Knowledge and Knowledge of Knowledge: What We Know, What We Think We Know, and Why the Difference Makes a Difference

Lawrence Feick, University of Pittsburgh
C. Whan Park, University of Pittsburgh
David L. Mothersbaugh, University of Pittsburgh
ABSTRACT - This paper is a summary of a special session titled, "Knowledge and Knowledge of Knowledge: What We Know, What We Think We Know, and Why the Difference Makes a Difference." The papers presented in the session focus on the constructs of knowledge, self assessments of knowledge, and over and underconfidence. The papers provide empirical results which shed light on these constructs, their interrelationships, and antecedents and consequences of the constructs.
[ to cite ]:
Lawrence Feick, C. Whan Park, and David L. Mothersbaugh (1992) ,"Knowledge and Knowledge of Knowledge: What We Know, What We Think We Know, and Why the Difference Makes a Difference", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 190-192.

Advances in Consumer Research Volume 19, 1992      Pages 190-192

KNOWLEDGE AND KNOWLEDGE OF KNOWLEDGE: WHAT WE KNOW, WHAT WE THINK WE KNOW, AND WHY THE DIFFERENCE MAKES A DIFFERENCE

Lawrence Feick, University of Pittsburgh

C. Whan Park, University of Pittsburgh

David L. Mothersbaugh, University of Pittsburgh

[The authors gratefully acknowledge the insightful comments provided by our discussant, Meryl Gardner.]

ABSTRACT -

This paper is a summary of a special session titled, "Knowledge and Knowledge of Knowledge: What We Know, What We Think We Know, and Why the Difference Makes a Difference." The papers presented in the session focus on the constructs of knowledge, self assessments of knowledge, and over and underconfidence. The papers provide empirical results which shed light on these constructs, their interrelationships, and antecedents and consequences of the constructs.

OVERVIEW

There has been much research directed at understanding the effects of knowledge on consumer behaviors such as information search and decision making (see for example, Beatty and Smith 1987, Bettman and Park 1980, Punj and Staelin 1983). Still, there is a lack of agreement and consistency about the definition and operationalization of the knowledge construct. For example, Alba and Hutchinson (1987) define two components of knowledge - familiarity (or experience) and expertise, while Brucks (1985) includes subjective knowledge as part of the knowledge construct. Alternatively, Beatty and Smith (1987) define consumer knowledge as self-assessed knowledge. However, the theoretical and empirical relationships among these constructs have not been fully explicated. For example, many researchers have used self-assessed knowledge as the sole measure of consumer knowledge. However, given the rather modest correlations (.4 to .6) that often have been found between measures of objective and self-assessed knowledge, the use of self-assessed measures of knowledge may obscure our understanding of the effects of objective knowledge on behavior. It may be more useful to consider the two variables as separate but related constructs that have important implications for consumer behavior. Since self-assessed kowledge is the consumer's perception of what or how much he or she knows, self-assessed knowledge may be the more important determinant of behaviors, particularly such behaviors as information search and decision making.

Consumers' perceptions of what or how much they know may be systematicaly biased. Specifically, consumers may be over or underconfident about how much they know about a given product class. Little research has focused on the determinants of over and underconfidence in knowledge assessment. The existence of over and underconfidence has important implications for consumer behavior researchers since these inaccuracies could lead to suboptimal consumer behaviors. For example, overconfidence could lead to insufficient information search, while underconfidence might cause consumers to be more susceptible to persuasion attempts. In addition, it is possible that the marketing actions of firms actually contribute to the bias through the amount, and/or type of information given.

The papers in this session focus on questions such as: what are the conceptual and empirical relationships among self assessments of knowledge, objectively measured knowledge, and product experience? What is the basis consumers use to assess their own knowledge? How does the basis they use affect their perceived extent of knowledge? What are the conditions under which consumers (and decision makers more generally) are likely to be over confident? Does overconfidence occur because of the effects of actual knowledge, perceptions of knowledge, or both? Are there ways to reduce overconfidence, i.e., to bring perceptions of knowledge more in line with actual knowledge?

The first paper, by Catherine Cole, Gary Gaeth, Goutam Chakraborty, and Irwin Levin, focuses on the relationship among the constructs of objective knowledge, self assessments of knowledge and product experience. The authors review their own empirical results which illustrate the interrelationship of the constructs across a number of product categories. In addition, they show the relationship of these constructs to other constructs of conceptual and practical importance.

The second paper, by Susan Broniarczyk, Wesley Hutchinson, and Joseph Alba, focuses on overconfidence. In two sets of laboratory experiments they examine the determinants of confidence in knowledge and the impact of types of presentation formats in a learning task on the effect of confidence of judgements.

The third paper, by Jayashree Mahajan, also focuses on confidence and overconfidence. In her paper Mahajan reports on a laboratory study which examines the sources of overconfidence in decision making about marketing strategy. She examines the impact of the type of evaluative feedback and the presence of contradictory evidence on overconfidence.

The final paper, by Park, Feick, and Mothersbaugh, examines the consumer knowledge assessment process. They gather data about the process consumers use to assess their own knowledge about C.D. players and examine the impact of the type of approach used on knowledge assessments.

More detailed abstracts of the papers follow.

 

EXPLORING THE RELATIONSHIP AMONG SELF-REPORTED KNOWLEDGE, OBJECTIVE KNOWLEDGE, PRODUCT USAGE AND CONSUMER DECISION MAKING

Catherine Cole and Gary Gaeth, University of Iowa

Goutam Chakraborty, Oklahoma State University

Irwin Levin, University of Iowa

Except for a rather small collection of experiments specifically designed to investigate relationships among different measures of knowledge, most studies operationalize the knowledge concept in whatever happens to be the most convenient fashion. However, partly due to recent theoretical work in the knowledge area, and partly due to indirect evidence emerging from our own work, in this paper we look more closely at these measures and report that there is a disturbing lack of shared variance.

First, we present data from a number of studies that have used a wide variety of product classes where we have measured the following three types of knowledge: self-reported knowledge (SK) - based on a response to rating scale measures; self-reported experience (SE) - based on reported frequency of use; and, object knowledge (OK) - based on a factual test. On average, the correlation is strongest (about .55), between SK and SE, and weakest (about .30), between SK and OK.

Second, we inspect the amount of variance these different measures of knowledge explain in a variety of dependent variables. Our results, again taken across a number of different studies, suggest that SK and SE relate best to dependent variables representing affective reactions. To illustrate, we found that involvement with a product class, (electronic typewriter), is significantly related to SK and SE (.23 and .46), but not related to OK (.02), while high basketball OK showed reduced context effects in assessing free throw proficiency. We speculate that OK may be related to task performance oriented measures (e.g., recall, accuracy, use of simplifying heuristics).

Two conclusions seem appropriate. First, it is risky to depend on only one of these three measures of knowledge. Our results would suggest substantively different conclusions may emerge depending on the measure chosen by the researcher. Second, our inconsistent results suggest prior work relating knowledge to other consumer behavior variables needs to be reconsidered with explicit attention paid to what type of knowledge measure was used.

 

WHAT CONSUMERS "REALLY" KNOW AND WHAT THEY "THINK" THEY KNOW: INVESTIGATIONS INTO THE DETERMINANTS OF CONFIDENCE AND PERFORMANCE

Susan Broniarczyk, J. Wesley Hutchinson and Joseph W. Alba, University of Florida

Two important problems for consumer research are (1) identifying the determinants of knowledge confidence and (2) developing interventions that can counteract the negative effects of this confidence when it is misplaced. In this paper, we present results from two sets of experiments that address these issues.

In the first set of studies a concept learning paradigm was used. Subjects took part in an initial learning phase in which they were exposed to two categories of goods within a single product class (e.g., high quality versus low quality stereo speakers). Afterward, a categorization test was administered in which subjects were presented with additional examples of the product class and were asked to determine the category to which each belonged. In addition, scale ratings of confidence for each judgement were taken. The percentage of correct categorization responses measured what subjects "really" knew, and the confidence ratings measured what consumers "thought" they knew. Results showed that our most powerful independent variables affected both measures. However, the manipulations of some factors affected only the objective measure, while others affected only the confidence measure. Interestingly, individual covariates such as gender, subjective knowledge, and objective knowledge primarily affected confidence. One explanation is that confidence is sensitive to salient manipulations and self-perceptions that subjects believe should affect performance but is not sensitive to "subtler" manipulations.

The second set of studies involved a prediction task in which we attempted to instill factual knowledge that was at variance with consumers' general beliefs. Specifically, subjects were presented with quality predictors for stereo speakers. In the learning phase, quality was highly correlated with level of advertising expenditure but was uncorrelated with price. This information was conveyed in one of three different ways: (1) through inspection of 25 fictitious brands and their associated levels of quality, price, and advertising, (2) through summarized central tendency information, i.e., the median prices and advertising expenditures of high quality and low quality brands, (3) through directly informing subjects of the correlations between price and quality and between advertising and quality. Afterward, 15 new brands were presented and subjects were asked to predict quality based on price and advertising levels. Results showed that prior beliefs exerted a strong influence in all cases but varied as a function of learning condition. When allowed to inspect individual brand information, low and nearly equal weights were placed on price and advertising as predictors of quality; that is, subjects were unwilling to rely heavily on either predictor despite the dominance of advertising in the data. When expressed as central tendencies, predictions were based almost exclusively on price. Finally, when expressed as correlations, subjects appropriately placed a high weight on advertising but also placed an equally high weight on price. Thus, even when provided with knowledge in an unambiguous manner, performance can be driven by what consumers think should be true. When prior beliefs are strong, attempts to debias consumers may be difficult.

 

THE OVERCONFIDENCE EFFECT IN STRATEGIC MARKETING PREDICTIONS

Jayashree Mahajan, University of Arizona

Estimating the likelihood of future events is a critical aspect of making strategic marketing decisions. An important bias that has emerged consistently in a variety of studies in the area of probability assessment is that of overconfidence (e.g. Fischoff 1982). Typically, individuals believe they know more than they actually know. These findings have implications for strategic market planning, as predictions of this type tend to be complex as compared to the types of judgment tasks reported in earlier work (Barnes 1984). More importantly, it is highly confident predictions that managers are most likely to act upon and commit resources to without pausing to consider additional information.

This study conceptually and empirically evaluates the sources of overconfidence by examining the psychological context within which strategic marketing predictions are made. Specifically, the processes of how individuals learn from past performance, generate evidence to assess the current situation, and subjectively construe the future are hypothesized to affect overconfidence. The study explores the robustness of the overconfidence bias in the context of predictions commonly made by managers from company and industry data contained in a strategic marketing plan. Three levels of evaluative feedback (Favorable, Neutral, Unfavorable) and two contradictory evidence conditions (Contradictory Evidence, No Contradictory Evidence) were investigated in a 3x2 factorial design. A total of 90 students participated in a study in which data were collected over 3 months.

The results suggest that evaluative feedback has a large impact on overconfidence, primarily through its impact on accuracy rather than through an impact on confidence. In addition, subjects who were required to provide contradictory evidence proved to be less overconfident than subjects who were not so required. There was no interaction between feedback and evidence.

To summarize the findings, overconfidence in strategic marketing predictions tends to be higher when: (a) prior performance is perceived to be superior, (b) no contrary evidence is generated, and (c) events are perceived to be more certain or familiar, or prediction behavior is atypical.

 

CONSUMER KNOWLEDGE ASSESSMENT: HOW PRODUCT EXPERIENCE AND KNOWLEDGE OF BRANDS, ATTRIBUTES, AND FEATURES AFFECTS WHAT WE THINK WE KNOW

C. Whan Park, Lawrence Feick and David L. Mothersbaugh, University of Pittsburgh

Researchers in consumer behavior have studied the effects of knowledge on a variety of consumption-related behaviors. Across studies, researchers often have used either a measure of actual knowledge or a self assessment of knowledge, or both. This paper focuses on the knowledge assessment process.

Results of an exploratory study designed to investigate the knowledge assessment process indicate that information ranging from specific knowledge about the product (e.g., brand names, attributes, and features), memory for experience events (e.g., ownership, usage, ad search), and involvement with the product category are used in making SAK judgements. The results indicate that of these, experience related information dominates consumers' responses - about 70 percent of the responses involved experience. The experience-based responses also seemed to have a large impact on the level of the respondent's SAK. That is, individuals who mentioned they had experience reported higher levels of SAK than those who didn't mention experience. Similarly, those who mentioned they lacked experience had lower levels of SAK than those who didn't mention experience.

Finally, some results of the study suggest that brand and attribute information was as available to individuals who used that information in their SAK assessment as it was to those who didn't. We discuss the implications of these results for future research.

REFERENCES

Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-454.

Barnes, James H. Jr. (1984), "Cognitive Biases and Their Impact on Strategic Planning," Strategic Management Journal, 5, 129-137.

Beatty, Sharon E. and Scott M. Smith (1987), "External Search Effect: An Investigation Across Several Product Categories," Journal of Consumer Research 14 (June), 83-95.

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

Brucks, Merrie (1985), "The Effects of Product Class Knowledge on Information Search Behavior," Journal of Consumer Research, 12 (June), 1-16.

Fischoff, Baruch (1982), "Debiasing," in D. Kahneman, P. Slovic, and A. Tversky (eds.), Judgement Under Uncertainty: Heuristics and Biases, New York: Cambridge University.

Punj, Girish N. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," Journal of Consumer Research, 9 (March), 366-380.

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