Information Examination As a Function of Information Type and Dimension of Consumer Expertise: Some Exploratory Findings

ABSTRACT - This paper reports on a study designed to examine information search patterns of differentially familiar subjects. The theoretical rationale predicts different patterns of search depending on whether subject familiarity is measured on terminology, attribute, or marketplace related factors and whether subjects have the option of searching for either intrinsic or extrinsic information. Using an electronic environment in which student subjects responded to personal computer based stimuli, it was found that terminology and attribute based familiarity generated "U-shaped" information search patterns for extrinsic information, while marketplace based familiarity generated a hyperbolic search pattern for intrinsic information.

Citation:

Akshay R. Rao and Eric M. Olson (1990) ,"Information Examination As a Function of Information Type and Dimension of Consumer Expertise: Some Exploratory Findings", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 361-366.

Advances in Consumer Research Volume 17, 1990      Pages 361-366

INFORMATION EXAMINATION AS A FUNCTION OF INFORMATION TYPE AND DIMENSION OF CONSUMER EXPERTISE: SOME EXPLORATORY FINDINGS

Akshay R. Rao, University of Minnesota

Eric M. Olson, University of Minnesota

[The first author gratefully acknowledges funding for this project from the Graduate School, University of Minnesota.]

ABSTRACT -

This paper reports on a study designed to examine information search patterns of differentially familiar subjects. The theoretical rationale predicts different patterns of search depending on whether subject familiarity is measured on terminology, attribute, or marketplace related factors and whether subjects have the option of searching for either intrinsic or extrinsic information. Using an electronic environment in which student subjects responded to personal computer based stimuli, it was found that terminology and attribute based familiarity generated "U-shaped" information search patterns for extrinsic information, while marketplace based familiarity generated a hyperbolic search pattern for intrinsic information.

INTRODUCTION

Existing theories of buyer behavior (Bettman 1979, Howard and Sheth 1969) have proposed that the degree of prior knowledge (or expertise) a consumer has about a product, product class, the marketplace and a variety of other facets of the purchasing process is likely to influence information search, examination, evaluation, and eventually, choice. This issue of consumer expertise or familiarity has recently received considerable scrutiny in the consumer behavior literature (Alba and Hutchinson 1987, Brucks 1985, Johnson and Russo 1984, Park and Lessig 1981, Punj and Staelin 1983, Rao and Monroe 1988, Sujan 1985). The principal issues of interest are a) the dimensions of consumer expertise and b) the influence of differential levels of expertise on behavior.

This paper will describe findings from an exploratory study that examines the relationship between various dimensions of expertise and consumers' information examination patterns. As the study was conducted in an electronic environment using personal computers, it was possible to use unobtrusive measures and disguise the purpose of the study quite effectively.

LITERATURE REVIEW

Expertise and Information Examination

Differentially knowledgeable consumers are expected to behave differently when searching for new information primarily because they have different amounts of information already available in memory. Studies that have examined this issue have essentially used single item measures of subjective knowledge and examined the relationship between knowledge and degree of information search.

Two conflicting hypotheses inform the research stream. On the-one hand, it has been argued and empirically validated that increased familiarity leads to enhanced search (Pun; and Staelin 1983). This, the enrichment hypothesis, rests on the logic that existing knowledge facilitates the acquisition of new information -- a "rich get richer" argument. The second hypothesis predicts an "inverted U" effect, as increased familiarity results in an initial increase in information search; however exceedingly familiar individuals do not need to acquire new information as-all relevant information is already available in memory (Bettman and Park 1980, Park and Less-1981). The essential difference in predictions between the two hypotheses is at the high familiarity level. While low and moderately familiar individuals are expected to search for low and moderate amounts of information according to both hypotheses, the high group is expected to search more than (or at least as much as) the moderate group according to one hypothesis (the enrichment hypothesis), while it is expected to search less than the moderate group according to the inverted U hypothesis.

To reconcile these seemingly conflicting perspectives, Rao and Monroe (1988) proposed information diagnosticity as a moderating variable. They argue that different items of information are differentially informative. Using the extrinsic versus intrinsic cue dichotomy (Cox 1962, Olson 1977) they hypothesize that differentially familiar individuals will rely on different types of information, depending on the degree to which the information is useful for the task at hand. Specifically, when assessing product quality, moderately and highly familiar consumers were expected to use intrinsic information (information directly related to product performance, e.g., nutritional content in cereal) to a greater degree than low-familiar consumers, because low-familiar consumers would be unable to interpret intrinsic information successfully. This prediction is analogous to the enrichment hypothesis. The degree to which low-familiar consumers would use extrinsic information (information not directly related to product performance, e.g., price or brand name) was hypothesized to be higher than the degree to which moderately familiar consumers would use extrinsic information. This was expected to occur because, while both groups could interpret extrinsic information successfully, the moderate group would prefer to use information already available in memory. However, whether highly familiar consumers would use extrinsic information to a greater or lesser degree than moderately familiar consumers was hypothesized to depend on whether the item of extrinsic information was diagnostic for the task. In other words, in an evaluative situation, brand name or price (both extrinsic cues) may or may not be accurate indicators of quality. If they are good indicators of quality, the highly familiar consumer would know so and therefore use them, not otherwise.

In a price-quality experiment using a product that displayed a positive price-quality correlation in the marketplace, Rao and Monroe (1988) showed that as familiarity increased, reliance on extrinsic cues first reduced then increased (a "U-shaped curve") while reliance on intrinsic cues increased with increased familiarity.

The principal argument seems to be that only information that is interpretable and believed to be (or known to be) useful for the task at hand, will be examined. This argument will be revisited after an examination of the construct "expertise".

Dimensions of Consumer Expertise

According to Alba and Hutchinson (1987) and Brucks (1986), consumer expertise is comprised of multiple dimensions. Brucks (1986) developed a typology of consumer knowledge consisting of eight components. Seven of these knowledge types are declarative in nature, e.g., knowledge about concepts, objects or events. The eighth is procedural in nature and refers to those rules consumers utilize when taking action. She reduced these components into three factors which can be conveniently labelled as follows: terminology related knowledge, attribute related knowledge and marketplace related knowledge. Note that the first two factors are related to knowledge of intrinsic information, while the third factor is related to knowledge of extrinsic information. Terminology related knowledge can be defined as knowledge of terms related to the construction of a product (e.g., overhead cams for an automobile). Attribute related knowledge can be defined as knowledge of the relationship between a term and another term or the relationship between a term and a benefit (e.g., 4-wheel drive in an automobile results in better traction in snow.) Marketplace related knowledge can be defined as knowledge of relationships that prevail between extrinsic information and other extrinsic information (e.g., BMW automobiles are above average in price).

The multi dimensional nature of the construct calls into question whether differential degrees of familiarity on each of the dimensions will result in similar information search patterns. The argument that information examination patterns are likely to be different will be developed in the next section.

CONCEPTUAL PREDICTIONS

The existing literature suggests that differences between the low, moderate and high familiarity groups are qualitative, as well as quantitative. In other words, moderately familiar consumers are more knowledgeable than low-familiar consumers about elements that low-familiar consumers are cognizant of, as well as elements about which low-familiar consumers are unaware. Similarly, highly familiar consumers are more knowledgeable than moderately familiar consumers about elements that moderately familiar consumers are cognizant of, as well as elements about which moderately familiar consumers are unaware. Rao and Monroe 1988 argue that the qualitative difference between the low and moderate groups is knowledge of product attributes (and how they relate to performance); and the difference between the moderate and high groups is the knowledge about market related factors (and how they relate to quality). Their argument is that, to evaluate quality, consumers must first learn what attributes are good indicators of quality. Consequently, knowledge that the number of heads on a VCR implies a better VCR has to be learned first, before a surrogate (such as brand name or price) or market related piece of information can be used to assess quality. -This enhanced ability (to correlate market-based information with performance) tends to occur after the individual has acquired knowledge as to what constitutes quality. However, the use of extrinsic (market related) information by low-familiar consumers occurs because such information is easier to interpret than intrinsic information. In sum, information that is interpretable and informative will be sought and examined.

The argument presented above rests on an intentional learning approach to familiarity. However, if familiarity or expertise can be acquired incidentally, as well as intentionally, it is conceivable that individuals can acquire covariation information such as the relationship between extrinsic information and quality, without first acquiring information about the relationship between intrinsic information and quality. In other words, it is possible to be knowledgeable about the high quality of SONY tape players, without having any knowledge of the specific chips used in the circuitry.

In the context of Brucks's (1986) typology of knowledge content, this distinction between learning paradigms becomes particularly relevant. Clearly, attribute related knowledge (i.e., the degree to which attributes relate to each other or affect performance) cannot be acquired without terminology related knowledge (i.e., the knowledge of the terminology used to describe attributes). However, it is quite conceivable that marketplace related knowledge need not result only after attribute related knowledge is acquired, as performance judgments can be formed based on word of mouth, usage and shopping experience, under the incidental learning paradigm. Consequently, enhancements in each of the three components of familiarity may result in different patterns of examination of different types of information. The specific patterns predicted are addressed below.

When differentially familiar subjects are asked to evaluate the quality of a product (for which, extrinsic information is informative about quality):

P1a: Terminology related familiarity will display a U shaped relationship with extrinsic information examination.

P1b: No relationship is predicted between terminology related familiarity and intrinsic information examination.

Here, low-familiar consumers are expected to be able to interpret extrinsic information and thus examine it. However, as familiarity on terminology increases, reliance on intrinsic information (in memory) would supersede reliance on extrinsic information, hence examination of extrinsic information would reduce. Finally, as familiarity reaches a high level, subjects are expected to be cognizant of the relationship between extrinsic cues and quality, and thus are likely to rely on extrinsic information to a greater degree than moderately familiar subjects. For intrinsic information, the low group will spend little time given their inability to interpret it, while the moderate and high groups will spend little time, as they already have the necessary information in memory.

P2a: Attribute related familiarity will display a U shaped relationship with extrinsic information examination.

P2b: No relationship is predicted between attribute related familiarity and intrinsic information examination.

The relationship predicted in P2 (a and b) is based on the same logic as that for P1. When making quality assessments increasing familiarity on the part of the consumer initially leads to a reduction in time spent on extrinsic information examination. However, subsequent increases in familiarity lead to an enhancement in time spent on extrinsic information examination. This occurs because low-familiar subjects have the ability to interpret extrinsic information and therefore will rely on it. High-familiar subjects know this information to be informative and will therefore utilize it. In contrast, moderately familiar subjects do not know this information is informative and will therefore rely on other (intrinsic) information in generating quality assessments. Further, there are no anticipated differences for intrinsic information examination, for the reasons mentioned in support of P1b.

P3a: Market related familiarity will display a hyperbolic relationship with intrinsic information examination.

P3b: No relationship is predicted between marketplace related familiarity and extrinsic information examination.

Here, low-familiar subjects will not be able to interpret intrinsic information. However, moderately and highly familiar subjects do not have intrinsic information in memory. Consequently, given their better developed knowledge structures, they will tend to examine intrinsic information to a greater degree than low-familiar subjects (the enrichment hypothesis). However, given their marketplace related knowledge, they would tend to spend relatively little time (like low-familiar subjects) on extrinsic information.

METHOD

To select a product that would lend itself to an examination of the issues mentioned above, a review of prior literature was undertaken. Automobiles (Johnson and Russo 1984), microwave ovals (Park and Lessig 1981) and woman's blazers (Rao and Monroe 1988) are some of the products that have been used in this research stream. Given the need to use a product for which there was a subject pool that ranged in familiarity, a woman's blazer was selected as the test product. This product allowed for the use of student subjects comprising males, females, and clothing and textile majors to represent the three groups of low, moderate and high familiarity (Rao and Monroe 1988, Sujan 1985).

A seventeen item familiarity scale was constructed using the typology developed by Brucks (1985) and operationalized in Rao and Monroe (1988). The multiple indicators were representative of the three components of interest: terminology, attributes, and marketplace knowledge.

Data was collected from subjects using a personal computer based methodology (Rao 1988). Subjects responded to stimuli on a computer screen by "clicking" a pointing device on appropriate menu options. Subjects first went through a series of orienting exercises to make them comfortable with the methodology. Then, a battery of items designed to assess their level of familiarity with the product was presented to them and their responses to these items were automatically recorded. Subsequently, in a product evaluation task context, they were exposed to an advertisement comprising a picture-word combination and asked to select further information if they so desired, prior to their evaluation. Unlike Johnson and Russo (1984) who used recall as their dependent measure, and Punj and Staelin (1983) who used self reports of hours spent in search, we measured the time spent in examining available information as the dependent variable of interest. [The stimuli, manipulations and familiarity scales are available from the authors.]

ANALYSIS AND RESULTS

Ninety participants responded to sixteen objective questions (and one subjective, self-assessment of familiarity) pertaining to different types of knowledge of women's blazers. Participants were allowed to either immediately make an evaluation of the product, or to access additional information and then subsequently make an evaluation. Half the subjects were exposed to intrinsic information, and half the subjects were exposed to extrinsic information, when they requested more information. (Because it cannot be determined whether those who elected not to search for more information failed to do so because of expertise and/or high self-confidence, or merely wished to conclude their obligatory time in the study as quickly as possible, their responses were treated as outliers and were not included for further analysis. This left a set of 78 usable responses.) Individual scores based on the three factors were generated and then regressed on the amount of time each individual spent examining information related to the product.

TABLE

RESULTS OF REGRESSION RUNS

Prior to performing a regression analysis, three independent variables were created by collapsing appropriate familiarity scale items. Thus, measures for terminology related, attribute related, and marketplace related knowledge were created, based on Brucks's (1986) typology.

Regression Analysis

Our intent in this study was to identify patterns of information seeking: an exponential (the enrichment hypothesis) shape or a U-shape. Rather than set arbitrary levels of relative expertise within our study population, and then test for differences between groups, we performed a regression analysis, as our independent variable (familiarity) existed in a continuous form, not a categorical form. The dependent variable was time.

The independent variable consisted of the total value of weighted (by degree of difficulty, as determined by a Professor of Clothing and Textiles) correct answers to questions associated with each of the three factors mentioned. A relatively high inter-item reliability (a = .7104) was obtained.

To test for exponential and quadratic forms (U-shaped), two regression models were run on the data. For the enrichment hypothesis, the appropriate model is hyperbolic:

1/y = b0 + b1(1/x).

For the U-shape, the appropriate form is quadratic:

Y = b0 - b1X + b2X² (Myers 1986).

The results of the regression runs are presented in the Table.

As the results indicate, there is some marginal support for P1 (and P2), and there seems to be some support for P3. The U-shapes are evident for terminology and attribute based familiarity, when examining extrinsic information. No patterns were found for terminology and attribute based information as they related to intrinsic information examination. Similarly, a hyperbolic relationship is manifested between market based familiarity and intrinsic information examination, but no pattern was found for the relationship between this factor and extrinsic information examination. The implications of these findings are discussed in the next section.

DISCUSSION

The results from this study are interesting, though tentative. Consumers who are differentially knowledgeable about terminology exhibit a "U-shaped" information search pattern for extrinsic information, with increasing familiarity. In other words, novices and experts tend to spend more time examining extrinsic information, than individuals who are moderately familiar. Attribute-knowledgeable individuals exhibit similar (though statistically weaker) results; the negative linear component of the equation suggests that increases in knowledge resulted in reduced time spent but the quadratic component of the model was not significant. Market- knowledgeable individuals display a hyperbolic form search pattern; an enhancement in familiarity leads to increased search and examination of intrinsic information (the enrichment hypothesis).

The statistical support for these findings is tenuous at best (p<.10). Perhaps the addition of a covariate (e.g., reading speed of subjects) would have enhanced the significance of the results. However, from a substantive, as well as methodological perspective, the results are interesting. The principal theoretical contribution of this study is that the coexistence of the inverted U and exponential hypotheses are moderated by a factor other than task (Johnson and Russo 1984) and information diagnosticity (Rao and Monroe 1988), e.g., familiarity dimension. As consumers become knowledgeable about market-based information, they tend to examine intrinsic (product related) information to an increasing degree, while consumers who become knowledgeable about product (attribute and terminology) related information tend to first reduce and then increase their examination of extrinsic information.

The rationale for the findings is theoretically defensible and has been suggested elsewhere (Johnson and Russo 1984, Rao and Monroe 1988). In essence, consumers at the low end of the familiarity continuum are unable to interpret intrinsic information, while those who are moderately and highly familiar are able to do so. Hence, information examination follows a hyperbolic function. In the context of our study, this was manifested in the market-knowledge dimension of familiarity when regressed against time spent on intrinsic information. It was not manifested in the attribute-knowledge or terminology-knowledge and intrinsic information scenarios simply because individuals possessing attribute- and terminology-knowledge would not need to spend much time on attribute- and terminology-related information (intrinsic information); hence, the absence of any trend for those two factors.

For the U-shaped trend for extrinsic information examination, the same logic prevails. Low-familiar consumers, not having much information, would tend to spend time examining this information, given the interpretability of the information and the evaluative task being performed. However, moderately familiar consumers would tend to dismiss this information as irrelevant to the task at hand. as their better developed knowledge structure would allow for evaluation using intrinsic information in memory. However, highly familiar consumers would examine extrinsic information, if they believed it to be informative. In the case of women's blazers, extrinsic information is indeed informative (Rao and Monroe 1988). Consequently, consumers at the high end of the familiarity continuum would examine this information to a greater degree than moderately familiar consumers, who are not aware of the degree to which extrinsic information is useful in assessing quality. In other words, as consumers become increasingly familiar, they first lower their reliance on extrinsic information, and then increase their reliance on extrinsic information, if this information is useful for the task at hand (Rao and Monroe 1988). This trend would only hold if, as a consumer moves from the moderate to the high condition, his/her enhancement of knowledge results in an understanding of the existing, marketplace (or ecological) correlation between extrinsic information and quality. In other words, the distinction between low and moderate familiarity is knowledge of intrinsic attributes, while the distinction between moderate and high familiarity is knowledge of ecological correlations and marketplace phenomenon. Consequently, this trend should only be manifested for non-marketplace related dimensions of familiarity (i.e., attribute and terminology). This trend would not be exhibited for the marketplace dimension, because increasing knowledge on the marketplace dimension would imply increased knowledge about the relationship between extrinsic cues and quality (performance), i.e., any ecological correlations that may exist. Hence, there would be no reason for increasingly knowledgeable subjects to spend any more or less time on examination of marketplace related information.

From the methodological perspective, the use of the personal computer allowed for unobtrusive, involving and refined measurement. The presentation involved both picture and word stimuli and allowed for the development of a credible cover story regarding computerized shopping. Further, the use of a continuous variable allowed for the use of regression and trend analysis, rather than between groups analyses (e.g., Rao and Monroe 1988).

The results from this study are tentative, at best. However, this is the first attempt at testing a specific functional form that is theoretically defensible; consequently the lack of significance at the .05 level is not a source of concern. Apart from replicating this study under other scenarios using different products, it would be of interest to examine the differential reliance on the two types of information available to differentially familiar subjects. In other words, theoretically justifiable functional forms can be derived to predict changes in relative differences of time spent on different kinds of information, as the degree of familiarity on various dimensions increase. These findings would be of considerable interest to purveyors of information (marketers, consumer advocates, public policy makers) as they attempt to determine what kinds of information their segment of interest should be provided, given their relative knowledge about various aspects of the product, product class and marketplace.

REFERENCES

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

Bettman, James R. (1979), An Information Processing Theory of Consumer Behavior, Reading, MA: Addison-Wesley.

Bettman, James R. and C. Whan Park (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes," 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.

Brucks, Merrie (1986), "A Typology of Consumer Knowledge Content," in Advances in Consumer Research, Vol. 13, ed. Richard J. Lutz, Provo, UT: Association for Consumer Research, 58-63.

Cox, Donald F. (1962), 'The Measurement of Information Value: A Study in Consumer Decision-Making," in Emerging Concepts in Marketing, ed. William S. Decker, Chicago, IL: American Marketing Association, 413-421.

Howard, John A. and Jagdish N. Sheth (1969), The Theory of Buyer Behavior, New York: John Wiley.

Johnson, Eric J. and J. Edward Russo (1984), "Product Familiarity and Learning New Information," Journal of Consumer Research, 11 (June), 542-550.

Myers, Raymond H. (1986), Classical and Modern Regression With Applications, Boston, MA: Duxbury.

Olson, Jerry C. (1977), "Price as an Informational Cue: Effects on Product Evaluations," in Consumer and Industrial Buying Behavior, eds. Arch G. Woodside et. al., New York: North-Holland, 267-286.

Park, C. Whan and V. Parker Lessig (1981), "Familiarity and Its Impact on Consumer Decision Biases and Heuristics," Journal of Consumer Research, 8 (September), 223-230.

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.

Rao, R. Akshay and Kent B. Monroe (1988), "The Moderating Effect of Prior Knowledge on Cue Utilization in Product Evaluations," Journal of Consumer Research, 15 (September), 253-264.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments," Journal of Consumer Research, 12 (June), 31 46.

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