Sequential decision-making strategies of expert and novice consumers.
In stark contrast to the tremendous amount of theoretical work that has been conducted in the field of decision making, very little is known about how consumers actually make decisions. Given the current state of knowledge, it is practically impossible to predict which brand will be chosen by a consumer based on the amount of information available to that consumer. The literature of information search and acquisition is mostly normative (i.e., prescriptive): Additional information should be sought as long as the benefits of the information acquired outweigh the cost of its acquisition (e.g., Stigler, 1961). Behavioral research, which is more descriptive in nature, is more concerned with the question: How do people actually make decisions? Indeed, consumers seldom consider all relevant information prior to making a purchase. Instead, they use various strategies to decide when to stop acquiring product information and to commit to one brand.
Payne (1982), for example, observed that consumers' choice processes are largely contingent upon the complexity of the task (i.e., how many brands and/or features are being considered), with different strategies being used for more complex decisions than for simpler ones. Consumers, however, can acquire information in many different ways. One way is sequentially, one piece of information at a time, one piece after another. How many pieces of information do consumers typically consider? Five? Ten? When consumers evaluate product information sequentially, the essential decision is not so much which alternative to choose, but when to stop acquiring information (Saad & Russo, 1996).
Different researchers have developed competing hypotheses to explain how consumers acquire and use information sequentially. One of them is the Core Attributes (CA) heuristic. According to the CA heuristic, consumers acquire and consider a pre-determined set of core attributes about several brands, then choose the brand that is ahead at that point (Busemeyer & Townsend, 1993). Another hypothesis is the Difference Model (DM), in which consumers stop looking after they reach a desired level of differentiation between two brands being compared (Aschenbrenner et al., 1984).
Saad and Russo (1996) found that both approaches were used, albeit by different decision makers. The crucial factor in determining when decision makers make use of the CA heuristic is whether they have control over the order in which information is acquired. Saad and Russo also found that decision makers utilizing a DM approach had stopping criteria which did not remain fixed. That is, the threshold for the degree of differentiation required for choosing between two competing brands was lowered as the number of attributes evaluated increased. Said a different way, smaller actual differences in the alternatives counted for more subjective differences as more and more information was considered without a clear winner appearing. So a difference between two alternatives that might not be sufficient to indicate a clear winner when just a few pieces of information had been considered, might be sufficient later in the process after a great deal of information had been considered and the alternatives appeared to be tied. That is because for the decision maker, as Russo and Saad (1999) point out, the choice process may change from, Which alternative is best? to, Can I stop acquiring information now and commit to the alternative that is currently leading?
These findings, however, are likely to be moderated by decision maker expertise. In general, experts are known to differ from novices in their decision-making ability. However, the question of whether they always differ in their use of information when making a purchase decision remains unanswered, especially for information acquired sequentially. Shanteau (1992), for example, noted that there is no difference in information usage between experts and novices. Devine & Kozlowski (1995), on the other hand, found that experts use more information, particularly in unstructured tasks.
The current study investigates the effect of product familiarity, or experience, on the sequential choice process. This is done by taking into consideration the characteristics of the consumer when making a purchase that is either complex, expensive, or otherwise important. If he or she is an "experienced" buyer, then that consumer may feel more comfortable acquiring less (but more useful) information than an inexperienced buyer. That suggests that, first of all, experienced (expert) consumers will acquire less information overall than novice consumers when they are engaged in a fairly structured task. Second, if the experienced consumer makes his or her choice only after finding a suitable difference (DM) between two brands, will that difference be greater or lesser than for the less-experienced buyer? Said another way, when experts utilize a DM strategy, will their required difference-thresholds decrease at a faster rate?
Thirty-five students recruited from an introductory marketing course at a medium-size West Coast state university agreed to participate in return for extra course credit. Participants were exposed to a well-structured decision problem in which they were directed to choose between two brands of compact-35mm cameras. The choice task was presented on a personal computer using a special purpose-built software program described below. The program recorded the activity of the subject in the task, which allowed the data to be retrieved later by the researchers. A moderately expensive, durable consumer product familiar to typical college students was used in the study to ensure a reasonably high level of involvement in the choice process. Compact cameras were selected from several products identified in pre-tests as potentially appropriate (price range: $25-$300).
The subjects' initial knowledge of the product was assessed by their performance on a short multiple-choice test. The questions were selected on the basis of earlier pre-tests. An example of a question from the knowledge screening test is: "The camera lens providing an image most nearly like that of the human eye is (choose one): 25mm, 35mm, 40mm, 50mm, 75mm?" Scores on the objective test were used to screen for low and high levels of knowledge. Those scoring low on the test were assigned to the "novice" condition (N = 29); those scoring high were assigned to the "expert" condition (N = 6). The difference between the numbers of expert and novice subjects reflects the difficulty of finding highly-knowledgeable participants. The existing gap in knowledge was subsequently widened by having participants in the expert condition study packets of information relevant to the product category (cameras), and having participants in the novice condition study packets of information relevant to another product category (bicycles).
During the experiment, subjects met in a dedicated computer lab equipped with 20 Macintosh computers running the Sequential Multi-Attribute Choice (SMAC) program (Saad 1996). First, an experimenter described the task, then provided a demonstration of the DSMAC program using a choice situation familiar to participants but different from the experimental choice task. The demonstration lasted about 15 minutes, during which the experimenter actively involved all or most of the participants in the demonstration task while explaining the features of the program. Questions were encouraged.
Second, subjects were directed to begin DSMAC on the computer in front of them, which began by asking them first to perform a Q-sort of the 25 camera features, then to assign weights to each on a scale having a range of 1-100, with 100 being the most important. Following the Q-sort, subjects engaged in 15 binary choice trials, one after the other, for 15 different pairs of cameras. At the end of each trial they indicated whether they preferred the first camera, the second, or neither.
In each of the 15 choice trials they were presented with, subjects acquired attribute information from a list of the 25 product features by clicking a mouse on the desired feature. By doing so, subjects were able to view attribute information for both alternatives in the "Information Integration" screen. In this window, they used the mouse to move a pointer on a scale (the "Cumulative Confidence Measure"), which ranged from "100 percent confidence in alternative A" to "100 percent confidence in alternative B," to indicate the likelihood of their choosing either alternative A or B at that time if no other attribute information were considered. The mid-point of the scale was "50 percent" likelihood.
Three options were available to subjects after viewing attribute information: (a) choose brand A or B; (b) choose neither; (c) acquire additional attribute information. If either of the first two options was selected, a new binary choice for two different brand was presented. If subjects elected to acquire additional information, then the list of 25 product features was again presented. The procedure described above was repeated until subjects indicated that they chose either brand A, brand B, or neither.
ANALYSIS AND RESULTS
First, how many pieces of information (attributes) were acquired by experts was compared to how many pieces were acquired by novices? Second, their stopping policies were examined in light of the difference Model. The hypothesis that experts acquire less information was tested by counting the number of attributes acquired by the two groups. Out of 25 available, the mean number of attributes acquired by experts was 13.60; for novices, the mean was 11.06. Experts, then, actually acquired slightly more attributes than novices, though the difference was not statistically reliable (t-test n.s.).
The hypothesis that when experts utilize a DM stopping strategy their difference-thresholds decrease at a faster rate (i.e., the required level of differentiation diminishes as the number of attributes acquired increases) was tested by comparing the levels of confidence of the decision makers with the number of attributes they acquired in each trial. Using a DM strategy implies that consumers stop searching once they reach a desired level of differentiation between brands. As in Saad & Russo (1996), this study uses the subject's stated degree of confidence (50-to-100% range) to represent the extent of the difference between the two brands. To test that this suitable differentiation threshold drops at a faster rate for experts than for novices, the relationship between level of confidence (the dependent variable) and the number of attributes examined (the independent variable) was estimated using non-linear regression modeling. If the level of differentiation diminishes as the amount of acquired information increases, then the model's beta is greater than 0. Larger values of beta indicate a faster decay of the difference-threshold (i.e., lower required levels of differentiation for any number of attributes examined). As predicted, the threshold decayed more rapidly for experts than for novices (mean beta = 0.228 vs. 0.197, p = .07). This means, first of all, that experts (and novices) accept a lower level of differentiation (i.e., a lower threshold) between two alternatives when they have acquired and considered many attributes than when they have considered fewer. Second, it means that, compared to novices, the rate at which the threshold falls is greater for experts. The implication is that expert consumers need less information than novices to distinguish the superior brand when it is closely tied with another brand.
The current study addresses a controversy that exists in the field today about whether expert consumers acquire more purchase information than novices. In a structured, repeated sequential choice task, expert consumers were found to acquire about the same amount of information as novice consumers. Because of the relatively small number of participants in the study, and because of the nature of the population from which the sample was drawn (undergraduate business students), readers are cautioned against over-generalizing the results of the study. This study can be regarded as a pilot study of how some knowledgeable consumers make decisions for a single product category that raises some interesting research questions. Clearly more studies are warranted using other product categories, at the very least. It would also be worthwhile conducting a similar study using "real-life" experts (e.g., members of a camera club, or professional photographers). Indeed, such a study would likely find differences between experts and novices that were even more pronounced. In contrast, this study found differences in search behavior between consumers who had relatively small differences in knowledge, as measured by a short questionnaire, and slightly amplified by the study of product-category information.
In summary, some differences in stopping strategies were found between experts and novices. Although experts acquired about the same amount of information as novices, they were found to more easily break ties between alternatives, and to do so with less information than novices.
Aschenbrenner, K.M., Albert, D. & Schmalhofer, F. (1984). Stochastic Choice Heuristics. Acta Psychlogica, 56, 153-166.
Busemeyer, J.R. & Townsend, J.T. (1993). Decision Field Theory: A Dynamic Cognitive Approach. Psychological Review, 100, 432-459.
Devine, D.J. & Kozlowski, S.W.J. (1995). Domain-specific knowledge and task characteristics in decision making. Organizational Behavior and Human Decision Processes, 64(3), 294-306.
Payne, J. (1982). Contingent Decision Behavior: A Review and Discussion of Issues. Psychological Bulletin, 92, 382-402.
Russo, J. E. & Saad, G. (1999). Consumer choice as a process of discrimnination. Proceedings of the Administrative Science Asociation of Canada, Saint John, NB, Canada.
Saad, G. & Russo, J.E. (1996). Stopping criteria in sequential choice. Organizational Behavior and Human Decision Processes, 67(3), 258-277.
Saad, G. (1996). SMAC: An interface for investigating sequential multiattribute choices. Behavior Research Methods, Instruments, and Computers, 28, 259-264.
Shanteau, J. (1992). How much information does an expert use? Is it relevant? Acta Psychologica, 81, 75-86.
Stigler, G.J. (1961). The Economics of Information. Journal of Political Economy, 69, 213-225.
Wald, A. (1947). Sequential Analysis. New York: Wiley.
NOTE: This research was funded in part by a faculty research grant from Central Washington University.
Peter Boyle, Central Washington University
Gad Saad, Concordia University
|Printer friendly Cite/link Email Feedback|
|Author:||Boyle, Peter; Saad, Gad|
|Publication:||Academy of Marketing Studies Journal|
|Date:||Jul 1, 2000|
|Previous Article:||Electronic commerce curricula: an overview of current electronic commerce courses and implications for marketing education.|
|Next Article:||Advertising effects on children's buying habits in the U.S.|