguments than by a few good arguments. However, Petty and Cacioppo (1984) demonstrated that argument number can dominate argument importance if subject involvement is sufficiently low. That is, many irrelevant arguments may be more persuasive than a few relevant ones if people do not process the information deeply. Thus, argument frequency may be considered a peripheral route to persuasion.

The importance of frequency has also been shown in choice contexts, as first suggested by Bettman and Park (1980). In their protocol analyses they noted that some subjects were prone to use a "counting" strategy, by which decisions were based in part on the number of good and bad attributes associated with each brand. Similarly, Russo and Dosher (1983) showed that binary choice decisions are sometimes based on a "majority-of-confirming-dimensions" heuristic. People who employ this heuristic choose the alternative that dominates the other one on the greater number of dimensions, sometimes at the expense of dimensional importance and the size of the interbrand differences on those dimensions. Further, the likelihood of adopting this heuristic seems to increase as the number of dimensions increases, and may also be related directly to the involvement and fatigue levels of the subjects.

Finally, consider the role of evaluative frequency in predictive judgment. The judgments consumers make about a product often involve predictions of it efficacy. Evidence suggests that when previous judgments have not been made and the outcome is uncertain, predictions about the likely occurrence of an event may be based partially on a comparison of the sheer number of reasons that can be generated for and against its occurrence (Hoch 1984).

Moreover, when a prior history of success and failure exists, it appears that prior event frequency greatly influences perceptions of future event probability (Estes 1976). That is, when making judgments of probability, people rely on the frequency rather than the proportion of times an event has occurred in the past. Though investigated in other domains, Estes notes the applicability of this finding to consumer situations. For example, when evaluating the preferability of two medicinal remedies, consumers may choose the one that has worked most often in the past, even though the two brands may have had an unequal trial rate. Estes suggests that the brand with the most successes will be chosen, even if its proportion of successes is lower than that of the competing brand.

EXPERIMENT

Though we believe that the encoding and use of frequency information is prevalent, much of what we have said heretofore with respect to consumer behavior has been speculative. Below, we present a preliminary study of the effect of frequency information on decision making. In particular, we examine evaluative category frequency in the context of memory-based consumer decisions. Further, we examine judgments that might be highly influenced by frequency information, i.e., judgments consumers make about the relative expansiveness of competing retail stores.

Briefly, subjects were presented with a list of grocery items along with the price of each item at several hypothetical stores. The critical comparison involved two of the stores, one of which was more expensive overall but cheaper on a greater number of items than the other one. To the extent that consumers use a frequency heuristic, they should describe the more expensive store as being the less expensive of the two.

Method

Stimulus. The stimulus was a two-page price advertisement which compared three supermarkets (A, 8, and C) on 60 items. The price of each item at Store A was always nine cents lower than the price at the cheaper of the remaining two stores.

Of focal interest were the differences between Stores B and C. In 36 of the 60 cases (i e., 60% of the items), Store C was cheaper than Store B. The mean advantage of Store C on these items was eight cents. These price differences were distributed approximately symmetrically and varied from two cents to 16 cents. Store B was less expensive than Store C on the remaining 24 items. However, the mean difference in these instances was 33 cents. Moreover, when Store B was the cheaper of the two, the price difference ranged from 28 to 38 cents. Thus, the two distributions of price differences were nonoverlapping, making the differences maximally observable

Overall, the 60 items cost approximately $5 (or about 5%) more at Store C than at Store B, but it was more frequently the case that Store C was cheaper than Store B. The items were arranged in the list such that within each group of fifteen, Store C was cheaper on a random subset of nine of them. The specific price difference attached to each item was also randomly determined.

Procedure. Subjects were asked to imagine that they had recently moved to a new city and that price was important when choosing a supermarket. They were provided with the comparative ad and were told that they would have 90 seconds to examine each page of it. The experimenter emphasized that they should examine the information about all three stores carefully in preparation for some questions about the stores' prices.

At the end of the 3-minute study phase, subjects were asked to rank the stores in order of their relative expansiveness. They also were asked to estimate the total cost of the 60 items at each store. Finally, they were asked to explain how they had decided whether Store B or Store C was the more expensive of the two. At no time during this test phase were subjects able to reexamine the stimulus

Subjects. A group of 30 undergraduate marketing students volunteered to participate in the study. Two of them failed to follow instructions and were eliminated from the analysis.

Results

Normatively, all subjects should identify Store B as being less expensive than Store C. Note that a correct response could result from either an accurate tally of the total price of each store or from reliance on the nearly four-fold difference in the average price difference between Stores B and C

To the contrary, a significant proportion of subjects behaved in a counter-normative fashion. That is, 57% of the subjects identified Store C as the less expensive store. This figure is significantly different from zero (p<.001). Because subjects were almost equally split, the mean estimates of the total cost of items at each store did not differ, though both differed from the estimate for Store A.

It could be argued that subjects randomly chose between Stores B and C, resulting in a significant amount of counter-normative behavior. However, an examination of the reasons subjects gave for their choices does not support such a claim. Subjects were classified according to whether or not they employed a frequency heuristic. Of the 16 subjects who made the counter-normative choice, 10 clearly relied on frequency information; the remaining subjects did not articulate their rationale in a clear enough way to allow classification. Of the 12 subjects who correctly chose Store B, only 4 made reference to frequency. These 4 subjects either confused Stores B and C, leading us to underestimate significantly the effect frequency had in this study, or they used frequency as a rationalization for their choices. It is not possible to identify the true explanation at this time. Regardless, the results indicate that frequency information was a significant factor in the decisions made by a number of our subjects.

DISCUSSION

The purpose of this paper has been to emphasize an aspect of knowledge that has been virtually ignored in consumer research. It is our contention that under many circumstances, particularly those that adversely affect information processing such as low involvement and high information load, knowledge of event frequency will play a significant role in consumer decision making. In cases in which other information is lacking, frequency knowledge will enable consumers to make decisions that are not completely arbitrary. In other cases, as in the present experiment or when advertisers intentionally manipulate frequency to overstate the advantages of their brand, frequency information may lead to suboptimal decisions.

REFERENCES

Alba, Joseph W. (1985), "The Learning of Frequency Information and Its Use as a Decision Heuristic," working paper, University of Florida. Gainesville. FL.

Alba, Joseph W., Walter Chromiak, Lynn Hasher, and Mary S. Attig (1980), "Automatic Encoding of Category Size Information," Journal of Experimental Psychology: Human Learning and Memory, 6(July), 370-378.

Alba, Joseph W. and J. Wesley Hutchinson (1985), "A Framework for Understanding Consumer Knowledge," working paper, University of Florida, Gainesville, FL.

Asam, Edward H. and Louis P. Bucklin (1973), "Nutrition Labeling for Canned Goods: A Study of Consumer Response," Journal of Marketing, 37(April), 32-3?.

Barsalou, Lawrence W. and Brian E. Ross (forthcoming), "The Roles of Automatic and Strategic Processing in Sensitivity to Superordinate and Property Frequency," Journal of Experimental Psychology: Learning, Memory, and Cognition.

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

Brucks, Merrie (1984), "The Effects of Product Class Knowledge on Information Search Behavior," Journal of Consumer Research, 12(June), 1-16.

Brucks, Merrie, Andrew A. Mitchell, and Richard Staelin (1984), "The Effect of Nutritional Information Disclosure in Advertising: An Information Processing Approach," Journal of Public Policy & Marketing, 3, 1-25.

Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 5(May), 752-766.

Duncan, Calvin P. and Richard W. Olshavsky (1982), "External Search: The Role of Consumer Beliefs," Journal of Marketing Research, 19(February), 32-43.

Estes, W. K. (1976), "The Cognitive Side of Probability Learning," Psychological Review, 83(January), 37-64.

Hasher, Lynn and Rose T. Zacks (1979), "Automatic and Effortful Processes in Memory," Journal of Experimental Psychology: General, 108(September), 356-388.

Hasher, Lynn and Rose T. Zacks (1984), "Automatic Processing of Fundamental Information," American Psychologist, 39(December). 1372-1388.

Hintzman, Douglas L., Georgie Nozawa, and Mark Irmscher (1982), "Frequency as a Nonpropositional Attribute of Memory," Journal of Verbal Learning and Verbal Behavior, 21(April), 127-141.

Hoch, Stephen J. (1984), "Availability and Interference in Predictive Judgment," Journal of Experimental Psychology: Learning, Memory, and Cognition, 10(October), 649-662.

Kanwar, Rajesh, Jerry C. Olson, and Laura S. Sims (1981), "Toward Conceptualizing and Measuring Cognitive Structures," in Advances in Consumer Research, Vol. 8, ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 122-12?.

Lichtenstein, Sarah, Paul Slovic, Baruch Fischhoff, Mark Layman, and Barbara Combs (1978), "Judged Frequency of Lethal Events," Journal of Experimental Psychology: Human Learning and Memory, 4(November), 551-581.

Malmi, Robert A. and David J Samson (1983), "Intuitive Averaging of Categorized Numerical Stimuli," Journal of Verbal Learning and Verbal Behavior, 22(October), 547-559.

Petty, Richard E. and John T. Cacioppo (1984), "The Effects of Involvement on Responses to Argument Quantity and Quality: Central and Peripheral Routes to Persuasion," Journal of Personality and Social Psychology, 46(January), 69-81.

Pitz, Gordon F. (1976), "Memory and Frequency Estimation Processes in a Decision Task," Memo n & Cognition, 4(March). 132-138.

Punj, Girish N. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," Journal of Consumer Research, 9(March), 336-380.

Russo, J. Edward and Barbara Anne Dosher (1983), "Strategies for Multiattribute Binary Choice," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9(October), 676-696.

Singh, Surendra N. and Michael L. Rothschild (1983), "Recognition as a Measure of Learning from Television Commercials," Journal of Marketing Research, 20(August), 235-248.

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