Difficulty of Pre-Purchase Quality Inspection: Conceptualization and Measurement

ABSTRACT - In this paper, we conceptualize and develop a measurement approach for assessing ''difficulty of pre-purchase quality inspection," a key construct in information economics. After reviewing the need for empirically defining this construct, we present a conceptual model to lay the foundation for our measurement procedure. Reliability of the measurement approach is acceptable. Evidence of convergent validity is provided by comparing rankings based on the DPQI measure with rankings provided by independent sources. Further, the measurement technique produces significantly higher scores (i.e., indicating easier pre-purchase quality inspection) for durables than for nondurables, as expected. Applications of such a measurement approach in examining predictions from information economics are considered.

Citation:

Arni Arnthorsson, Wendall E. Berry, and Joel E. Urbany (1991) ,"Difficulty of Pre-Purchase Quality Inspection: Conceptualization and Measurement", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 217-224.

Advances in Consumer Research Volume 18, 1991      Pages 217-224

DIFFICULTY OF PRE-PURCHASE QUALITY INSPECTION: CONCEPTUALIZATION AND MEASUREMENT

Arni Arnthorsson, University of South Carolina

Wendall E. Berry, University of South Carolina

Joel E. Urbany, University of South Carolina

ABSTRACT -

In this paper, we conceptualize and develop a measurement approach for assessing ''difficulty of pre-purchase quality inspection," a key construct in information economics. After reviewing the need for empirically defining this construct, we present a conceptual model to lay the foundation for our measurement procedure. Reliability of the measurement approach is acceptable. Evidence of convergent validity is provided by comparing rankings based on the DPQI measure with rankings provided by independent sources. Further, the measurement technique produces significantly higher scores (i.e., indicating easier pre-purchase quality inspection) for durables than for nondurables, as expected. Applications of such a measurement approach in examining predictions from information economics are considered.

Marketing researchers make frequent reference to Phillip Nelson's work on information and consumer behavior (see Nelson 1970; 1974). These references focus on the useful distinction between "search" and "experience" goods (e.g., Huber and Elrod 1981; Bettman 1982) and the key construct underlying the difference between them: difficulty of pre-purchase quality inspection (DPQI).

DPQI is a search cost fundamental to Nelson's research and subsequent work in the information economics area (Schmalensee 1978, Wilde 1980; 1981) and in the marketing literature (Hauser and Wernerfelt 1990). We develop a measurement procedure for DPQI here with the goal of facilitating tests of the propositions from these rich theories. Before detailing the measurement procedure, we consider the roots of the construct and a conceptual definition for it.

The Original Theory

In his widely cited theoretical and empirical research, Nelson proposed that consumer search differs quantitatively and qualitatively for products which differ in the ease/difficulty with which quality can be inspected prior to purchase. Search goods are easier to inspect prior to purchase than are experience good -- therefore consumers will sample more offerings, be better informed in general about quality, and will less likely use external cues for quality (like amount of advertising) for search goods.

This has wide implications for how sellers of search goods behave. According to the theory (Nelson 1970; 1974), they will tend to be more clustered in retail locations, will provide more informative advertising, and their industries will be less heavily concentrated than those of experience goods sellers (goods for which quality is more difficult to discern prior to purchase). In the marketing literature, concepts related to DPQI have been incorporated recently in Hauser and Wernerfelt's (1990) model of evoked set size, Tellis and Fornell's (1988) model of the advertising-quality relationship, and Zeithaml's (1988) discussion of price, quality, and perceived value. In spite of the importance of the construct in testing predictions from the theoretical work, only limited effort has been spent attempting to conceptualize and measure it.

Nelson's Empirical Operationalization

In Nelson's (1970) often-cited original paper, he empirically categorizes durable goods as either experience or search based upon aggregate ratios of nonmerchandise receipts-to-sales with the following logic: assuming the nonmerchandise receipts reflect in large part post-purchase repairs, then goods with higher ratios have a higher incidence of repairs and, therefore, a larger predominance of characteristics which could not be inspected prior to purchase. Durables with higher nonmerchandise receipts-to-sales ratios, then, were classified by Nelson as experience goods. This empirical definition suffers from (1) some leaping assumptions about the interpretation of nonmerchandise receipts and (2) the use of aggregate statistics to define a concept that effectively depends upon consumer perception. With the exception of Albion and Farris (1979), few researchers in marketing have questioned Nelson's empirical work.

A Conceptual Model of DPQI

DPQI is defined here as the difficulty of assessing or verifying the quality of a product prior to purchasing it. This definition requires specifying what is meant-by "inspection" and what is meant by "quality." Inspection is defined as the consumer's effort to physically examine, assess, and verify the quality of the product at the point of purchase. Following Zeithaml (1988), quality is defined as the value or payoff that the consumer obtains from purchase and consumption of the product. Pre-purchase quality inspection, then, involves examining the product in an effort to determine (before purchasing) how well the product will deliver on the benefits the consumer expects from it.

Q for Quality = Whether the Product is Expected to Deliver the Benefits Desired. One could alternatively define quality in terms of the technical characteristics of a product or the more abstract dimensions -- i.e., the benefits (value or payoff) one derives from using the product. We focus on the latter under the assumption that, at the point of purchase, the consumer's primary objective in examining quality is to assess the product's value-in-use. We assume that the pre-purchase assessment of value-in-use involves the evaluation of lower level "intrinsic" attributes as a step, but the ultimate judgment relates to whether the product will deliver the benefits the consumer expects. For example, a consumer may consider an intrinsic or lower level attribute of cola (e.g., whether or not it has caffeine) as a step in assessing whether or not the product delivers a desired benefit (e.g., "helps me to stay awake"). The consumer uses the information about the intrinsic attribute to judge the product's "quality" -- i.e., whether it will deliver the desired end benefit.

FIGURE A

CONCEPTUAL MODEL OF "DIFFICULTY OF PRE-PURCHASE QUALITY INSPECTION"

D for "Difficulty" of Inspecting Quality without the Help of Extrinsic Cues. As Zeithaml (1988) notes, consumers judge product quality prior to purchase via extrinsic attributes (e.g., brand reputation, price) as well as via physical inspection of the product's intrinsic attributes. However, we contend that the "difficulty of inspection" measure must ignore the use of extrinsic attributes because such cues are search attributes, are generally available across product categories and, therefore, provide little help in attempting to distinguish product categories based upon DPQI. Further, many of the predictions from the information economics theories to be tested argue that consumers use extrinsic cues to judge quality more for products which are higher in DPQI. To test this proposition, researchers must first have a way of separating lower DPQI products from higher DPQI products independent of the extrinsic cues attached to those products. Consistent with Nelson's original intention, then, our approach to defining DPQI focuses on the natural "inspectability" of a product's intrinsic characteristics at the point of purchase.

The Determinants of DPQI: A Simple Model

The measure of DPQI is, in essence, an empirical indicator of the "inspectability" of a product's intrinsic characteristics, stripped of brand name and price information. This is depicted in Figure A. Independent of the product's inspectability, however, DPQI may vary across consumers and even within consumers across different situations. Prior familiarity with the product should, all else equal, reduce DPQI, as experienced consumers who have gone through the consumption process may be able to more confidently estimate prior to purchase the chance that a product will perform adequately in the consumption setting (Tellis and Fornell 1988).

Further, characteristics of the purchase situation may influence the difficulty that a consumer has in inspecting a product's quality prior to purchase. For example, time pressure will generally have the effect of making pre-purchase quality assessment more difficult, potentially even for an experienced consumer evaluating a product which tends to have predominantly search characteristics. Finally, the availability of external information sources naturally influences DPQI. A salesperson or Consumer Reports can provide useful product information even (or, perhaps, especially) for the most difficult to evaluate products. Consistent with this, Nelson (1974) proposes that advertising to sales ratios will be greater for experience goods industries.

The survey procedure for measuring DPQI described below also incorporates the measurement of individual respondent familiarity with the product, as we naturally want to examine the influence of familiarity on DPQI. We either ignore or hold constant the other determinants of product category "inspectability" described in the Figure A model. We ignore situational characteristics given the random nature with which they are likely to occur in the marketplace (i.e., it is not clear that time pressure would be associated systematically with certain product categories and not others). We hold constant the external information sources (e.g., sales people, where relevant) by instructing respondents to assume that no such source is available as they are evaluating the product prior to purchasing it.

Predictions. While the major purpose of this paper is to develop and evaluate a scale to measure DPQI, two predictions are also assessed. The predictions simply provide a basis for determining whether the scale produces face valid results. The first is based upon Nelson's (1970) contention that durables have more "searchable" qualities than do nondurables. The second is based on the intuition from our model that familiarity should be negatively related to DPQI. Formally stated:

P1: Durables should have lower DPQI scores than nondurables.

P2: DPQI scores for low familiarity respondents should be higher than those for high familiarity respondents.

METHOD

DPQI Scale Development

Our procedure for measuring DPQI evolved over a series of attempts to ask consumers how easy/difficult it was to judge a given product's quality before purchase. The failures led to a procedure which focused on end benefits. In this final procedure, questions were framed in terms of how sure respondents could be (prior to purchase) that a specific end benefit would be obtained if they purchased a new brand X after evaluating it only at the point of purchase. Sureness was measured on a 10 point scale ranging from "I'm UNSURE before I buy" to "I'm SURE before I buy." To illustrate, the Appendix provides the page from the questionnaire which asked respondents to evaluate a new brand of VCR.

Pilot Study. A pilot study was done for the purposes of (1) determining product categories with which students would be most familiar and (2) determining what attributes/end benefits should be included in the measures for each product category. The pilot study involved a paper and pencil survey of 80 undergraduate business majors. The survey asked respondents to rate their familiarity with 29 durable and nondurable product categories and to list the important considerations in evaluating the quality of the five product categories with which they were most familiar; Based upon the familiarity ratings and the five most familiar products identified, ten products were selected for the study -NONDURABLES: beer, soda (2 liter, plastic bottle), toothpaste, cold cereal (flakes), deodorant; DURABLES: a VCR, a television set, a head stereo set (Walkman), jeans, and a wrist watch.

As the pre-purchase assessment of quality ultimately relates to the expected benefits, a list of benefits was generated for each of the ten products. The pilot study responses, along with Garvin's (1987) eight general dimensions of product quality (performance, features, reliability, conformance, durability, serviceability, aesthetics, and perceived quality) provided a framework for enumerating end benefits for each product category. In evaluating each product category, we attempted to generate statements of benefits/attributes for the relevant dimensions from Garvin's framework, making sure to address all the issues mentioned by the pilot study respondents. Five to eight questions were generated for each product.

Survey Administration

One hundred and four undergraduate business students each responded to the final questionnaire, which included the following measures for each of ten product categories: (1) the sureness measures for each of several benefits/attributes, (2) three measures of product familiarity (see the Appendix), and (3) rankings of several brands on quality, advertising, and price (not-reported in this paper). Four respondents were eliminated from the sample for providing incomplete or incoherent responses. For half the questionnaires, a random ordering of the products was used. This ordering was reversed for the other half. It was found that product ordering had no effect on any of the key measures in the study.

The survey was administered in a classroom setting, with all respondents present at once. Prior to beginning the questionnaire, respondents were given explicit instruction on the use of the measures in the questionnaire. Overhead transparencies were used to explain the measures and their meaning. The questionnaire took from between 25 and 45 minutes to complete.

RESULTS

The questionnaire administration involved having respondents judge their "sureness" in evaluating a new brand's quality prior to purchase on several benefits/attributes. Reliability is evaluated via a test-retest procedure for four of the ten products used. Convergent validity is evaluated by comparing the ranking of the 10 products produced by four "experts" (based upon a separate measurement procedure) with the ranking produced by our respondents (based upon their average "sureness" ratings). We then assessed the general a priori expectation that the five durables in the product list would be easier to evaluate prior to purchase than the nondurables and that respondents more familiar with the product category would report being more sure about pre-purchase quality evaluation than respondents less familiar.

TABLE 1

SURENESS ABOUT JUDGING QUALITY PRIOR TO PURCHASE

TABLE 2

OVERALL RANKINGS OF THE 10 PRODUCT CATEGORIES ON DPQI

Reliability

DPQI for the ith consumer and the jth product was empirically determined by calculating the average sureness for that consumer across the attributes for that product category. To evaluate the reliability of this indicator of DPQI, a second administration of the questionnaire including four of the original product categories was undertaken with 88 of the original respondents. Just four product categories were chosen to reduce the required time for answering the second questionnaire. The product categories most familiar to pilot test respondents were selected (toothpaste, wrist watch, soft drink, jeans). The overall test-retest correlation for the average sureness measure (aggregating the data across the 4 products) was .67, an acceptable level for basic research (Nunnally 1967).

Convergent Validity

Three faculty members and one doctoral student were asked to provide an overall ranking of the 10 product categories based upon their general perceptions of the ease/difficulty of inspecting the product's quality at the point of purchase (with no brand name or price information available). The experts' aggregate ranking (based upon the average ranking of each product category across the four experts) is presented in the last column in Table 2, next to the rankings provided by the low and high familiarity respondents based upon their average sureness measures (to be discussed later). The Spearman rank order correlations between the experts' and respondents' rankings are .72 for low familiarity respondents (t=2.92, p<.05) and .75 for high familiarity respondents (t=3.19, p<.05). In short, there was strong agreement between the independent product rankings provided by the experts and the rankings produced by the DPQI scale, providing evidence of convergent validity.

Descriptive Results

For each product category, each respondent rated their perceived "sureness" in judging (prior to purchase) whether a new brand would deliver on five to eight end benefits. The overall mean position of each attribute (or end benefit) for each product category on the sureness scale is presented in Table 1. In the table, nondurables are presented first and durables second. The attributes of each product are categorized based upon the sample's mean sureness ratings on the 10 point scale. Each attribute is categorized as either in the middle of the scale (not significantly different from the middle point 5.5), or toward the low/high end of the scale (significantly different from the middle point). In Nelson's terminology, the attributes on the lower end of the scale would tend to be experience attributes (i.e., respondents less sure about their ability to evaluate the benefits prior to purchase) while the attributes on the higher end would tend to be search attributes.

In examining the attributes falling in the column marked "low sureness," it is clear that they tend to relate to such unobservable (prior to purchase) characteristics as performance, taste, reliability, and durability. The attributes falling in the third column, "high sureness" tend to be more observable, relating to available product features, product color, style, size, and benefits delivered via packaging. Just two of the ten products failed to support our general expectation that the nondurables' attributes would tend toward low sureness and that the durables' attributes would tend toward the high sureness end of the scale. For cereal, only taste falls into the low sureness category, while judgments of nutrition and hunger satisfaction were on the higher sureness end of the scale. This latter finding could be a result of package information and/or respondents' confidence in generalizing from experience with other brands. The other exception is the VCR category, which predominantly has features which are difficult to evaluate outside of the consumption context. In hindsight, the key attributes of a VCR do tend to be more difficult to observe (or more difficult for a salesperson to demonstrate) than, say, the key attributes of a TV.

Explaining the Variance in DPQI

An ANOVA model was run which examined differences in DPQI across the ten product categories and two familiarity groups. The three measures of familiarity for each product category ("familiar," "experienced in using," "experienced in shopping for") had an overall alpha of .87 across the 10 product categories. For each product category, these three measures were summed to form an index, and respondents were split into low and high familiarity groups for each product category based upon a median split.

The 10 X 2 ANOVA model explained 23 percent of the variance in DPQI. Both the product and familiarity factors had significant main effects (P{9, 10063 = 28.37, p < .001 for product; F{1, 1006} = 53.41, p < .001 for familiarity) and, unexpectedly, the interaction was significant as well (F{9, 10063 = 2.65, p < .01). The significant product effect appears due to the lower average sureness scores received by nondurables. [A separate 2 X 2 ANOVA substituting a two level durability factor for the product factor indicated that nondurables had significantly lower sureness scores than did durables {p<.001}, and found no interaction between durability and familiarity.] The interaction occurs because familiarity increased sureness for some product categories (beer, soda, wrist watch, vcr, head stereo set) but not others. The reason for this is not entirely clear. In sum, the scale passes our test of face validity in detecting differences in DPQI between durables and nondurables. The results are consistent for half of the product categories with the expectation that DPQI would be lower for consumers more familiar with the product category. [A reviewer has noted correctly that the relationship between familiarity and DPQI may be driven in part by the fact that the constructs were measured on the same page of the questionnaire.]

DISCUSSION

A limitation of the proposed procedure for measuring DPQI is that respondents are asked to answer questions which require them to "pretend" that they are evaluating a new brand at the point of purchase. Further, they are asked to assume that they have no external information available (other than packaging). This approach asks a lot of respondents' ability to imagine the situation. In spite of this potential shortcoming, however, the DPQI measure showed reasonable reliability and convergent validity and, further, was sensitive enough to identify differences between durables and nondurables.i Future development of the measurement approach will focus on identifying aspects of the questioning approach that are not entirely clear to respondents. While there appeared to be no confusion among our respondents during administration, the test-retest reliability suggests that improvements can be made in making the questions more concrete.

Why does one care to measure DPQI? We contend that DPQI is one of a class of costs facing consumers which may affect consumer behavior in important ways (Hauser and Wernerfelt 1990; Zeithaml 1988). Further, the empirical identification of how product classes vary in DPQI holds the potential of examining many propositions regarding how sellers behave under varying conditions of consumer search/evaluation cost.

APPENDIX

ILLUSTRATION OF DPQI SCALE QUESTIONS: VCR

REFERENCES

Albion, Mark S. and Paul W. Farris (1979), "Appraising Research on Advertising's Economic Impact," Marketing Science Institute Working Paper 79-115, Cambridge, MA.

Bettman, James R. (1982), "A Functional Analysis of the Role of Overall Evaluation of Alternatives in Choice Processes," in Andrew Mitchell (ed.), Advances in Consumer Research, Vol. 9, Chicago: Association for Consumer Research.

Hauser, John R. and Birger Wernerfelt (1990), "An Evaluation Cost Model of Evoked Sets," Journal of Consumer Research, 16 (March), 393-408.

Huber, Joel, and Terry Elrod (1981), "Consumer Learning through Experience: A Study and Experimental Paradigm," in Kent Monroe (ed.), Advances in Consumer Research, Vol. 8, Chicago: Association for Consumer Research.

Nelson, Philip (1980), "Advertising and Ethics," in Tom L. Beauchamp and Norman E. Bowie (eds.), Ethical Theory and Business, Englewood Cliffs, NJ: Prentice-Hall, 343-7.

Nelson, Philip (1978), "Advertising as Information Once More," in David Tuerck (ed.), Issues in Advertising, Washington, D.C.: American Enterprise Institute, 133-60.

Nelson, Philip (1975), "The Economic Consequences of Advertising," Journal of Business, 48 (April), 213-41.

Nelson, Philip (1974), "Advertising as Information," Journal of Political Economy, 81 (July/August), 729-54.

Nelson, Philip (1970), "Information and Consumer Behavior," Journal of Political Economy, 78 (March/April), 311 -29.

Nunnally, Jum C. (1967), Psychometric Methods, New York: McGraw-Hill.

Parasuraman, A., Valerie A. Zeithaml, and Leonard L. Berry (1985), "A Conceptual Model of Service Quality and Its Implications for Future Research," Journal of Marketing, 49 (Fall), 41-50.

Tellis, Gerard J. and Claes Fornell (1988), 'The Relationship between Advertising and Product Quality over the Product Life Cycle: A Contingency Theory," Journal of Marketing Research, 25 (February), 64-71.

Zeithaml, Valerie A. (1988), "Consumer Perceptions of Price, Quality, and Value: A Means-End Model and Synthesis of Evidence," Journal of Marketing. 52 (July), 2-22.

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