Are Abstract - Product Qualifications More Subjective? a Test of the Relation Between Abstract - Ion Level and Opposite Naming

Dirk Snelders, Delft University of Technology
Jan P. L. Schoormans, Delft University of Technology
ABSTRACT - The hypothesis is investigated that the more abstract a product qualification, the more diverse opposites consumers can give for it. Content analysis on a number of qualifications and their opposites (N=72) shows that there is indeed such a relation. These results show that abstract product qualifications can lead to a higher degree of subjectivity in consumer responses because they are related to a more personal background of consumers.
[ to cite ]:
Dirk Snelders and Jan P. L. Schoormans (1999) ,"Are Abstract - Product Qualifications More Subjective? a Test of the Relation Between Abstract - Ion Level and Opposite Naming", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 584-587.

Advances in Consumer Research Volume 26, 1999      Pages 584-587


Dirk Snelders, Delft University of Technology

Jan P. L. Schoormans, Delft University of Technology


The hypothesis is investigated that the more abstract a product qualification, the more diverse opposites consumers can give for it. Content analysis on a number of qualifications and their opposites (N=72) shows that there is indeed such a relation. These results show that abstract product qualifications can lead to a higher degree of subjectivity in consumer responses because they are related to a more personal background of consumers.


Consumer research often consists of collecting information about the product attributes that are valued by consumers. Different techniques are suggested in the literature to find out which attributes consumer use to judge products. Consumers can be directly asked what meaningful differences exist between brands or products (Reynolds and Gutman 1988) or they can be asked to cite the attributes on the basis of which they have sorted a list of brands or products in a rank order (Reynolds and Gutman 1988), a grouping task (Snelders and Stokmans 1994), or a triadic sorting (Kelly 1955). Projective techniques (Churchill 1991; Giannelloni and Vernette 1994) can be used to find attributes that consumers are not likely to come up with by thmselves. Finally, in order to find attributes that consumers only use implicitly, decompositional perceptual mapping can be used (Churchill 1991; Hair, Anderson, Tatham and Black 1995). The information collected through these techniques, i.e. a list of product attributes, can be evaluated for its importance or salience and thus guide the development of new products, brand names, or advertising campaigns.

In the case that consumers come up with very concrete attributes, like cars that only use X gallon per 100 miles, it is completely clear what is meant. Often however, the gathered information is more difficult to understand and to use. This is especially the case when the attributes are more abstract, like the reliability or the elegance of a car. These abstract product qualifications can lead to problems when they have to be compared with the answers of other consumers. For example, consumer A and B may both mention that a car is elegant, but mean something different with it. This can lead to the problem of subjectivity, which we will define technically as the problem of not clearly understanding what consumers mean when they qualify a product. One possible way to overcome this problem is to use the technique of opposite naming (ON).

ON is a questioning technique in which people are asked to name the opposite of what they have just answered, in order to get a clearer view of the meaning of the answer. The study of Riley and Palmer (1975) is an early example of a study in which the meaning of seaside resorts is uncovered through ON. Other examples of the ON technique in consumer research exist: American high class foods are opposed to low class foods (Levy 1981); store A is hated because of its contrast with store B (Thompson 1989); and business flights are opposed to vacation flights (Durgee 1985).

In the present paper, we want to find out to what degree ON can be used to clarify the meaning of abstract attributes. We propose that the abstraction level of the attribute will positively influence the diversity of opposites that consumers will name for it. There are two reasons for believing this. A first reason is provided by a theory about word meaning, called #Context Availability Theory’ (Kieras 1978; Schwanenflugel and Shoben 1983). This theory states that the information from prior knowledge that is associated with an abstract word is cued to a wider range of contexts (i.e. situations of use). A number of empirical findings support this assertion. First, abstraction ratings of words are negatively correlated with context availability ratings and positively with the rated number of contexts to which a word can apply (Schwanenflugel and Shoben 1983). Second, it takes longer to identify a picture of an object as an example of an abstract word than as an example of a concrete word, but only if the object is presented in isolation. When the picture displays the object in a meaningful context, then the difference in identification times disappears (Murphy 1991; Murphy and Wisniewski 1989). These findings show that the meaning of abstract words is less tied to specific contexts than the meaning of concrete words. When we assume that ON is a way of uncovering these different contexts, we can predict on the basis of this that abstract words have more opposites than concrete words. However, this assumption is disputable, because the availability and range of contexts that can be used to get to the meaning of a word are not necessarily related to a diversity of opposites that have to be thought up by a consumer in an ON task. For this reason, we turn to a second reason for why abstract attributes have a wider range of opposites.

A second reason to think that more opposites will be named for abstract attributes is that abstract qualifications of products are thought to be less utilitarian and more expressive of one’s personality (Hirschman and Holbrook 1982). In other words, abstract product qualifications are more subjective, in the sense that they express someone’s personality. If that is the case, one can argue that abstract attributes are highly personal, much in the way that Kelly (1955) proposed. For Kely, the answers of people can be understood as qualifications of "similarity [that] can only be understood in the context of difference" (Bannister, Burman, Parker, Taylor and Tindall 1994, p.73). For example, when someone qualifies a chair as comfortable, we can understand what this means if we know that this statement is constructed from the opposition comfortable rest vs. active work. Thus, ON can help here to show how a qualification (comfortable) derives its meaning from what it is not (active). Kelly assumed that these personal constructs arise from highly personal backgrounds. If this is true, we can expect that abstract product qualifications have more diverse opposites, as each opposite describes the consumer’s personal background against which the qualification can be understood.

Kelly’s theory about ON comes in hand to explain how abstract attributes can be characterized as subjective product qualifications that express the consumer’s personality. However, it does not make it more plausible that subjectivity is preferably expressed through abstract attributes. Therefore, we are left only with a popular belief, and the speculation of Hirschman and Holbrook (1982) that subjectivity is the result of the expression of the consumer’s personality through abstract words. In order to make the relationship between subjectivity and abstract attributes more plausible, we turn to another view of ON, which has lived in peaceful co-existence with Kelly’s personal construct approach for the last forty years. This view is the cultural structure approach by LTvi-Strauss (1958). For LTvi-Strauss, ON is more a means to uncover cultural structures than personal constructs. Opposites inform us about the #structures TlTmentaires’ or rudimentary structures (Greimas 1966), in which an answer acquires its meaning. According to LTvi-Strauss (1949), the reason why these structures are thought to consist of opposites is that they are rooted in the gestalt-like perception of objects as distinct from their background (e.g. Koffka 1935). This means that an answer has an opposite that serves as a background against which the meaning of the answer becomes clear. Thus, where Kelly’s personal construct approach assumes that the opposites are individual, developed through one’s personal life, the cultural structure approach of LTvi-Strauss assumes that the opposites are the manifestations of a collective culture. Within the cultural structure approach it may seem improbable that individual variation in ON can be meaningful, but that is not the case. Later findings from cross-cultural psychology show that people from highly individualistic cultures (like the U.S. or the Netherlands) tend to describe people in terms of abstract personalities, while people from highly collectivist cultures (like Turkey or India) describe people in terms of their concrete behaviors (Miller 1987; Schweder and Bourne 1984; Zwier 1998). This means that if consumers are expressing their distinct personality through product qualifications, we can expect that these will be abstract, rather than concrete. [Note that it is not a semantic necessity that the product characteristics that express personality can only be communicated by qualifying the product in abstract terms. The research of Zwier (1998), for example, shows that the same thing can be expressed as easily in concrete terms as in abstract terms. Thus, a characteristic of a product (e.g. a car) can be expressed both by concrete qualifications (e.g. runs 50 miles to the gallon) as abstract qualifications (e.g. is very economical). Therefore, the consumer's choice between concrete and abstract qualifications is a pragmatic choice-not a semantic one.]

To conclude, abstract attributes can be seen as more expressive of the consumer’s personality, and it is for this reason that these can be regarded as more subjective. If so, ON should lead to more diverse opposites for abstract attributes. In the present study we limit ourselves to a one-culture study, and within that culture, concrete attributes will have a limited number of opposites. In order to test the relation between abstraction level and ON, we compare the number of opposites we collect with the ON technique for concrete and abstract product attributes. This is expressed in the following hypothesis:

H1:  The more abstract a product attribute, the more different opposites respondents will give for the qualifiction.


Seventy-two respondents of a household panel were invited to the university to be interviewed on two product categories (telephones and clock radios). Ages ranged between 18 and 60, and half were men, half were women. Thirty-six respondents received an interview for the telephone, another 36 for the clock radio. Respondents were interviewed in three stages. At the beginning of the interview they were asked to name attributes of the products. Next, they were given the ON task and another, unrelated task about another product (the order of the last two tasks was reversed for half of the respondents).

After a brief instruction, the respondent was first given a brochure with general commercial information about the product category. After about one minute, the interviewer asked the respondent to name attributes of the product s/he would consider when buying the product (respondents were specifically told not just to consider what they had just seen in the brochure, but also what they were thinking of themselves). In the ON task, respondents received short instructions about the task and were then asked to pick out the five most important attributes (unless the respondent had given five or less attributes, in which case all attributes were used). Next, the respondent was asked four questions about each attribute. The first question (in Dutch) was "kunt u kenmerken noemen van een (telefoontoestel, wekkerradio) die het tegenovergestelde zijn van een (telefoontoestel, wekkerradio) met (attribuut X)" [can you name characteristics of a (telephone, clock radio) that are the opposite of a (telephone, clock radio) with (attribute X)]. This is the question that was used for further analysis. [The last three questions that were asked in the ON task dealt with how a product would be without the attribute (question 2), what one could do with a product without the attribute (question 3), and what someone who did not want the attribute would be like (question 4). Following Durgee (1985), these three questions can also be asked to get a better picture of the background of the attribute. However, in the present study, with its focus on attributes rather than whole products, many subjects did not understand these questions. In the case of question 2, many thought it was the same as question 1 and they did not answer it or gave the same answers as to question 1. Most subjects' response to question 3 was simply that they would not buy the product if it did not have the attribute, so there was almost no variation in it. With respect to question 4, a lot of subjects did not understand the question and because of that they did not give an answer to it.] In the remainder of the paper, the attributes are referred to as product qualifications, and the answers to question one are referred to as the opposites. All respondents were debriefed about the purpose of the experiment in a letter they received after the experiment had ended.


The respondents’ answers (both qualifications and opposites) were put on 442 coding cards for telephone, 553 for the clock radio (only counting the qualifications and the answers to question 1). For each product, two independent judges were asked to create answer categories that contained the same answers. Judges were asked to do this with precision, to avoid concrete and abstract words ending up in the same answer category. After each judge had individually categorized the answers, they discussed what to do with those answers that they had put in different answer categories. Categories were merged or split, according to the final agreement or disagreement among judges. For the telephone, 66 answer categories were created, and 92 for the clock radio. Among these answer categories, 31 distinct product qualifications were listed for the telephone, and 38 for the clock radio. The remainder of answers consisted of opposites.

The product qualifications given by respondents were rated by five judges on a seven point scale ranging from "zeer concreet" (very concrete) to "zeer abstract" (very abstract) (after Johnson and Fornell 1987). The ratings of the five judges were very consistent and, as a consequence, showed high alpha levels: for telephone, alpha=.936, for clock radio alpha=.875. This allowed us to consider the average scale values of the five judges as a measure of the abstraction level of the answer categories.

Two variables were created that varied over the qualifications that were named in the ON task: the average abstraction ratings of each qualification and the mean number of distinct answer categories that were given as an opposite to each qualification. The mean number of disinct answer categories was taken, to correct for the effect that if more people name the same qualification, more distinct opposites will occur as a result. Next, the average abstraction of each qualification was correlated with the average number of opposites of each qualification. The correlation between the abstraction level and average number of opposites was, for telephone .250, p (one-sided)<.088, and for clock radio .487, p (one-sided)<.001. So it turns out that there is a strong effect of the abstraction level on the number of opposites for clock radios, in that the more abstract a qualification is, the greater number of distinct opposites it can have. For telephones, the same trend exists.

Looking at the average number of opposites instead of all opposites is better because some people can give more opposites than others can and by averaging, we have given all respondents equal weighting. However, it can be argued that each person in the experiment was given the chance to name all the opposites that he or she knew, and could thus have exhausted all possible opposites that exist for that qualification. In addition, if concrete qualifications are listed by more people than the abstract qualifications, then the average number of opposites will downplay the real number of opposites that were given for concrete qualifications and thus make our test more liberal than it should be. Tests show that this is not the case: the correlation between the abstraction level of a qualification and its absolute number of opposites (regardless of how many respondents it had taken to produce that absolute number) was similar to the correlations that we found earlier. For telephone it was .220, p (one-sided)<.118, and for clock radio .472, p (one-sided)<.002.

Finally, a small group of purely idiosyncratic answers existed that could not be placed in a category: 52 for telephone, and 33 for clock radio. Many of these answers were nonsensical expressions, but there is a risk that the results of the ON task are affected by not considering these idiosyncratic answers (most often with no apparent meaning) as viable opposites. To control for this, the number of opposites was recalculated, now also containing the idiosyncratic answers. Again this did not affect the results very much: the correlation between abstraction level and average number of opposites for telephone are now .171, p (one-sided)<.178, and for clock radio it is .496, p (one-sided)<.001. We can thus conclude that H1 is partially supported, in that the more abstract an qualification, the more different opposites respondents can give for the qualification.


The research has shown that the abstraction level of product qualifications is linked to the number of opposites: more abstract qualifications are correlated with higher numbers of opposites. So, our study indicates that more opposites exist for abstract qualifications. This higher number of opposites can be seen as an aspect of subjectivity in consumer responses. Namely, the more opposites that can possibly be named for an attribute, the more the response is multi-interpretive, and the less it will be understood by everyone in the same way. So for this aspect of subjectivity, we can say that the abstract level of a product attribute has an influence on the subjectivity of the consumer response.

This conclusion may sound obvious, but it is not. First, we have to remind the reader that we did not find perfect correlations between abstraction and the number of opposites: for clock radio, the abstraction level accounted for 24% of the variance in the number of opposites, and for telephone this was only 6%. This means that we have found a statistical relationship, not a determinist one. Second, theories about abstract attributes are not equivocal about the higher subjectivity of abstract qualifications. Alternate explanations about the abstraction level exist where abstract attributes are explained in terms of hierarchical categorization (e.g. Johnso 1989). Within this view, abstract attributes are thought to be inclusive of more concrete attributes, but whether or not this leads to more subjective answers is left in the open.

Abstract product qualifications can be regarded as expressions of more personal (and thus more variable) relevance, and in that sense they are more like personal constructs. At the same time, however, it must be acknowledged that the whole concept of subjective personal constructs may be the exclusive product of individualistic cultures (Markus and Kitayama 1991; Triandis 1989). Thus, using abstract product qualifications in judgments may be a way for consumers to express that they are really unique, independent selves. But the idea of unique independent selves may itself be a cultural product that can best be understood in terms of structural anthropology. This means that personal constructs can be seen as a special type of cultural structure, where value is placed on variability and distinction.

To the extent that individualistic cultures place a high value on materialism, we can expect that products are used to express the individuality of consumers. Since abstract product attributes are thought to do this more than concrete attributes, we can expect abstract attributes to be mostly used to describe the most visible aspects of the product: form, color, eye-catching details, etc. However, the problem with these abstract attributes is that they are multi-interpretive, and this means that during the development of products and promotion campaigns, they do not easily find their way into the final product and promotions. It is more difficult for a team of product developers or campaign strategists to decide what to do with, for example, the elegance of a car than with the size of car bumpers. Even if consumers place a high value on elegance, team members may still be tempted to work on the bumpers at the expense of car’s elegance. Our results indicate that ON gives researchers more insight in the subjective meaning of product qualifications given by consumers. Especially when information is needed about the value of abstract product attributes ON is a valuable research tool.


Banister, Peter, Erica Burman, Ian Parker, Maye Taylor, and Carol Tindall (1994), Qualitative methods in psychology, Buckingham, UK: Open University Press.

Durgee, Jeffrey F. (1985), "Depthinterview techniques for creative advertising," Journal of Advertising Research (December/January), 237.

Giannelloni, Jean-Luc and Eric Vernette (1994), Etudes de MarchT, Paris: Vuibert.

Gilbert A. Churchill (1991), Marketing Research: Methodological foundations (5th ed.), Fort Worth, TX: Dryden.

Greimas, Algirdas J. (1966), STmantique structurale: Recherche de mTthode, Paris: Librairie Larousse.

Hair, Joseph F., Rolph E. Anderson, Ronald L. Tatham and William C. Black (1995), Multivariate data analysis (4th ed.), London: Prentice Hall.

Hirschman, Elisabeth C. and Morris B. Holbrook (1982), "Hedonic consumption: Emerging concepts, methods and propositions," Journal of Marketing, 46 (Summer), 92-101.

Johnson, Michael D. (1989), "On the nature of product attributes and attribute relationships," in Advances in Consumer Research, Vol. 16, ed. Thomas K. Srull, Provo, UT: Association for Consumer Research, 598604.

Johnson, Michael D. and Claes Fornell (1987), "The nature and methodological implications of the cognitive representations of products," Journal of Consumer Research, 14 (September), 214-228.

Kelly, G.A. (1955), Thepsychology of personal constructs (Vol. 1 and 2), New York: Norton.

Kieras, David (1978), "Beyond pictures and words: Alternative information processing models for imagery effects in verbal memory," Psychological Bulletin, 85 (3), 532-554.

Koffka, Kurt (1935), Principles of gestalt psychology, New York: Harcourt Brace.

LTvi-Strauss, Claude (1949), Les structures TlTmentaires de la parentT, Paris: Presses Universitaires de France.

LTvi-Strauss, Claude (1958), Anthropologie structurale, Paris: Librairie Plon.

Levy, Sidney J. (1981), "Interpreting consumer mythology: A structural approach to consumer behavior," Journal of Marketing, 45, 3 (Summer), 49-61.

Markus, H.R. and S. Kitayama (1991), "Culture and the self: Implications for cognition, emotion, and motivation," Psychological Review, 98, 224-253

Miller, J.G. (1987), "Cultural influences on the development of conceptual differentiation in person description," British Journal of Developmental Psychology, 5, 309-319.

Murphy, Gregory L. (1991), "Meaning and concepts," in The psychology of word meanings, ed. Paula J. Schwanenflugel, Hillsdale, NJ: Lawrence Erlbaum, 11-35.

Murphy, Gregory L. and Edward J. Wisniewski (1989), "Categorizing objects in isolation and in scenes: What a superordinate is good for," Journal of Experimental Psychology: Learning, Memory and Cognition, 15 (4), 572-586.

Reynolds, Thomas J. and Jonathan Gutman (1988), "Laddering theory, method, analysis, and interpretation," Journal of Advertising Research (February-March), 11-31.

Riley, Stuart, and John Palmer (1975), "Of Attitudes and Latitudes: a Repertory Grid Study of Perceptions of Seaside Reports," Journal of the Market Research Society, 17 (2).

Schwanenflugel, Paula J. and Edward J. Shoben (1983), "Differential context effects in the comprehension of abstract and concrete verbal materials," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9 (January), 82-102.

Shweder, R.A. and E.J. Bourne (1984), "Does the concept of the person vary cross-culturally?" in Culture theory: Essays on mind, self and emotion, eds. R.A. Shweder et al., New York: Cambridge University Press.

Snelders, H.M.J.J. and Mia J.W. Stokmans (1994), "Product perception and preference in consumer decision making," in Correspondence analysis in the social sciences, eds. Michael Greenacre et. al., New York: Academic Press.

Thompson, Craig J. (1989), "The role of context in consumers’ category judgements: A preliminary investigation," in Advances in consumer research, Vol. 16, ed. Thomas K. Srull, Provo, UT: Association for Consumer Research, 542-547.

Triandis, H. (1989), "The self and social behavior in differing cultural contexts," Psychological Review, 96, 506-520.

Zwier, Sandra (1998), "Patterns of language use in individualistic and collectivist cultures," dissertation, Amsterdam: Free University of Amsterdam.