Prior knowledge and complacency in new product learning."

Our research examines the role of prior knowledge in learning new product information. Three studies demonstrate that, compared to consumers with lower prior knowledge, those with higher prior knowledge learn less about a new product. Further, higher knowledge consumers are able to learn more but learn less due to motivational deficits; inferior learning of new product information by those with higher prior knowledge is caused by inattention at encoding rather than reconstructive errors at retrieval. These results hold both when prior knowledge is manipulated experimentally (studies 1 and 2) and when it is an individual difference factor (study 3). M ost practitioners see consumer knowledge as an advantage, targeting many new products at expert heavy users. This strategy seems intuitively appealing when based on the assumption that experts have a learning or information processing advantage, proportionately higher levels of interest or involvement, and a greater likelihood of opinion-leadership. As Rogers (1995, p. 166) states, "When an adequate level of how-to knowledge is not obtained prior to the trial and adoption of an innovation, rejection and discontinuance are likely to result. To date, few diffusion investigations are available that deal with how-to knowledge." But are those with higher prior knowledge better able to learn about a new product offering? Fifty years of expertise research have culminated in two conflicting pictures Our research examines the role of prior knowledge in learning about new products in situations where new information makes existing product knowledge obsolete. We posit that, compared to consumers with lower prior knowledge, those with higher prior knowledge may learn less about the new *Stacy L. Wood is assistant professor of marketing, University of South Carolina, Columbia, SC 29208 (e-mail: wood@moore.sc.edu). John G. Lynch, Jr., is Hanes Corporation Foundation Professor of Business Administration, Duke University, Box 90120, Durham, NC 27708-0120 (email: john.lynch@duke.edu). The authors thank Joe Alba, Bill Bearden, Jim Bettman, Lisa Bolton, Wes Hutchinson, Randy Rose, the editor, associate editor, and three reviewers for helpful comments, and Scott Swain and Danny Wadden for exemplary research assistance. This research was partially funded by the Moore School of Business and the Fuqua School of Business. product. More important, we present evidence that this inferior learning is due to motivation at encoding rather than to retrieval errors. Those with higher prior knowledge incorrectly generalize from knowledge of existing products and assume that they already know how to use the new product properly. With the presence of certain cues at encoding, those with higher prior knowledge learn more. We demonstrate this result both when prior knowledge is manipulated experimentally and when it is a measured individual difference factor. There are almost as many definitions of "expertise" as researchers who study it (Shanteau 1992). Similar to Spence and Brucks (1997), we define degree of expertise as a function of the amount of domain-specific knowledge acquired through experience or training. This definition is not materially different from the concept of prior knowledge (PK). Thus, we first test our hypotheses by comparing consumers with experimentally induced levels of PK to avoid confounding with correlated constructs of involvement or self-perception of goals. We then replicate these results when real prior experience is measured, allowing us to tie our findings back to experience-based definitions in the expertise literature (e.g., Alba and Hutchinson 1987). HYPOTHESIS DEVELOPMENT Advantages of High Prior Knowledge Cognitive science provides many examples of advantage in learning due to high PK Disadvantages of High Prior Knowledge Experts often fail to perform in accord with these processoriented advantages Overconfidence is a prevalent bias (Fischoff, Slovic, and Lichtenstein 1977); typically people assume that they know more than they do (e.g., Moorman 1999). One might expect that consumers with higher PK would be more overconfident (cf. Keren 1987). Repeated problem-solution patterns facilitate the formation of possibly inappropriate inference heuristics, which can subsequently lead to systematic biases (Kahneman, Slovic, and Tversky 1982; The feeling-of-knowing phenomenon (Hart 1965) provides a further reason to expect poor performance by experts. Feelingof-knowing is a metacognitive preretrieval process in which one assesses one's memory for a memory Thus, overconfidence, use of heuristics, or FOK effects may cause knowledgeable consumers to inappropriately rely on self-generated inferences. Poor performance in this context could arise due to inference making at encoding of new information or at retrieval. For example, overconfidence might cause encoding errors due to superficial processing of new information or cause retrieval errors based on insufficient effort to retrieve new product information. Prior Knowledge Effects and New Product Innovation Little expertise research has examined reactions to product innovations. Will those with prior product category information be better able to learn how to use new products? Research in other domains has shown that expert superiority in learning or problem solving is strongly impacted by the external characteristics of the given task (e.g., Shanteau 1992). With a cognitive science approach, one might expect that consumers with a high degree of product category knowledge would be best able to learn about and use new products in that category. Behavioral decision researchers have reported results that seem on the surface to conflict. It is unclear, though, whether classic findings of expert disadvantage in consumer research should be viewed as reflecting a curse of expertise or completely adaptive behavior on the part of more knowledgeable consumers. A similar argument can be made for Bettman and Park's (1980) result that search is lower for high PK than for moderate PK consumers. Several of the disadvantages of PK noted are based on the knowledgeable consumers' complacency in reliance on old knowledge. Higher PK may lead to overconfidence (e.g., "I will learn this new software program in one night"), and this may abbreviate search or processing in a dysfunctional, superficial way. Similarly, the use of inappropriate schemas may be exacerbated by a strong familiarity-induced FOK. Thus, we hypothesize: H1: When obsolescence of PK is not cued explicitly, higher PK may lead to lower scores for new product learning compared to those consumers with lower PK. The argument that PK is not detrimental to learning when change is explicitly cued assumes that the negative effect of PK on learning new product information is due to shallow processing at encoding. In other words, when consumers with higher PK do not recognize that the new product represents a substantial change within the product category (i.e., PK has become obsolete or does not apply to the new product), they may not devote sufficient attentional resources to the learning task. This is in accord with We reason that, when motivated by recognition of change, higher PK consumers may devote the necessary resources to benefit from their enriched cognitive resources. H2: When obsolescence of PK is explicitly cued at the time of new product information exposure, higher PK consumers' scores for new product learning will improve relative to uncued scores more than is true for lower PK consumers. This motivation to process new information may occur naturally via the change cues. Moreau, In real innovation adoption contexts, expert consumers may be affected by the correlated constructs of prior domain knowledge and increased product category involvement STUDY 1: PRIOR KNOWLEDGE AND NEW PRODUCT LEARNING The goal of the first study was to test the influence of PK about allergy medications on the learning of information about a new allergy remedy (hypotheses 1 and 2). To avoid confounds with involvement, we chose to manipulate PK. We PRIOR KNOWLEDGE AND NEW PRODUCTS 419 FIGURE 1 CONCEPTUAL-PROCEDURAL DIAGRAM NOTE.-The dotted arrows indicate that incentive timing differed by study. Motivational incentives occurred only at encoding in study 1 and either before or after encoding in studies 2 and 3. The dashed arrows represent conceptual influence of prior knowledge on new product learning. chose a product category about which our respondents would have low PK and administered a training exercise to the high PK group prior to receipt of new product information. We manipulated the observable newness of the new product by altering superficial similarity of the new product to the old product. The purpose of this manipulation was to determine if a salient newness cue would promote more careful processing by higher PK participants. If higher PK participants make inappropriate inferences or use shallow processing because they are unaware of substantive changes in the product category, this newness cue might trigger better performance by higher PK than by lower PK participants. Without the cue, we expected experts to learn less new product information than novices. We chose allergy medications because there is a clear relationship between proper use and efficacy with pharmaceutical products. If a drug that should be taken on an empty stomach is taken with food, it may not work effectively, or it may cause unexpected side effects. Thus, if subjects score poorly on a test of usage instructions (and this is indicative of their actual behavior), we can plausibly assume that these subjects risk subpar product performance and perhaps even severe illness or death. For ethical reasons, the new product we introduced was fictional at the time of the studies, but it is similar to Claritin (loratadine), introduced in 1994. Method Design. A 2 (Higher versus Lower Prior Knowledge) # 2 (Drug Form) # 2 (Side Effects) between-subjects design tested the influence of expertise on learning and intended usage. Subjects were randomly assigned to one of eight conditions. To manipulate PK, participants read an information booklet on either allergy medications or toothwhitening processes. Those who read about allergy medications were designated as higher PK, while those who read about tooth-whitening processes were lower PK. The new product introduced later in the session was a new hybrid antihistamine. Two newness cue factors, Drug Form and Side Effects, manipulated the superficial similarity of the new medicines to existing medicines. For Drug Form, the new product was shown to be either a pill (similar to existing products, thus no newness cue) or a topical patch (dissimilar to existing products, providing a newness cue). For Side Effects, the new medicine was reported to have 420 JOURNAL OF CONSUMER RESEARCH few side effects (similar to existing products, thus no newness cue) or no side effects (dissimilar to existing products, providing a newness cue). Drug Form produced no effects on any dependent variable, so the results reported below collapse across this factor. One hundred and eighty-eight students at the University of South Carolina participated in the experiment for course credit. Sixty-five subjects who indicated that they had suffered from allergies in the past were eliminated from the data analysis. Mean responses for both the eliminated sufferers and the remaining subjects will be reported in the results section. Results of reported analyses replicate when these sufferers are included, however, we exclude them because, when the training manipulation is layered on existing PK, it is theoretically nonobvious whether the potential resultant increase in knowledge will outweigh the potential increase in overconfidence. Procedure. Each session lasted one hour and was conducted in groups of two to 12 participants. After a study introduction, participants read general product category information booklets, ostensibly as a warm-up task. Participants in the higher PK condition read about allergy medications. Participants in the lower PK condition read about tooth-whitening processes. Both information booklets were similarly structured and contained similar amounts of information. After this, the booklets were taken from the participants, and the manipulation check-a short general knowledge test on allergies and allergy medications-was administered. Finally, participants read an information booklet that contained information about a new product, a hybrid antihistamine allergy medication. This information was prefaced with the true statement that most allergies are developed in the early to mid-twenties, and it was hoped that this knowledge would motivate active consideration of the new medication. Participants were given as much time as they desired to read about the new product. The brochure did not differ between conditions except for the picture of the medicine (shown as a pill or a patch) and the description that "Certizol does not interact with known medications and has few (no) side effects." The text contained some new product information and usage instructions that were congruent with existing products; however, some information and instructions differed from the PK. This represented the obsolescence of some PK common in product innovation. Then, the product information was removed, and participants responded to a survey about the new product in which items were embedded pertaining to current/past experience with allergy medications, confidence, participants' purchase intentions if an allergy were later developed, and a quiz concerning proper usage of the new medication. This quiz constituted the important dependent variable to measure new product learning. (See example questions in table 1.) The quiz tested subjects on their knowledge of how to use the new medication properly (i.e., "this medicine should be taken at night") and only covered information that was similar across all conditions