Association Measures As Predictors of Product Originality

Dirk Snelders, Delft University of Technology
Paul Hekkert, Delft University of Technology
ABSTRACT - This study was designed to find underlying measures for product originality. It is argued that the common predictors of originality, i.e. novelty and appropriateness, can be caught in terms of relatively unique and appropriate associations of a product with a list of domains. In an experiment these associations were assessed for 14 telephones. When novelty of the telephones was statistically controlled for, two of the three association measures developed were significantly related to originality. These measures probe the relative uniqueness of the associations, regardless of the appropriateness of those associations.
[ to cite ]:
Dirk Snelders and Paul Hekkert (1999) ,"Association Measures As Predictors of Product Originality", in NA - Advances in Consumer Research Volume 26, eds. Eric J. Arnould and Linda M. Scott, Provo, UT : Association for Consumer Research, Pages: 588-592.

Advances in Consumer Research Volume 26, 1999      Pages 588-592


Dirk Snelders, Delft University of Technology

Paul Hekkert, Delft University of Technology

[The authors wish to thank Mia Stokmans for her help in formalizing the measures.]


This study was designed to find underlying measures for product originality. It is argued that the common predictors of originality, i.e. novelty and appropriateness, can be caught in terms of relatively unique and appropriate associations of a product with a list of domains. In an experiment these associations were assessed for 14 telephones. When novelty of the telephones was statistically controlled for, two of the three association measures developed were significantly related to originality. These measures probe the relative uniqueness of the associations, regardless of the appropriateness of those associations.


There is a widespread belief among both businesses and consumers that a new and original product appearance sells. This belief is not so much based on commercial proof (for a review of the value of imitation, see Kremen Bolton 1993), but more on a generally accepted feeling that things that are new and original are more appreciated than those which are old and boring. Two commercial advantages of a new product appearance ae usually provided in the literature. Although these listed commercial advantages may well serve as rationalizations of a more general liking of new designs, we list them here briefly. First, there is the argument that technical product functions are becoming more alike (Pilditch 1976) and component miniaturization is making appearance less dependent on function (Creusen 1998). Because of this, a marked appearance is a more important and feasible way of differentiating the product from that of competitors. Second, there is the argument that consumers (especially highly involved consumers) get bored with prototypical products, and products that are new and original draw the attention of consumers (Veryzer and Hutchinson 1998). Two ideas circulate in the consumer research literature as to why a new and original design can lead to positive affect. The first is that the initial incongruity of the new product with previously held ideas about the product can lead to a higher level of arousal, and a resolution of this incongruity will therefore be experienced more intensely (Mandler 1982; Meyers-Levy and Tibout 1989). A second explanation is that consumers may interpret the incongruity of the new product as exemplary of incongruities in their own lives, and this can lead them to sublimate their own feelings into the product (Durgee 1988).

The focus of this study is on product originality. [The research reported here is part of a series of studies in which aesthetic pleasingness is the main dependent variable. In the other studies (in preparation) subjective originality appears to be an important predictor of product liking, next to typicality.] More specifically, the aim of this study is to identify underlying measures that contribute to the perceived originality of a product. In the introduction above, and in the literature (especially in the field of creativity research), originality is treated as being similar to novelty, uniqueness or a-typicality (e.g., Guilford, 1968; Runco & Charles, 1993), and is normally assessed by self report scales ranging from not original to original (e.g. Hekkert & van Wieringen, 1996). In our view, however, there is a subtle, yet important distinction between originality and novelty or uniqueness. [Such a combination of novelty and 'fit' or usefulness has often been proposed as a necessary basis of creativity (e.g. Bruner, 1972; Runco & Charles, 1993). Although in those studies, originality is defined solely in terms of novelty, we strongly believe that for a product (or idea) to be viewed as original, it must also to a certain extent be appropriate or fitting. This believe may be biased by the Dutch use of the term original. In our language, the unusual product 'baseball' would not be considered as an original response to the question "Name all of the things you can think of which are square" as Runco and Charles (1993, p.537) do in their paper.] Artifacts must be considered original if they are novel or a-typical, but this novelty should at the same time 'make sense’, i.e., be appropriate. This means that original artifacts communicate something that is new and deviant in a way that is somehow appreciated. For example, a new car may look like a chicken, and this may be very unusual and new. However, this may not be enough for a car to be original since consumers may not like cars to look like chicken. Thus, originality is considered here as appropriate novelty.

Note that the criterion of appropriateness also says something about the kind of novelty that is promoted. Namely, if a deviation in a new product is meaningful, it must refer to something that reminds the consumer of something else. This means that these theories do not treat novelty as something that is absolutely new and that has never been seen before. [Note that we do not want to imply that absolute novelty does not exist. For example, a product made of a completely new material may have an absolutely new look and this may well form an attraction by itself. Examples for such absolute newness, however, tend to be rare, and most new products must be regarded as incremental innovations or modifications of old products (Bessant 1992).] Instead, the criterion of appropriateness implies that novelty is a relative quality that reminds the consumer of things that other products in the same category do not remind them of. Novelty and appropriateness can therefore be treated as associative qualities. The present paper will discuss how these qualities can be measured and to what extent these new association measures predict self-reported originality.


Before we arrived at the association measures used in the main study, we first tried out various operationalizations. These operationalizations are briefly reported here. The first aim was to arrive at a measure of novelty on the basis of reported associations. This means that we are looking at a way to assess to what extent objects are associated with objects from other domains in a relatively unique way.

Participants used during the development of the measure were departmental staff and students, all in the age group of 20-35 years. Each tested group consisted of equal portions of men and women. The stimulus set consisted of 14 slides of telephones, including old modes, new models, and prototypes that had never been on the market.

A first attempt to measure associative novelty (N=4) was to ask subjects to list all possible associations that they had with each of the 14 telephones. Next, they were asked to compare each association with each telephone and tick those cases where the association could be applied to the telephone. This procedure did not work because subjects found it difficult to judge whether a specific association given for one telephone could also be applied to another telephone in the set.

In a second attempt to measure associative novelty it was determined in advance what type of things could be associated with products like telephones and these associations were given to subjects so that they could tick those that applied to each of the telephones in the set. A list of 136 objects (products, toys, musical instruments, places, furniture, animals, foods, people, plants and emotions) was checked by new subjects on redundancy and completeness for things that one could possibly associate with telephones (N=10). Seventy-five of the most cited associations were kept for a final list. New subjects (N=4) were then asked for each of the telephones to tick those objects on the list of 75 that the telephones reminded them of. One problem arose immediately with this procedure: all four subjects noted that the list of objects to associate the telephones with was much too short, while the task itself was very fatiguing and should certainly not be longer (it now took about 45 minutes per subject).

A solution to this problem was to replace the list of 75 objects with a list of a comparable number of domains (places, times, activities, fictional settings) that regrouped the objects on the first list. So, instead of objects like 'cactuses,’ 'banana’s,’ 'lego’ and 'children’, we would now present whole domains on the list, like 'kindergarten,’ 'hospital’ and 'the fifties’. A first list of 89 domains was checked by new subjects on completeness and redundancy (N=4). The question here was: "Could a telephone have a form that reminds you of things that you especially encounter ...?" This resulted in a selection of 75 domains that were subsequently used in the main study. The domains are listed in the appendix.

Association measures

We depart from a matrix of 14 telephones and 75 domains. Let i be a specific domain and j a specific telephone, then aij is the association between the ith domain and the jth telephone (with value one if the association is made and value zero if it is not made). In this matrix, a first thing to look at is the absolute number of associations (NA) that each telephone has. NA can be expressed as:


To arrive at our association measures of originality, a first thing to look at is the relative uniqueness of the associations (UA) that can be expressed by:


UA describes the degree to which a telephone has relatively unique associations with domains. This means that a particular telephone has a high UA score if its associations with the list of domains are relatively unique, which means that not many other telephones have associations with the same domains. A critique that can be given of UA is that it depends too much on NA (i.e. the total number of associations that a telephone has with domains). Because of this, a telephone with a large amount of non-unique associations (i.e. associations that are given for a lot of other telephones as well) can receive a UA score that is higher than a telephone with a few unique associations.

A first measure of originality, based on associations, can be obtained by calculating the mean uniqueness of associations (MUA), which is independent of the amount of associations that a telephone has:


It is easy to see that MUA is equal to U / NA. MUA describes the relative uniqueness of the average association that a telephone has. Note that MUA only takes into account to what extent an association is uniqueBnot the extent to which the association is appropriate. Here, the appropriateness of each association is established by asking participants to what extent each domain is liked on a seven point scale. The appropriateness of each domain is expressed as wi.

A second measure of originality is therefore based on the mean weighted uniqueness of the association (MWUA):


MWUA describes the relative uniqueness of the average association that a product has, weighted by the appropriateness of that association. This means that a telephone is considered original if, on average, it has associations with other telephones that are unique (other telephones don’t have them), and if those associations are appreciated. All of the measures developed in this section (NA, UA, MUA, MWUA) will be used in the main study.


The goal of the main study is to correlate our new originality measures with a regular self-report scale of originality. Two more self-report scales were administered to control for the effects of absolute novelty and typicality. This was done because NA (and thus MUA and MWUA) was thought to be dependent on absolute novelty (new products will have fewer associations) and typicality (typical products will have more associations). Since self-reported originality was thought to be positively correlated with novelty and negatively correlated with typicality, we have included these two variables as control variables.


The same 14 telephones were used as in the previous section. Twelve participants took part in the study, all in the age group 20-35 years, and six were men, six were women. After a brief explanation of the procedure, participants were asked to perform two tasks. Task 1 was that participants had to check each telephone on the list of domains and tick those instances where that telephone could be associated with a domain. The exact question was: "Could a telephone have a form that reminds you of things that you especially encounter ... [in the specific domain]?" After the last telephone was checked, participants were asked to rate how much they liked each domain on a seven point scale (ranging from "not like at all" to "like very much"). NA, UA, MUA, and MWUA were all calculated on the basis of Task 1. During Task 2 participants were asked to rate each telephone on seven point scales of originality (ranging from "not original" to "original"), novelty (from "old model telephone" to "new model telephone"), and typicality [from "bad example for the category" to "good example for the category," administered after instructions similar to those of Rosch and Mervis (1975, p. 588)]. Task 1 and 2 were changed systematically in order.






Table 1 presents the mean scores and standard deviations for the 14 telephones. The correlations between all scale ratings and the various originality measures are presented in Table 2. As can been seen from this table, the Number of Associations (NA) is strongly related to Typicality (r=.73) and has a negative correlation with both Originality (r=-.78) and Novelty (r=-.72). These combined results clearly indicate that typical, older models of a telephone could be associated with more domains than original and new models. This finding supports our theoretical notion that an appropriate originality measure based on associations should becorrected for by NA. This decision is furthermore backed up by the strong correlation (r=.78) between NA and Uniqueness of Associations (UA).

The first proposed association measure of originality, i.e., Mean Uniqueness of Associations (MUA), was not related (r=.38, ns) to Originality. By adding a domain-preference weight to this measure (MWUA), the correlation slightly improved and approached significance (r=.50, p=.07). This result at least suggests that a measure based on the mean uniqueness of associations with domains and weighted by the preference for the associated domains, explains some of the variance in the originality ratings.

A tentative explanation for these fairly low correlations between Originality and our association measures can be drawn from the high correlation between Originality and Novelty (r=.91). Since a variety of old, new and prototype models were included in the stimulus set, it can be predicted that the expected relations between Originality and the association measures were suppressed by the large variation in absolute novelty of the models. To test this prediction, partial correlation coefficients between the two association measures and Originality were computed. After Novelty was partialed out, both MUA (r=.81, p<.01) and MWUA (r=.63, p<.05) correlated significantly with Originality. The high partial correlation between MUA and Originality indicates that this association measure explains a unique amount of variance in the Originality ratings. This amount of variance does not increase as a result of correcting the association measure by a weight based on domain preference. The high correlation between Novelty and Originality becomes even higher when either MUA (r=.96, p<.01) or MWUA (r=.93, p<.01) is partialed out.


Our results confirm the prediction that the originality of a product is related to the mean uniqueness of an association with a domain. Originality ratings of telephones are determined by their absolute novelty as well as by the novelty of the associations they can have. Both factors explain a significant amount of unique variance in the originality ratings.

The evidence obtained does however not indicate that the domain with which the product is associated has to be appropriate. A measure of this association uniqueness corrected for by an appropriateness weight based on domain preference did not result in a better fit. On the basis of this result one might be tempted to conclude that the nature of the associated domain, in terms of liking, is not important with respect to the originality of the product. Such a conclusion is however premature. In the remainder of this discussion we will present arguments for an alternative interpretation of the appropriateness of the associated domains.

One of the telephones in the stimulus set is red and has the shape of a heart. Although this design solution is not rated as novel, this model received the highest MUA rating of the set. Apparently, it made a number of associations possible the other models did not have. Examples of such unique associations are a brothel and the bedroom of a teenager. In this study, the appropriateness of these domains was assessed by liking ratings for the domains. However, consumers may dislike brothels but still consider the association appropriate. The tentative reason for this is that telephones share a (highly abstract) feature with brothels and teenager bedrooms, which is expressed in the design of this model. This feature could be characterized as 'being connected to each other, a means to communicate or experience love (or sex), a place where sharing of 'love’ is held important’. Seen in this way, the association is understood in terms of a metaphorical relationship. Based on theory of metaphorical comparisons, it is argued that this feature sharing contributes to the appropriateness of an association.

Following the terminology of Richards (1936), a metaphor is the characterization of one object, called tenor, by another object, called vehicle. The basis for comparison between both objects is called the ground. A comparison is only classified as a metaphor if the tenor and vehicle are members of different domains (e.g., Lakoff, 1993). According to Lakoff (1993), "The locus of metaphor is ... in the way we conceptualize one mental domain in terms of another" (p. 203). In this sense, the phrase "an eagle is like a hawk" is a characterization that is not a metaphor, while the phrase "an eagle is like a lion" is a metaphor.

There is some dispute among experimental psychologists over the ground on which a tenor can be characterized by a vehicle from other domains (for a review, see Tourangeau & Rips, 1991). The ground of comparison has been presented as a set of features shared by tenor and vehicle that can range from the very concrete (Ortony 1979) to very abstract (Tourangeau and Sternberg 1982). Consider the phrase "men are wolves" as a metaphor based on abstract features. The ground here is not a set of concrete features of wolves like biting and growling, but more abstract ones like aggressiveness and competitiveness. These features can be very salient for both the tenor and the vehicle. Often however, the metaphor introduces new, 'emergent’ features that are not ordinarily seen as characterizing either the tenor or vehicle (Tourangeau & Rips, 1991). In this way, the metaphor reorganizes the understanding that one had of the tenor, i.e. the basis of comparison is created by the metaphor itself. An example is a business suit, designed by Vivienne Westwood, that looks like an armor used by knights in the middle ages. Such a suit introduces a sense of feudal hierarchy and connects this to contemporary business practice. In this case, the association between tenor (business suit) and vehicle (feudal knight) is created by the metaphor, which is the Westwood suit, and it is an original association, in the sense that it is unique (no other suit made this association), while at the same time appropriate for those with an appreciation of the feudal qualities of the business suit.

From these theoretical arguments on the interpretation of a metaphor, it can be argued that it is not the preference for the domain as such that makes an association appropriate, but instead, whether the domain (the vehicle) shares (emergent) features with the product (the target). An association with a particular domain may be considered appropriate if that domain 'makes sense’, i.e., if exemplars of the domain have features that can in some logical way be transposed to the target. If these features are emergent, i.e., if they were not seen as characteristics of the target before, but emerge from the comparison (the design), the originality of the association is furthermore enhanced. The example of the telephone shaped as a red heart clearly reveals this notion of appropriateness. Future experiments with new association measures based on this interpretation of appropriateness must confirm these predictions.




Bessant, J. (1992). "Big bang or continuous evolution: Why incremental innovation is gaining attention in successful organisations," Creativity and Innovation Management, 1 (2), 59-62.

Bruner, J.S. (1972), "The conditions of creativity," In On knowing: Essays for the left hand, ed. J.S. Bruner, Cambridge, MA: Harvard University Press.

Creusen, Marielle E.H. (1998), "Product appearance and consumer choice," Dissertation, Delft: Delft University of Technology, Fac. OCP.

Durgee, Jeffrey F. (1988), "Product drama," Journal of Advertisin Research, (Feb-March), 42-49.

Guilford, J.P. (1968), "Intelligence, creativity and their educational implications," San Diego, CA: Knapp/EDITS.

Hekkert, Paul and Piet C.W. van Wieringen (1996), "Beauty in the eye of expert and nonexpert beholders: A study in the appraisal of art," American Journal of Psychology, 109 (3), 389-407.

Kremen Bolton, Michele (1993), "Immitation versus innovation: Lessons to be learned from the Japanese," Organizational Dynamics, (December), 30-45.

Lakoff, George (1993), "The contemporary theory of metaphor," In Metaphor and thought (2nd edition), ed. Andrew Ortony, Cambridge: Cambridge University Press.

Mandler, George (1982), "The structure of value: Accounting for taste," In Affect and cognition, ed. Margaret S. Clark, Hillsdale, NJ: Erlbaum.

Meyers-Levy, Joan and Alice M. Tybout (1989), "Schema congruity as a basis for product evaluation," Journal of Consumer Research, 16 (June), 39-54.

Ortony, Andrew (1979), "Beyond literal similarity," Psychological Review, 86, 161-180.

Pilditch, James (1976), "Talk about design", London: Barry and Jenkins.

Richards, I.A. (1935), "The philosophy of rethoric," Oxford: Oxford University Press.

Rosch, Eleanor and C.B. Mervis (1975), "Family resemblances: Studies in the internal structures of categories," Cognitive Psychology, 7, 573-605.

Runco, M.A. and R. Charles (1993), "Judgments of originality and appropriateness as predictors of creativity," Personality and Individual Differences, 15, 537-546.

Tourangeau, Roger and Lance Rips (1991), "Interpreting and evaluating metaphors," Journal of Memory and Language, 30, 452-472.

Tourangeau, Roger and Robert Sternberg (1982), "Understanding and appreciating metaphors," Cognition, 11, 203-244.

Veryzer, Robert W. and J. Wesley Hutchinson (in press), "The influence of unity and prototypicality on aesthetic responses to new product designs", Journal of Consumer Research.