Sie sind auf Seite 1von 9

Food Quality andPreference (1995) 191-199 Elsetier Science Limited Printed in Great Britain. ELSEVIER 0950-3293(94)00023-9 095~3293/95/$9.50+.

00

DlMENSlONSOFSENSORYQUALITY:ACRlTlQUE
Harry Lawless
Department of Food Science, New York State College of Agriculture and Life Sciences, Cornell University, Ithaca, NY 14850, USA

ABSTRACT
This essay argues that quality is a complex term used in

different ways. In addition to sensory quality, there are


factors such as nutritional content, safety, shelf-life and reliability, that contribute to the consumers overall opinion of a food product. Product quality is multidimensional. Some sensory quality determinations have taken an analytic approach through the identification of defects by experts or descriptive measurement of conformance to a target. However; consumer opinion must be one benchmark for good quality. Yet this also presents problems. Consumers are rarely analytical they perceive poducts in an integrated fashion, show halo effects and use language imprecisely. Product strategists would do well tofocus on specific aspects of consumer appeal. A comp-ehensive a@roath to quality entails a consideration of consumer neeo3 and expectations, careful translation of consumer language, and optimization of attributes through good sensory measurement. The quality of mercy is not straind It droppeth as the gentle rain from heaven Upon the place beneath Portia, The Merchant of Venice, IV: 1 The late Dr Morley Rare, Director of the Monell Chemical Senses Center, was fond of arranging social gatherings for his staff at which the more philosophical issues of taste and food acceptance would be discussed. Always an instigator for lively debate, one of his favorite questions was the following: Well, young man, what is your definition of quality? As a young scientist who was expected to operationalize and measure the conceptual, I was embarrassed by this question. Rare was wise in his choice - this troubling word can produce an intellectual minefield. My position on the matter is that no single definition of quality is entirely satisfactory, in spite of the firmly held opinions one reads in this volume. Quality is complex and multidimensional, contains both subjective and objective components, is situation-specific, and is fluid and dynamic across time. While quality must at some point
191

refer to consumer perceptions, objective characteristics are also important (a point of disagreement with Cardello in this volume). All other things being equal, a food with better nutritional properties has higher quality regardless of consumer perception (cf. Romans 1: 20 Cods invisibh [sic] qualities). To ignore objective aspects of quality is to deny objective reality. A second problem is that attributes of a food are considered desireable and functional depending on the situation. The wine and cheese I choose for a summer picnic are different than those I choose for a midwinter party. Finally, consumer perceptions are not static in time or consistent across people. Segments exist, and psychophysical approaches that look only at measures of central tendency on relative-to-ideal scales are bound to miss different consumer groups. People change their ideas about quality with experience. As we learn to appreciate fine wines, our definitions of whats appealing will certainly evolve. These issues pose strong challenges to the measurement and optimization of food product quality. The interplay of the senses and food quality was reviewed in a symposium given at the 1990 meeting of the Southern Association of Agricultural Scientists, published as a collection in the Journal of Food Quality. In the lead article, Civille (1991) paraphrases Plsek [ (1987) see also Garvin, 19841 in listing the dimensions of quality and how they relate to food. These dimensions are easily related to durable goods, to manufactured items like cars and are shown in Table 1. Civille discusses how these dimensions apply to food quality. Perfonance includes sensory attributes as well as nutrition and wholesomeness. This reminds us that there is more to food quality than merely sensory quality. Consumers expect food to be safe and provide
TABLE 1. Dimensions of Quality Performance Features Conformance Reliability Durability Serviceability Response Aesthetics Reputation

192

H. Lawless

nutrients. Features might include attributes like convenience in preparation. Conformance Civille suggests is a match between product concept, consumer expectations, and what the product delivers. Reliability concerns the degree to which the product is consistent upon repeat tastings or repeat purchases. Durability may imply a good shelf-life. Smwiceabilityfor a food may imply ease of preparation or recipe tolerance -wide limits on the abuse of the product during cooking due to failure to measure accurately or follow preparation directions. Response would involve customer support such as a telephone hot-line. Aesthetics include not only sensory attributes but package design. Finally, reputation is the continuing goodwill of consumers toward the integrity of the brand name. Obviously, there is a better fit of some of these concepts to durable goods like cars or home appliances. However, the list serves to remind us that sensory quality, as perceived by a person in a single sensory test, is only one aspect of a much larger picture. Furthermore, the perception of quality is not static. Factors such as the consistency of delivery will engender consumer loyalty in the long run - the time dimension is very important. The next section will focus on the issue of sensory quality per se, and begins with some definitions and their inherent problems.

QUALITY

AS CONSUMER

APPEAL

Any definition of sensory quality runs into problems sooner or later. The word has different meanings in different circumstances. The most simplistic approach is to look at quality as a unitary characteristic indicated by market performance. Shewfelt (1991)) in paraphrasing Kramer (Kramer & Twig, 1970) reminds us that quality is a composite of those characteristics that affect the degree of acceptability to the buyer. If consumers are the ultimate arbiters of whats good and whats poor, they will cast their vote by what they purchase. However, most culinary artists will be quick to point out that the product that sells the best to the public is rarely deemed high quality by people who have experienced more interesting flavors and textures in their food. Examples abound. Many mass-produced jug wines sell in huge volumes, yet few people would regard them as being wines of the highest quality. The same holds for mass-produced baked goods like white breads, which are functional in some circumstances (toast, childrens sandwiches) but are rarely accepted as the equal of home-baked or small production breads. Furthermore, the consumer responds not to quality in a vacuum, but to the quality-to-price ratio, which often determines value or perceived utility, and thus market performance. A second benchmark is blind-labeled consumer acceptance.Taking out the price and brand information

might lead to an evaluation of consumer appeal that is independent of economic concerns and based strictly on sensory factors. This is simply what mainstream sensory evaluation of consumer acceptance measures we give volunteers the chance to register their sensory pleasure or distaste for a product, as if there was no overriding frame of reference or price concern in the test. Many market researchers would find this ludicrous. Of course, all products are evaluated within a context, even if it is merely the context of the usual fare served in the sensory test facility. Human beings are natural comparators and always bring a frame of reference to a product test. A second measurement problem concerns whether the consumers in the test are representative of a valid reference group. In our sophomoric statistical training, we are repeatedly given the assumptions about random samples, but those are never achieved in a sensory acceptance test. A truly random sample would imply an equal probability of sampling all those present in the target group (for example, users of the product). AI1 experimental samples have some convenience aspect (e.g. people in shopping malls, people with telephones) so they are hardly random. Do we want a random sample anyway? Who says that John Q. Public is capable of judging quality? Since product appeal is judged only from a frame of reference, the average consumer may not have access to or experience with the finer, more complex, better crafted examples of culinary practice. Given such experience, the average consumer would come to recognize the appeal of the upscale or hand-crafted product, and eschew their previous choices. So perhaps the appeal to the public is not entirely satisfactory.

QUALITY

AS EXPERT

OPINION

This brings us to the traditional definition of quality, which is that quality is what experts say it is. We would not ask the public to score the quality of the performance of a figure skater or rate the participants in a gymnastic exhibition, so why would we have them be the arbiters of what a fine cheese or great wine should be? The definition of quality by connoisseurs is particularly annoying to those who practice sensory evaluation since the very methods they employ were designed to supplant the historical reliance on expert judges. Traditional definitions of quality for commodities have focussed on the identification of defects, and deviation from some ideal. Such methods are rooted in the historical traditions of quality grading by government inspectors. These programs were attempts to assure the consumer that products or production practices had been audited by someone outside the selling company and were found to be consistent with traditional expectations. The notion of typicality or resemblance to a prototype is not without its discomforts as well. How do

Dimensions of Quality

193

we know what is typical or ideal? Since foods change over time and have a limited shelf-life, the standard for quality is necessarily a mental one, which means it is kept in the minds of those with longstanding experience in evaluating the product in question. The only way to pass along this knowledge is through apprenticeship, leading to a fraternity of expert quality graders. The roots of their original knowledge base are lost in the arcane and mysterious traditions of smaller scale food production in the last century. Such systems have been roundly criticized for decades by workers in sensory evaluation, (6 Mahony, 1979; Side1 et al, 1981) and yet their traditions persist via the insularity of product-oriented professional organizations. The criticisms fall into four areas: lack of correspondence between experts and consumer opinion, lack of applicability to new engineered foods, lack of a psychophysical basis for measurement, and mixing of test objectives. What experts consider good or bad is often not aligned with consumer opinion (McBride & Hall, 1979). Even the status of defects may not attain universal agreement when it comes to consumer appeal. One persons defect may be anothers marketing bonanza. Substandard glue leads to the invention of the Post-it note. Cottage cheese with shattered curd may be more suitable than normal (good quality) cottage cheese for stuffing a blintz. This raises the issue of marketing niches and consumer segmentation, and creates problems for orthodox definitions of product standards. Newly designed foods present challenges to those who have a standard or ideal based on traditional production and ingredients. Is it fair to use a standard quality scoring system for vanilla ice cream to evaluate the quality of low fat frozen yogurt? What about goats milk ice cream? Do we need new quality systems for each new product? If there is no tradition for a product, we have no experts to consult, so consumers will have to be the arbiters of goodness. However, few food processors or academic labs have the resources to investigate all the variables that will affect consumer appeal, and furthermore, that may be a moving target anyway. Consumer familiarity may breed a whole new set of criteria for what is expected and what is familiar (the supplanting of shaving creams (foams) by shaving gels is a good example). Truly innovative products may face a long period of consumer adjustment. Quality becomes a moving target. The measurement problems inherent in quality grading arise from the practice of giving an overall score (a unidimensional scalar number) on the basis of multidimensional changes in product character. For example, a common quality scale for cake has the following grading points: IO-moist cake, &slightly dry, 6-gummy, 4-very dry. One wonders how gummy got inserted in a dryness scale. Obviously, such scoring guides do not lead to statistical treatment. The average

of a score of 4 and a score of 8 would be 6, implying the nonsensical result that gummy (soft and cohesive) is an average of dryness scores. This clearly has no basis in psychophysical measurement, where numbers are assigned to represent increases and decreases in perceived stimulus intensity along a single perceptual dimension. The terms used to describe defects may be phenomenologically complex since they will often refer to root causes, rather than perceptual experience. Such concepts are difficult to master, and once again, are multidimensional. A good example is the term oxidized in dairy product judging (similar to rancidity, as used in cereal science and studies of meat lipids). As milk becomes more and more oxidized, it will turn from producing sensations like cardboard or paper, to tallowy and painty, and finally to a fishy character. Each of these would be scaled as a separate variable in descriptive analysis, yet the dairy judge lumps them together as examples of oxidized flavor. Extensive work by Claassen has shown that complex defect-oriented terms are often not as discriminating as simpler unidimensional associative terms and that they are less well correlated with consumer opinion (Claassen & Lawless, 1992; Lawless & Claassen, 1993).

QUALITY

AS CONFORMANCE

In the quality control of durable goods and many nonfood products, it is confmnce to specifications that determines quality. Since physical measurement and mechanical performance are easy to quantify, most quality control approaches involve the setting of simple deviation limits in order to reject bad products. Limits may be set on the basis of statistics such as manufacturing variability (a common but curiously circular approach to validation) or by some externally imposed standards such as management philosophy or consideration of consumer complaints. It is also possible to assess conformance to sensory measurements, provided that reliability and accuracy are good enough. Good enough usually entails a trained panel using very specific attributes. In other words, a descriptive analysis. This approach was a primary focus of the article by Civille (1991) and is a major technique described in the book on sensory evaluation and quality control by Muiioz et al. (1992). These authors use the Spectrum approach to descriptive analysis. In the Spectrum descriptive procedure, panelists are calibrated not only to the conceptual referents of the attribute words, but also to intensity references for all scales. This is similar to the high level of intensity calibration in the texture profile method (Brandt et d., 1963), where individual points on the scales are illustrated by reference products. If this level of calibration is attained, it would seem possible to use descriptive ratings much like any

194

H. Lawless diet soda to the sugared version, some prefer artificial dairy dessert toppings to old-fashioned whipped cream and some prefer skim milk to whole milk, because it is more refreshing. The approach of using descriptive analysis to look at conformance to specifications must also rely on consumer opinion at some point to provide its validity check. The relation of trained panel data to consumer opinion presents statistical challenges. Sampling concerns such as range effects are important in design. For example, in conjoint measurement there is a well known effect of the number of levels chosen for the attribute to affect the resulting utility weights (Wittink et al, 1989). Second, individuals may change their minds about what to pay attention to. Schutz (1988) proposed that the appeal of food products is to some degree determined by their appropriateness for a given situation. This raises the possibility that quality reflects suitability for a purpose. Quality does not exist in a vacuum, but can only be appreciated if one understands the frame of reference of the end user. I may produce very high quality cheddar cheese by all traditional standards, but if I want to make nachos by melting that cheese on my corn chips, a processed cheese analog might suit better (cheddar does not always melt uniformly). Intended uses and appropriateness for a specific purpose should always be probed in consumer studies. The possibility of measurement contamination is a potential problem here. A common maxim of questionnaire design is that the more time you spend on an issue, the more salient it becomes in the respondents mind. This is the primary justification for asking consumers overall opinion of a product before you probe any specific characteristics - they might not have considered those specific characteristics until you brought them to mind. Another maxim of questioning is to avoid the double-barreled question, e.g. Do you think ice cream is wholesome and nutritious? In spite of the Boolean logical conjunction (and) in the sentence, consumers may decide to answer on the basis of wholesomeness or on the basis of perceived nutritional value, and you wont know which is important without further questioning. However, we often ask consumers complex questions we think are related to overall appeal such as how refreshing or how creamy a product is perceived to be. We know from the work of Kokini and others that creaminess is a function of lubricity, viscosity and smoothness (lack of particles) (Kokini & Cussler, 1983; Kokini et al., 1984). Aromatic qualities associated with dairy products may also affect judgments of creaminess (Lawless & Clark, 1993). How do consumers weight these different factors? Well, in a single experiment, its much like a double-barreled opinion question - its up to them, and you dont know their weighting scheme without further probing. The act of probing itself may affect the answer you get, so we get stuck in a Heisenberg-like conundrum of the measurement process affecting the measured quantity.

other analytical tool or instrumental measure and compare product ratings on specific attributes to predefined tolerance limits. A related, but somewhat simpler approach, was taken by Gillette and Beckley (1993), where scales may be used to indicate too little or too much of an attribute relative to the standard product. As in the case of quality grading, an overall score is given. In addition, the just right scales can be used by panelists as diagnostics, to indicate aberrations that resulted in lower overall scores being given. This kind of technique is rapid and simple to perform, and suitable for in-plant quality control efforts. The just-right or relative-to-ideal scales have a basis in the psychological literature (Booth et al., 1983; Booth et al, 1985). The primary difficulty with these approaches is the time and cost necessary to train and calibrate the panelists. In the case of the descriptive approach, training may be extensive. Also, turnaround time for analyses will be cumbersome, unless the data handling is automated. Finally, since the data rely on the assembly of an entire panel, they may be difficult to use at all times when a quality audit is required, e.g. on night shifts in the manufacturing location. Even for simpler diagnos tic procedures, the panelists still must be trained as to the nature of the standard product, to fix it in their memory. It also helps to have the sample present in every evaluation, which is sometimes difficult. Finally, these panelists must also understand the range of acceptable variation. To do this they must undergo a training procedure that shows the range of expected variation in every important attribute.

CONSUMERS CHALLENGES

REVISITED: SOME AND DIFFICULTIES

What choices are left to serve our need for a definition of quality? Experts become impractical in a milieu of thousands of processed food products. As suggested above, we cannot have experts to define quality for all forms of ice cream, frozen yogurt, goats milk ice cream and every other frozen dessert. In a practical sense, they probably could serve to reject unsafe or unsalable products from those that are salable, as is done in fish inspection. But it would certainly be impossible to differentiate whats good from whats superior for so many product types. A common anecdote among the cheese graders in my department is that when it comes to judging specialty cheeses, it boils down to a matter of their opinion (the same individuals are quite rigid about standard, traditional commodities like cheddar). So once again we are left with consumer opinion as a benchmark. Using consumer data raises a number of difficulties and challenges. First, consumer segments exist. Some people prefer

Dimensims

of Quality

195

Quality as an integrated impression


Another challenge that we get from using consumer opinion is that product quality is an integrated impression. Consumers are very good at arriving at an overall attitude, but they are poor analysts. They are prone to showing halo effects and it is questionable whether their diagnostic questionnaire information is of much value for the assessment of individual products (Lawless & Claassen, 1993). They simply find it hard to relate to product quality as if it were a collection of independent characteristics. Figure 1 shows a typical halo effect in which the addition of a minute amount of aromatic flavor appears to change sweetness, texture and even liking for a product. Under most circumstances, consumers treat individual attributes on a questionnaire as if they were interrelated. An example of conceptual fuzziness is seen in Fig. 2, where the profile of a fragrance is compared among a trained panel and two consumer panels. The trained panel is very specific in their use of attribute descriptors, while consumers are very diffuse (see Murioz & Chambers, 1993), in this case using all available terms to describe the smell. Some traditional quality evaluation schemes attempt to dissect products, assigning quality points for appearance, aroma, flavor, etc. The U. C. Davis 2@point wine scoring system is one example (Amerine & Roessler, 1981). But if consumer opinion is a global impression, does it make sense to do this dissection? Furthermore, where do the attribute weightings (2 points for appearance, 4 points for flavor, etc.) originate? Such details are often lost in the arcane practices of expert judges. A simpler approach is to use the overall impression of consumers as the required task. This was the approach of Craig Goldwyn at the Beverage Testing Institute (Goldwyn & Lawless, 1991). He decided to use small panels of experienced wine evaluators, but only to ask for their hedonic impression, a simple rating on a scale 1
13 r ;11 *
l l

Plantpanel
(untrained)

Office panel (untrained)

-.-.

-.-,-

-,-.

Trained Panel

FIG. 2. The descriptive profiling of a popular air freshener fragrance by trained and consumer panels.

,
0 ADDED VANILLA n CONTROL MILK * DIFFERENT
FROM CONTROL, Paired t-test, N = 19.

+ P F 2

7 5

5 s F/i 3 1 SWEET CREAMY THICK LIKE

SCALED

ATTRIBUTE

FIG. 1. The addition of a barely perceivable amount of vanilla flavoring to lowfat milk (1% fat) produces significant changes in consumer ratings for sweetness, thickness and creaminess (from Lawless & Clark, 1993, with permission).

for liking vs. disliking much like the commonly used 9-point hedonic scale. To the extent that the judges form a cultural community whose members agree on whats likable and whats not, this system seems to work well. This tendency to perceive a global impression of a product, combined with the imprecision in consumer language presents yet another challenge to those who want to measure quality on the basis of consumer appeal. Here is the dilemma: more reliable measurement can be attained through the use of trained panels in techniques like descriptive analysis. However, these trained panels at some point become unrepresentative of consumers. So there is a choice to be made - precise measurement or valid measurement, and they probably cannot both be attained in a single experiment. Of course, consumer opinions can be correlated to trained panel responses, and the regression of consumer acceptance against descriptive panel attributes has a long history in sensory evaluation. Given the new and more powerful techniques for multivariate data reduction and relation such as partial least squares, the relationship of data sets may become more commonplace, easier and more accurate (Murioz & Chambers, 1993).

196

H. Lawless

One attack on this multidimensional problem might be to work directly from data that captures an integrated impression. Multidimensional scaling approaches look at some global response (overall similarity, preference) and construct spatial models of products (Schiffman et a/., 1981). Dimensions in the model or vectors in the space may recover important quality factors that differentiate products. However, multidimensional scaling is a technique that has not always lived up to its promise. Early demonstrations of the feasibility of this approach relied on simple stimulus domains such as Morse code symbols, colors or tones, where two or three obvious dimensions were salient. Attempts to apply multidimensional scaling to more complex problems such as defining the dimensions of odor quality have not been so successful. In the case of odor quality, hedonic and intensity dimensions would usually result (Davis, 1979), a finding that is of little utility.

CONCLUSIONS
What then should be a reasonable strategy for product scientists who want to insure product quality? First, a certain amount of intellectual honesty is required to admit that quality is more than a single issue. H. F. Pratt wrote The term quality, noun, has been defined by the American Society for Quality Control as the totality of features and characteristics of a product that bear on its ability to satisfy given needs. As such, the term is useless (Pratt, 1987, p. 20). This essay echoes his lament that the totality of features is not an actionable concept. An appreciation of the factors that contribute is essential. Also, one must adopt the consumers frame of reference. This includes an appreciation of segments of consumers who may differ as to what they think is important and how the product is to be used. The foibles of consumers bear constant watching-consumers are poor diagnosticians, the language they use is fuzzy, and they tend to treat products as integrated wholes, rather than collections of attributes. Can overall quality be measured? With great difficulty. Quality as a whole can only be understood metaphorically (mercy is like rain). Individual aspects of quality can be quantified and optimized, but even this will not guarantee product success.

AN AFTERTHOUGHT: QUALITY AND FRANCHISE LOYALTY


In the broad view, quality is a complex proposition. It is frequently used in current business terminology and has become one of the watchwords in the expanding international competitive arena. US automotive manufacturers, given a sound thrashing by the Japanese in the 197Os, began to examine the Japanese use of quality improvement practices deriving from Deming and others. Thus quality circles and total quality management have become part of the everyday corporate jargon. This is using the term in a very global sense. It refers to the combination of characteristics that engender goodwill and confidence among consumers in the companys brand image. Such was the idea behind Portias defense of Antonio from the debt claim of Shylock. Metaphorically, the essenceof mercy is that it is not constrained. This is powerful imagery, and quality itself is a powerful symbol. Quality is bound to consumer perception, opinion and attitude. This was captured in the Profit Sharing Day speech by H. F. Johnson, Sr., in 1927 (Johnson, 1988): When all is said and done, this business is nothing but a symbol and when we translate this we find that it means a great many people think well of its products and that a great multitude has faith in the integrity of the men who make this product. The goodwill of the people is the only enduring thing in any business. It is the sole substance . . . The rest is shadow!
While sensory professionals attempt to dissect quality into more useful and specific characteristics, the global term as used in corporate public relations is likely to persist.

REFERENCES
Amerine, M. R. 8c Roessler, E. B. (1981). Wines, 7X& Sensory Eualution (Second Edn). W. H. Freeman, San Francisco. Booth, D. A., Thompson, A. & Shahedian, B. (1983). A robust, brief measure of an individuals most preferred level of salt in a foodstuff. Appetiti, 4,301-12. Booth, D. A., Conner, M. T., Marie, S., Griffiths, R. P., Haddon, A. V. & Land, D. G. (1985). Objective tests of preferences amongst foods and drinks. In Meamrement and Determinants of Food Habits and Food preferences, ed. J. M. Diehl & C. Leitzmann. Stichiting Nederlands Instituut voor de Voeding, Wageningen, The Netherlands. Brandt, M. A., Skinner, E. Z., & Coleman, J. A. (1963). Texture profile meth0d.J. Food k-i., 28,4049. Civille, G. V. (1991). Food quality: Consumer acceptance and sensory attributes.J. Food @al., 14, l-8. Claassen, M. & Lawless, H. T. (1992). Comparison of descrip tive terminology systems for sensory evaluation of fluid mi1k.J. Food Sci., 57,596600,621. Davis, R. G. (1979). Olfactory perceptual space models compared by quantitative methods. Chmn. Senses, 4,191-206. Gatvin, D. A. (1984). What does product quality really mean? Sloan Management Rev., 26,25-43. Gillette, M. G. & Beckley, J. P. (1993). Sensory methods for in plant quality control. Report of the Cornell Agricultural Experiment Station at Geneva. Goldwyn, C. & Lawless, H. (1991). How to taste wine. ASTM Standardization News, 19(3), 32-7.

Dimensions of Qua& Johnson, S. C. (1988). The Essence of a Family Enterprise. Curtis Publishing, Indianapolis. Kramer, A. & Twigg, B. A. (1970). Quality Controlfor the Food. Industry, Third Edn. Van Nostrand/AVI, New York. Kokini, J. L. & Cussler, E. L. (1983). Predicting the texture of liquid and melting semi-solid fo0ds.J. Food Sci., 48, 1221-4. Kokini,J. L., Poole, M., Mason, P., Miller, S. & Stier, E. F. (1984). Identification of key textural attributes of fluid and semi-solid foods using regression ana1ysis.J. Food Sci., 49,47-51. Lawless, H. T. & Claassen, M. R. (1993). Validity of descrip tive and defect-oriented terminology systems for sensory analysis of fluid mi1k.J. Food Sci., 58, 108-12, 119. Lawless, H. T., & Clark, C. C. (1992). Psychological biases in time intensity scaling. Food Tech., 46(11), 81,84-6,90. McBride, R. L. & Hall, C. (1979). Cheese grading versus consumer acceptability: an inevitable discrepancy. Austr. J Dairy Tech., June: 668. Murioz, A. M. & Chambers, E. C. (1993). Relating sensory measurements to consumer acceptance of meat products. Food Tech., 47(11), 128-31, 134. Murioz, A. M., Civille, G. V. & Carr, B. T. (1992). S.en.soly Eualuation in Quality Control.Van Nostrand Reinhold, New York. 6 Mahony, M. (1979). Psychophysical aspects of sensory analysis of dairy products: a critique.J. Daily Sci., 62, 195462. Plsek, P. E. (1987). Defining quality and the marketing development interface. QualityProgress,20,28-36. Pratt, H. F. (1987). Some thoughts on the term quality. ASTM Standardization News, 15( lo), 20. Schiffman, S. S., Reynolds, M. L., & Young, F. W. (1981). Introduction to Multidimensional Scaling. Academic Press, New York. Schutz, H. G. (1988). Beyond preference: Appropriateness as a measure of contextual acceptance of food. Food Acceptability, ed. D. M. H. Thomson. Elsevier Applied Science, London, pp. 115-34. Shewfelt, R. L. (1991). Editorial: perception of food quality by the five senses.J. Food QuaL, 14, v-vi. Sidel, J. L., Stone, H. & Bloomquist, J. (1981). Use and misuse of sensory evaluation in research and quality control. J. Daily Sci., 61, 2296302. Wittink, D. R., Krishnamurthi, L. & Reibstein, D. J. (1989). The effect of differences in the number of attribute levels on conjoint results. Marketing I&%x, 1, 113-23.

197

COMMENTARIES
Perceived People quality is a complex and difficult conceptual concept. definidiffer considerably in their

tions of the concept, operationalize

and as a corollary

in how they

it. Lawless discusses some of these approaches at length, mainly emphasizing their weaknesses and shortcomings. The reader, who probably has an interest in measuring (perceived) quality, might feel slightly depressed after reading Lawless contribution. Basically, his point can be summarized as: Perceived quality is the eye of the beholder (the consumer), it is highly subjective and complex, it represents perceived fitness for use (but the degree of fitness may differ from one situation to the other), and consumers have great difficulty in expressing in some de-

tail what they consider high quality. As academic marketing and consumer behavior researchers, we could not agree more with Lawless notion that consumer opinion must be the benchmark for what constitutes good quality and that perceived quality may differ among consumers and from one situation to the other. Lawless discusses three approaches to quality measurement (consumer appeal, expert opinion, and conformance to specification) and gives many reasons why each of these three approaches falls short, some of which are very practical and can easily be improved (e.g. phrasing in questionnaires) and others are more fundamental. Would this imply that quality perception cannot be measured? We are far less pessimistic about quality measurement than Lawless. We agree with Lawless that the multidimensional nature of the quality concept provides a great challenge to those trying to measure it. Fortunately considerable progress has already been made in this respect (e.g. Steenkamp, 1989). Measurement of perceived quality is admittedly a difficult task, particularly if it is approached from a sensory perspective. Sensory quality is only one aspect of quality perception although sensory researchers tend to give it a highly central place. If we accept that quality perception is highly subjective and variable in time, we should leave the position that perceived quality is a characteristic of the product or resulting from its consumption in any static way. Rather we should conceptualize quality as the resultant of an interaction between the consumer and the product. The nature of this interaction goes way beyond the (relatively) simple interaction between characteristics of the product and the human senses (as it is often operationalized by sensory researchers). It includes the interaction between the product and the consumers cognitive structures, the intended usage situation, and possibly other factors. It is our conviction that, if we are willing to leave the position that quality is something inherent in the product or statically resulting from its consumption, and accept that perceived quality is a resultant of complex interaction between the product and other aspects of consumer behavior, we need not be overly pessimistic about its measurement. We would like to refer the interested reader to Steenkamp (1989) and our own contribution to this special issue, for more detailed information about how perceived quality can be made measurable. Peter A. M. Oude Ophuis Hans C. M. Van Trijp

The essence of food quality: goodness and mercy


It is difficult to argue with the basic contentions of Lawless paper-that food quality is a complex term, that the concept poses difficult measurement problems,

198

H. Lawless

and that, when defined in terms of the tot&y of features of a product, food quality becomes an inactionable and meaningless concept. Moreover, his wellbalanced presentation of the issues provides solace for researchers of varying points of view. However, the nature of his critique is such that he does not offer a unitary, schemapiric view of food quality. Rather than having a better understanding of Lawless own position on food quality after reading the paper, I feel as if he has chosen to gingerly walk the fence on the issue. While this approach is useful for identifying critical issues, it may be unsatisfying to the reader seeking an operationally based and internally consistent approach to this difficult problem. The most serious problem with Lawless exposition is the lack of a definition of food quality. In his opening paragraph he recalls his feelings of embarrassment when as a young scientist he was asked by the late Morley Kare, What is your definition of quality? Unfortunately, Lawless never reveals his reply to Dr Kare, nor does he provide an operational definition of food quality elsewhere in the text. His only reference to definition appears in the conclusions, where he dismisses any definition of food quality that embodies the notion of totality of features. It is not clear from his omission whether Lawless is adopting the philosophical view that Nevertheless, quality or goodness is indefinable. so unsaddled by definition, Lawless moves freely and effectively from one position on the topic to another from the experts view to the quality assurance view to the consumers view. In lieu of definition, he appeals to Shakespearean metaphor for insights into the nature of food quality. Unfortunately, while Shakespeares words, i.e. that the quality of mercy is not strained and that mercy is like rain have literary appeal, science demands an operational definition of both llzercyand food quality before the veracity of Shakespeares and Lawless statements can be assessed. In addition to the lack of definition, there are two other points raised by Lawless paper that are worthy of discussion. One of these is practical and relates to the empirical aspects of how one measures quality. The other is philosophical and concerns the inherent nature of value words like quality and goodness.

they do not purchase due to their expense. Purchase behavior is a poor index of both liking and quality. Lawless goes on to note that culinary artists are quick to point out that the product that sells the best to the public is rarely deemed high quality by people who have experienced more interesting flavors and textures in their food. At first glance, the statement appears to operationalize the special characteristic of experts in terms of their greater exposure to interesting flavors and textures. However, upon closer examination the statement can be seen to be an example of circular reasoning. Homeless people who eat from restaurant dumpsters are routinely exposed to interesting flavors and textures, yet they are certainly not considered food experts. This is because the phrase interesting flavors and textures in this statement does not refer to any type of interesting flavors or textures, but rather, it refers to those associated with better quality foods. Thus, the argument becomes circular and meaningless: Culinary experts believe that high quality foods are those which culinary experts believe are high quality. The entire notion of expert judges for food quality smacks of the tertiary and archaic English definition of quality as aristocratic or of high social position. Perhaps the best pima facie argument that Lawless uses for the use of experts to judge food quality is his analogy to the judging of figure skating and gymnastic competitions. Here, Lawless correctly states that we would not think of using the public as judges. However, the judging of figure skating and other sports competitions is more representative of the definition of quality as conformance to standards. Such judges are judging precision, i.e. the degree to which the skaters (gymnasts) leg is straight or bent upon landing or whether a pitched ball passes through a circumscribed spatial area. I have no doubt that training in these areas will produce more reliable and, perhaps, more valid data. However, the standards by which these judgments are made are best established according to public notions of what constitutes good quality in skating, diving and the like.

Thephilosophical issue: objectivism vs. subjectivism


The other major issue that Lawless paper raises concerns the philosophical notion of quality or goodness. Lawless eloquently states that the essence of quality can only be described metaphorically (mercy is like rain). His choice of wq as the object of the metaphor is not accidental, for mercy, like goodness, is an abstract value judgment. Indeed, moral philosophers have long dealt with the issues surrounding the nature and definition of goodness (and mercy). What Lawless and, indeed, all who have contributed to this volume have done is to try to assess the concept of quality using principles and approaches that have been derived through science. However, as one notable philosopher has pointed out, questions as to what is good or bad on its own account...

Theempirical issue: experts vs. consumers


Concerning the practical issue of how to measure food quality, Lawless seems to come down on the side of the consumer, especially in his comment that one must adopt the consumers frame of reference when dealing with food quality. I wholeheartedly agree. However, in his critique of food quality as consumer appeal, Lawless states that If consumers are the ultimate arbiter of whats good and whats poor, they will cast their vote by what they purchase. This is not necessarily true. There are many foods and wines that people enjoy but that

Dimensions of Quality

199

lie outside the domain of science (Russell, 1935). Thus the concept of quality, which concerns what is good or bad, is also beyond the objectivism of science. Therein lies the problem. Russell, in keeping with the philosophical view that all value judgments, like goodness, quality and mercy, are subjective, states that prima facie, anything that we all desire is good and anything that we all dread is bad. His use of the word all clearly positions these concepts within the domain of popular opinion. In fact, Russell goes on to argue that every attempt to persuade people that something is good (or bad) in itself, and not merely in its effects, depends upon the art of rousing feelings, not upon an appeal to evidence. This emotive view denies any descriptive or cognitive meaning to these terms. Lawless, in his essay, reflects this approach to quality when he states that consumers, with proper experience with finer, better crafted food items, may come to recognize the appeal of these products. Similarly, the opinions on food quality found in this volume can be seen to appeal more to emotions than to reason, since much of the evidence can be

argued to fit almost any model of food quality. The emotive or subjectivist view has several advantages over the objectivist view when applied to food quality. On the one hand, we dont have to decide who the expert is. In the second place, while one person may view a product or service as high quality, another person may view it as low quality, yet neither can charge the other with intellectual error. Lastly, the subjective view unshackles the concept of quality from the constraints imposed by omnipotent experts and other guardians of Truth and Goodness. In the subjectivist view, quality, at last, becomes like mercy, unrestrained and free to drop as the gentle rain from heaven upon the place beneath. Armand V. Cardello

REFERENCE
Russell, B. (1935).
Press, London.
Religion and Science.

Oxford University

Das könnte Ihnen auch gefallen