A Cognitive Model of Innovative Behavior: the Interaction of Product and Self-Attitudes

Citation:

Marvin E. Goldberg (1971) ,"A Cognitive Model of Innovative Behavior: the Interaction of Product and Self-Attitudes", in SV - Proceedings of the Second Annual Conference of the Association for Consumer Research, eds. David M. Gardner, College Park, MD : Association for Consumer Research, Pages: 313-330.

Proceedings of the Second Annual Conference of the Association for Consumer Research, 1971     Pages 313-330

A COGNITIVE MODEL OF INNOVATIVE BEHAVIOR: THE INTERACTION OF PRODUCT AND SELF-ATTITUDES

Marvin E. Goldberg, McGill University

[This research study was a part of a doctoral dissertation completed at the University of Illinois, 1971. The author is grateful to Professor Joel B. Cohen, the University of Illinois, for his considerable help in all aspects of this study. The assistance of the Colgate Palmolive Company, Canada Ltd., is also acknowledged.]

[Marvin E. Goldberg is Assistant Professor Marketing, Faculty of Management, McGill University.]

Past research in the area of new product innovation (for example: Pessemier, Burger & Tigert, 1967; Robertson, 1967; Robertson & Meyers, 1969) has shown marketers that individuals who are younger, better educated, wealthier and more cosmopolitan tend to buy more new products earlier. Unfortunately, these rather general guidelines are often of limited value. The manufacturer who, for example, introduces a new skin care cream for teenagers, or a new spray hair coloring for men, is concerned largely with specific segments of the market, and cannot base his decisions upon such general demographic guidelines as "more cosmopolitan".

One direction research can take is to focus each time anew upon the specific segment of the market perceived to be relevant to the product of immediate concern. Any company (with a large enough research budget) is capable of taking this approach. What is learned probably enables a company to market each such new product more effectively than had it proceeded by mere intuition.

The cost paid for this precision, however, extends beyond the financial cost of each separate study. Another "cost" is the limited ability to generalize findings either from one product or product category to the next, or from one market segment to the next. The resulting inefficiency and repetitive outlay of research dollars seem to warrant efforts at establishing an alternative approach which would allow for generalization.

One such alternative would be to seek a general understanding of consumer purchase behavior vis-a-vis new products. Beyond the parsimony and efficiency of such an approach, it seems appropriate for marketing theoreticians to adopt such a strategy in order to develop a cumulative, theoretically integrated body of knowledge.

In order to do so, marketers must work with a concept (or concepts) common to all consumers. One such concept which may prove useful is personality. It has recently been suggested (Marlow & Gergen, 1969) that general statements concerning social behavior perhaps ought not be developed without personality constructs serving as important and essential qualifiers:

. . . general laws [may be]. . . entirely misleading in that they reflect only the ways in which groups of persons sharing similar personality characteristics react to a given situation. In this sense, personality constructs might be used to replace all other intervening variables . . . personality comes to play an important part in understanding almost all aspects of social processes.

With respect to any one specific object or situation, there are several other important variables besides personality that help explain an individual's behavior. These include: externally, strong instigation from the situation itself; and internally, states of deprivation, and momentary sets. These other factors notwithstanding, an individual's basic orientations towards his environment are so pervasive it would be unusual if they failed to exert an influence upon the general pattern of his purchase decisions.

NEED FOR A THEORETICAL FRAMEWORK

Recent attempts in marketing to use personality as a common unit of analysis (Evans, 1959; Kaponen, 1960; Massy, Frank, Lodahl, 1968; Robertson & Meyers, 1969) have not been very successful. This is due at least in part, to the failure of researchers to apply a formal theoretical framework. [In two recent papers both Kassarjian (1970) and Jacoby (1970) have noted the "atheoretical-shotgun" approach taken in studies attempting to link personality variables and purchasing behavior.] The goal of this project is to develop and empirically validate a theoretical model wherein personality variables are linked to particular new product attributes.

Failure of past studies using personality to predict purchase behavior is partially attributable to the inappropriate nature of the tests used. Tests such as the Edwards Personal Preference Schedule (Edwards, 1957), frequently used by marketers, and the California Psychological Inventory (Gough, 1957) have a number of distinct limitations insofar as their application to marketing iq concerned.

First, use of these multiple factor tests has encouraged an unstructured a theoretical approach. Of the numerous personality traits that compose these test batteries, many are theoretically irrelevant with respect to the product or products involved. These traits should not be expected to correlate significantly with purchase behavior, and in fact they usually do not e Such significant findings as are revealed are likely to be rather disjointed pieces of information. For example, it is difficult to draw any meaningful conclusions when Robertson and Meyers (1969) note that:

Appliance innovators may be somewhat more self-accepting, with a greater sense of personal worth and they may also be more dependable, moderate and tactful. They seem to be somewhat less socially mature, modest and serious.

Precisely why these traits are linked to each other and to appliance innovation, and by what underlying construct they are linked, are questions left unanswered. As Jacoby (1970) notes:

. . . if . . . investigators were to select specific personality traits for study on the basis of theoretically derived hypotheses, were to make specific predictions as to how these traits ought to interact with, or affect specific aspects of consumer behavior, and were to utilize experimental rather than correlational paradigms, the likelihood of obtaining significant and usable findings should be enhanced.

The test batteries commonly employed are composed of about twenty different scales, each assumed to be independent. If scores on each of the twenty scales are compared with one or more criterion variables such as purchase behavior, or attitude, one or two out of twenty comparisons ought to be significant on the basis of chance alone (depending upon the level of confidence adopted). If three or four are revealed, which ones are mere artifacts? The lack of theoretical guidelines makes it difficult both to interpret the meaning of the findings and to relate these findings to a larger body of knowledge.

A second limitation of personality tests such as the Edwards Personal Preference Schedule or the California Psychological Inventory, relates to the nature of some of the needs they measure. Needs can be conceptualized at various levels ranging from those deep within the individual, to the "quasi-needs" to which Lewin (1951) refers, that arise as a consequence of the individual's interaction with his environment.

Some of the factors in tests such as the California Psychological Inventory attempt to measure deeper level needs (such as "self-acceptance", "sense of well-being", "self-control"). The problem is that the individual can act out and fulfill these needs in a variety of ways or situations, often as a result of interaction with other less central needs and important environmental stimuli.

Not only may it be difficult, it may not be entirely relevant for marketers to assess deeper level needs in order to predict purchase behavior. The lack of success in working with deeper level needs suggests two directions for future research: first, use of tests that stress the relationship between the consumer and relevant people and objects in his environment; and secondly, the importance of considering products and product attributes as particularly critical stimuli within that environment.

Inappropriate application of personality tests has not been the only reason for the lack of success in relating personality to purchasing behavior. It is also likely that failure to properly define the relevant stimuli, that is, product and product attributes, has contributed to the problem. As Blake et al. (1970) recently hypothesized, the inconsistent results of past studies linking personality characteristics and the acceptance of new products ". . . is partially due to the fact that the relationship is mediated by the type of new product analyzed".

In most studies of new product purchasing, the dependent measure has been created in one of two ways: either l) the study limits itself to a single product, or 2) it summates across a number of products or even product categories. Both approaches assume that the purchase of new products is an undifferentiated phenomenon.

The Single Product Study

A typical example of the first approach is the Lafayette Consumer Behavior Research Project (Pessemier, Burger, & Tigert, 1967). In this study, hypotheses predicting the purchase (or non-purchase) of a new detergent by a group of housewives are acknowledged to have been derived from the diffusion of innovations literature, including studies dealing with farmers' purchases of new farm equipment, and doctors' prescriptions of new drugs. While both the nature of the product as well as the class of consumers shifts radically from study to study, these differences are largely overlooked. But can one deny that the consequences for a venturesome doctor who prescribes a new drug may differ considerably from the consequences for a venturesome housewife who buys a new detergent? The relative importance of venturesomeness, or of any personality variable, as a determinant of purchase behavior is likely to vary considerably as a function of such factors as the nature of the product and the constraints inherent in the consumer's role. [Kegerris et al. (1970) recently studied the introduction of a rather performance oriented new product, the automotive diagnostic center. They concluded ". . . innovators, far from being reckless or impulsive consumers, appear to be more careful planners than the population in general." By contrast, Robertson and Kennedy (1968), studying the touchtone (pushbutton) telephone found that venturesomeness, operationally defined as "willingness to take risks in the purchase of new products . . . [is one of two variables to] account for most of the innovative behavior difference between innovators and noninnovators . . ." It is highly likely that it is the nature of the product in each study is responsible for these contrary conclusions.]

The Summation Approach

In the "summative" approach, individuals are categorized as innovators or non-innovators on the basis of how many of a list of new products they have purchased. Such an approach fails to take into account specific subclasses of products. This is an important shortcoming, for a Robertson and Meyers (1969) note, the inter-relationship of innovativeness with respect to the three broad product categories they studied (electrical appliances, clothing, and food) ". . . are pragmatically low and seriously dispute [the notion that] innovativeness is a general trait possessed by the individual".

If there is no ready equivalence between all new products, and if the undifferentiated concept of innovativeness is incorrect, then some means of discriminating among products or groups of products is imperative. There are a practically infinite number of ways [One way is to use the manufacturer's standard classifications as do Robertson and Meyers (1969) in the study noted above.] to categorize products and product attributes, but probably the most relevant way to do so is according to the way consumers categorize them. Since individuals tend to perceive and categorize their environment in part on the basis of their needs and values (Bruner, 1958) salient product categories are most likely to be identified by defining both consumer needs and product attributes interdependently. As Massey et al. (1968) note:

. . .one of the most fruitful directions for future research is the study of characteristics that are idiosyncratic to both the consumer and the product, and not to the customer alone as in the case of general personality characteristics.

Consumer purchase behavior is not simply a function of consumer needs considered by themselves, nor merely of perceived product attributes, but rather of the interaction between these two factors. Ideally, the same theoretical model should identify the relevant (and interactive) dimensions of both consumer needs and perceived product attributes.

THEORETICAL MODEL

Anyone concerned with presenting a model interactively linking consumers and products would find several methods of categorizing consumer needs and product attributes discussed in the literature. Kassarjian (1965) has used Reisman's (1961) "inner" and "other" directed social character types to predict differential responses to advertisements that may be categorized in the same manner. Cohen (1967) has linked Horney's (1937) "compliant", "aggressive", and "detached" personality types to products whose functions are likely to make them more appealing to one or another of these personality types. Wilding and Bauer (1968) have suggested "problem-solving" and "psycho-socializing" as two different games people play in making purchase decisions, or in reacting to persuasive communications which may be similarly differentiated. Brody and Cunningham (1968) have differentiated between a product's "performance" and "social" risk and have related this dimension to the consumer's level of self-confidence and to the likelihood of brand loyal behavior. Bourne (1957) has posited a product's social conspicuousness as a determinant of the extent to which reference groups are taken into account in purchasing decisions. In light of these varied approaches, how might consumer needs with regard to new products and new product attributes best be categorized? The model outlined here begins by positing four product dimensions.

Product Dimensions

Two questions may be asked of any new product. First, how new is the product? How novel is it perceived to be? Secondly, how is the product new? What attributes does it have so that it is perceived as being new?

Product dimension I: novelty. The question "how new is the product" suggests that not all new products should be treated as equally new. Nor should newness be measured as a function of the length of time the product has been on the market. A product recently introduced may involve little that is unique or novel, while a product of a few years standing may still be perceived as relatively novel.

Of the various conceptualizations of novelty, one version seems especially relevant to the measurement of novel products: that of the perceived discrepancy between a novel and a familiar stimulus

A novel object is often surprising. i.e. it differs from what was expected . . . (Berlyne, 1960)

There is a discrepancy between information embodied in expectations and information embodied in what is perceived. (Berlyne, 1965)

Novelty is conceptualized here as the perceived discrepancy between new and familiar versions of a product. [Blake et al. (1970) differentiate between a product new in terms of its "recency" or its "novelty". They define the former as those products that are "incongruous with expectations of the typical product in that class". They argue that the two dimensions are independent: "A product's recency should not affect its associated uncertainty". While "recency" and "novelty" may not be perfectly correlated, it seems reasonable to assume that however "novel" a product is, the less "recent" it is, the less it will be perceived as incongruous or discrepant. With the passage of time the novel product is likely to be perceived as less and less novel. At least for the purpose of the model outlined herein, it would seem advantageous to treat novelty as a unidimensional concept.]

With regard to the second question, "how is the product new?", one may ask: "Regardless of how new the product actually is, in what way is it changed so that it is perceived as being new? In the model developed here, two attributes - the product's appearance and performance features - are considered critical. Products are perceived and labelled as new, either because they have novel appearance features, novel performance features, or both. [The dichotomy between appearance and performance has appeared previously in various related forms, most recently (Jacoby, 1970) as "fashion" versus "function". The above definition of "appearance" is intended to broaden the scope of "non-functional" product changes to include any appearance-oriented change, regardless of whether or not it is part of a general fashion. Moreover, the term "fashion" has often been unduly restricted to articles of clothing.]

Product dimension II: appearance. Products new in terms of appearance, appeal to consumers because they are perceived as more attractive, becoming or aesthetically pleasing than their older, more familiar counterparts.

Product dimension III: performance. Products new in terms of performance, appeal to consumers because they embody the latest product advances or improvements. They enable the consumer to perform some task more easily, efficiently or thoroughly. [Conceptually, the extent to which a product is novel (our first question) is a function of the extent to which both its appearance and its performance features are novel. Operationally, the product's total (gestalt) novelty value is measured independently so that the relative contribution of novel performance and appearance attributes may be functionally linked to the product's total novelty value.]

Product dimension IV: social conspicuousness. The appearance or performance features of a new product may be more or less socially conspicuous; that is , visible to and discussed by ones' friends and acquaintances. Products that are socially conspicuous help the consumer relate to others -by enabling him to create a desired impression upon them. In other words, by allowing the consumer to demonstrate something unique or novel about himself, the socially conspicuous product changes the nature of the product's novelty into an "expressed novelty".

Consumer Dimensions

In Lewinian Field Theory, objects in our psychological environment have either positive or negative value for us, depending upon our needs. The value of an object, in coordination with our needs, creates a force acting upon us. Objects that possess positive value attract us; those with negative value, repel us.

A product's appearance, performance, and social conspicuousness all represent potentially positive forces. Just as food appeals more to a hungry person, products new in terms of their appearance should appeal more to consumers with a relatively high "aesthetic" orientation. [This and the other consumer orientations are defined presently.] Products with improved performance capabilities should appeal more to consumers with a relatively high "practical" orientation. New products that are socially conspicuous should attract consumers with a relatively high interaction orientation. Highly novel products ought to be more appealing to the more venturesome consumer. [While novelty has been conceptualized by Berlyne (1960, 1965) and others as having both positive and negative valence, the paradigm outlined here simply specifies that novelty will be less positive for those who are less venturesome and more positive for those who are more venturesome.]

In our interactive model, we are guided to the relevant consumer dimensions by the four product attributes:

Product Attribute                      Consumer Dimensions

Total Novelty . . . . . . . . . . . . . . . Venturesomeness

Appearance. . . . . . . . . . . . . . . . . Estheticism

Performance . . . . . . . . . . . . . . . . Practicality

Social Conspicuousness. . . . . . . . Interaction orientation

Consumer dimension I: venturesomeness. This attribute refers to a general predisposition to evaluate new products favorably. Berlyne (1960), referring to a curiosity need, or a need for novelty, indicates that "the actual level of novelty to which an individual is geared will vary within relatively wide limits depending . . . on differences in personality".

In the diffusion of innovations literature, much the same concept has been termed "venturesomeness". Rogers (1962) defines venturesomeness as "the major value of the innovator. He must desire the hazardous, the rash, the daring and the risky".

Similarly, in the marketing literature, Robertson and Kennedy (1968) have defined venturesomeness as willingness to take risks in the purchase of new products. The definition of venturesomeness used in this study is intended to be more general, in that it includes a predisposition to evaluate a new product favorably and is not limited to overt purchase behavior.

Consumer dimension II: esthetic orientation. The aesthetic individual is interested in products that are attractive and decorative. He/she is more attracted by those products that have a new shape, color, scent or sound than ;.c the individual who is less aesthetic.

Consumer dimension III: practical orientation. The practical individual is interested in products that help him/her in their day-to-day tasks. He/she is relatively more attracted by new products that help complete a job quickly, efficiently and thoroughly than is the less practical individual. [The salience of this concept has been noted by Harrison Gough (1970), author of several personality batteries including The California Psychological Inventory (Gough, 1957) and the Adjective Check List (Gough, 1952):

Practicality [as outlined above] is something that ought to be measured and measurable, although I know of no scales directly targeted on the concept . . . Perhaps you could construct a new scale for practicality . . . I'd urge you to think of this possibility.]

Although some consumers may be either high or low on both dimensions, the constraints faced in daily purchase behavior often necessitate a choice between products of a more aesthetic nature and those of a more practical nature. Decisions in this regard are likely to be, at least in part, reflections of basic needs or predispositions. It is on this basis that the dichotomy is made between the practical and the aesthetically-oriented individual. [To take advantage of this market segmentation strategy, marketers might establish: 1) the distribution of aesthetically and practically-oriented consumers in the target market; 2) the extent to which the addition of appearance or performance attributes would increase the product's appeal (or intention to purchase) for each of these market segments; and 3) weigh the potential increase in market share against the increase in unit costs.]

Consumer dimension IV: interaction orientation. The interaction-orientated individual is interested in being with people. He/she seeks social activities and gains satisfaction from them. As Cohen points out (1968):

There is very little behavioral or attitudinal response made by a consumer that is not a response to significant "others" present either physically or referentially, at the time . . . The things which we own, just as the opinions we hold fit the image we have of ourselves in relation to others.

To the extent an individual is interaction-oriented these "significant others" will be taken into account in purchasing behavior.

HYPOTHESES GENERATED BY THE MODEL

In general, the model hypothesizes that where there is a "matching" or compatibility between product attributes as perceived by the consumer and consumer needs, the product will be evaluated more favorably than where there is a "mismatching" or incompatibility.

Of the four product dimensions, appearance and performance novelty are focused upon generating the following more specific hypotheses:

- A product, new in terms of its appearance attributes will be evaluated more favorably by consumers who are aesthetically-oriented than by consumers who are practically-oriented.

- A product, new in terms of its performance attributes will be evaluated more favorably by consumers who are practically-oriented than by consumers who are aesthetically-oriented.

Considering all four product dimensions, the following broader hypotheses are generated:

- For any given consumer, products that match his/her needs more closely (on all four dimensions) will tend to be evaluated more favorably than products that match his/her needs less accurately.

- For any given consumer, groups of similar products (representing a potential "lifestyle"), will tend to be evaluated more favorably where this lifestyle more accurately reflects the consumer's needs than where it reflects his/her needs less accurately.

METHOD

Subjects

The sample consists of 192 English-speaking Montreal women drawn from nine voluntary associations. The mean age of the sample is thirty-eight. The mean household income is almost $14,000. About half the sample had at least some college education. In general, the group represents the middle and upper-middle socio-economic strata.

Instruments

Venturesomeness is measured by a sixteen item Likert instrument, developed specially for this study in a series of pilot tests. One form of the instrument has been demonstrated to correlate .53 with the Change scale of the Adjective Check List (Gough, 1952). The venturesomeness scale achieved a reliability of .83 (Kuder-Richardson Coefficient Alpha (see Nunnally, 1967).

Estheticism versus Practicality is measured by a twenty-three item Likert instrument. The items were selected on the basis of face-validity and their demonstrated predictive validity in pilot studies. [For further information regarding the development of this and the Venturesomeness scale see (Goldberg, 1971).] The Kuder-Richardson coefficient alpha for the scale is .62.

In his study of the role played by compliant, aggressive and detached personality types in purchasing behavior, Cohen (1968) suggests that "the aggressive person should desire more distinctive brands . . . . Acceptance by others is not enough . . . . He wants to establish his separate identity and style of behavior". This conception of an interaction orientation may be linked directly with the concept of the socially conspicuous product. Consequently, this study uses the fifteen item "aggressiveness" scale of Cohen's C.A.D. instrument (1967). [For criterion III (below), the Detached scale was used instead; see (Goldberg, 1971).]

Procedure

The model outlined above was tested using three distinct and independent criterion.

Criterion I: verbal product alternatives. Eighteen pairs of alternative versions of new products were presented to the S's. In each pair, one version was "performance-oriented" and one version was "appearance-oriented" (for example, a portable T.V.; version "a": ten pounds lighter; version "b": with a walnut cabinet). S's were asked which version they preferred and whether they preferred it strongly or only somewhat.

These are two advantages to this criterion:

1) it attempts to eliminate all variance other than the variance relevant to the appearance - performance dichotomy;

2) it groups individual products along meaningful dimensions, creating a more reliable criterion.

Criterion II: selection of alternative gifts. S's were asked to select two products as gifts: either Ultra Brite Toothpaste or Colgate Fluoride Toothpaste; and either Florient pine-scented air freshener or a new disinfectant air freshener. In both cases, theoretically, the S was choosing between a "practical" alternative (Colgate; disinfectant) and a more aesthetically or sense-oriented alternative (Ultra-Brite; Florient).

The advantage of this second criterion is evident. It permits prediction of actual behavior vis-a-vis the relevant dimensions.

Criterion III: evaluation of product clusters (lifestyles). S's were shown slides (with verbal descriptions) of twenty new products. These ranged from a glass-top stove to frozen pudding. Products were selected in pre-tests to provide variation with respect to each of the four product dimensions outlined in the model. S's were first asked to describe each product in terms of its newness along the four product dimensions and then to evaluate the product in terms of the extent to which the product was perceived as an improvement over its older counterpart.

This third criterion affords maximum flexibility by including products that vary along each of the four product dimensions. Since these dimensions are defined on the basis of consumer needs and perceptions, the chance of interaction between product and personality measures is maximized and the predictive Dower of the model should be enhanced.

Moreover, as with criterion I, it is possible to group the products along a particular dimension into meaningful clusters. Each cluster of similar products may be regarded as presenting a potential "life-style". This approach removes the restrictive assumption that personality need predict behavior vis- a-vis a single stimulus. As is the case with attitudes (Fishbein,1966), personality is more likely to be predictive of a cluster of related behaviors (products), rather than any single specific act (product).

RESULTS

Criterion I: Verbal Product Alternatives

The criterion score in a multiple regression analysis consists of a simple index: the number of performance alternatives selected minus the number of appearance alternatives selected (weighting each alternative checked "prefer strongly" by two).

The general paradigm is as follows:

Y₁ = a + b₂X₂ + b₃X₃ + b₄X₄ + b₅X₅

where Y₁ = the criterion index; X₂ = Venturesomeness; X₃ = Estheticism-Practicality; X₄ = Interaction orientation; X₅ = Age.

The adjusted multiple correlation coefficient (R_1.2345) is .44(F = 11.32, p<.05, 4, 187 d.f.). As is revealed in Table 1, the four predictor variables, three of them personality measures, explain one-fifth of the variance (R2 = .195).

TABLE 1

CRITERION I: VERBAL PRODUCT ALTERNATIVES

Criterion II: Selection of Alternative Gifts

The attempts to discriminate among selectors of two toothpastes and two air fresheners were not very successful.

Two variables, Venturesomeness and Estheticism-Practicality were statistically significant discriminants of Ultra Brite and Colgate Selectors (Mahalanobis D2 = 20.16; p<.001) (see Table 2). However, only two percent (.016) of the variability in discriminant space is relevant to group differentiation (Tatsuoka, 1970). Additionally, (although this criterion is not without ambiguity; see Morrison, 1969) only 63% of the cases were correctly classified.

Results of the discriminant analysis in the case of the two air fresheners (Florient and the disinfectant) were equally poor. Here too, only two percent (.020) of the total variability of the total variability of the discriminant function is attributable to group differences. Fifty-seven percent of the cases were correctly classified. Mean scores of the predictor variables for each group were not statistically significant.

Criterion III: Evaluation of Product Clusters (Lifestyles)

Each product is clustered into low, medium and high levels of each of the four product dimensions, (i.e. 12 clusters in all). This is accomplished using the mean score provided by the sample in describing each dimension of each product.

The model predicting each S's evaluation of the product cluster is as follows:

Y₁ = a + b₂X₂ + b₃X₃ + b₄X₄

where Y = the score obtained for each S by standardizing (across all S's) the ratings1 for each product; and for each S, summating her standardized scores for all products in the given cluster; X₂ = Estheticism-Practicality; X₃ = Interaction Orientation; X₄ = Venturesomeness.

TABLE 2-A

CRITERION II: MEAN VALUES OF CHARACTERISTICS: A) TOOTHPASTES

TABLE 2-B

CRITERION II: MEAN VALUES OF CHARACTERISTICS: B) AIR FRESHENERS

Table 3 indicates the adjusted multiple correlation coefficient for each cluster of products. As is revealed in the table, the proportion of variance (R2 2 4) accounted for ranges from three percent to fifteen percent. Better prediction is obtained with the product clusters that are clearly appearance-oriented (high appearance cluster: R2 , .13; low performance cluster: R2 = .15). [Although a few products are both "appearance-oriented" and "performance-oriented", in general, those that are highly appearance-oriented are low in terms of performance orientation and vice-versa. Consequently, the clusters: "low-performance" and "high-appearance" consist of almost identical sets of products.] R's in nine of the twelve clusters are significant at or beyond the .01 level.

TABLE 3

CRITERION III: EVALUATION OF PRODUCT CLUSTERS (LIFESTYLES)

On the average, the three predictor variables account for 8.4 percent of the variance. By contrast, when predicting with respect to each of the twenty products individually, the same three variables account for an average of only three percent of the variance.

DISCUSSION

It has been noted (Kassarjian, 1970) that most personality studies in consumer behavior have accounted for no more than five to ten percent of the variance. Several of the reasons how prediction might have been improved were noted above. It is important to recognize, however, that the multiple influences upon a purchasing decision constrain any single predictor from explaining much more of the variance. In addition to the S's personality traits, factors such as the momentary needs and enduring personality traits of others in the family, past experience with various brands in the product category, and various marketing variables such as price and advertising are all likely to be at least equally powerful predictors. Ten percent of the variance is likely to prove an accurate estimate of the role the consumer's personality traits play in the overall decision process.

For the researcher attempting to achieve a general understanding of consumer behavior as one of the many facets of human behavior, personality is potentially an important theoretical construct. It should facilitate our comprehension of the ties between consumer decision processes and decision processes in other spheres of life. On the other hand, it is probably illusory for the marketer searching for a single predictor with regard to individual products to expect general personality measures alone to hold the key. In part for this reason, it has often been suggested (Brody & Cunningham, 1968; Massy, Frank, & Lodahl, 1968; Kassarjian, 1970) that consumption-specific instruments ought to be developed. Perhaps the answer for the marketer does lie in this direction. It is important to take note, however, of the potential disadvantages inherent in such an approach.

First, there is a danger that this strategy, if followed to the exclusion of others, might foster an unwanted and unnecessary schism between consumer behavior and the general body of personality research.

More important, however, is the questionable reliability and validity of many consumption-specific instruments, especially those of a psychographic nature. (Low reliability of the instruments is often almost assured by the very small number of items used to measure each trait.) Moreover, the number of "factors" one might invent in the absence of a systematic program of convergent and discriminant validation (Campbell & Fiske, 1959) seems almost endless. To trade what reliability and validity there is in general personality batteries for the usually lower standards of consumption-specific tests would hardly seem to represent a step forward. [This is not to ignore the many serious questions involving the validity of general personality tests. For example, in one study (Mitchell, 1963) where the Sixteen Personality Factor Test (Cattell & Eber, 1957) and the California Psychological Inventory (Gough, 1957) were administered to the same set of subjects, the factors labelled "Dominance" on each test were correlated only .32; see also (Goldberg, 1971).] In the interests of parsimony alone, it should be incumbent upon the researcher to first demonstrate the absence of an appropriate scale among general personality instruments. In the event a consumption-specific instrument is developed, at least some minimal standards of reliability and validity ought to be maintained.

Just as the nature of the predictor variables are likely to affect prediction in a given study, so too is the nature of the criterion variables. It has been noted that an atheoretical-shotgun approach has not been limited to the selection of predictor variables. Selection of the products incorporated as criterion variables has also been rather haphazard.

When the differences between product alternatives are at least clearly perceived by the consumer there is a reasonable chance of discriminating between buyers of each product. [This is probably why convertible buyers and standard car buyers can more readily be discriminated between (Westfall, 1962) than can Ford and Chevrolet owners (Evans, 1959).] In this study, where the criterion successfully isolates the variance along a specific dimension, as with the verbal product alternatives (criterion I), personality variables alone account for fifteen percent of the variance.

Criterion I and III highlight the advantages of working with groups of products clustered along relevant dimensions. Operationally, these prove to be more reliable criterion than are those consisting of individual products. Theoretically, as indicated above, we are far more likely to predict clusters of related behaviors than we are to predict any single behavioral act.

The two product alternatives offered as gifts (criterion II) demonstrate the difficulty in obtaining powerful personality instruments that discriminate amongst buyers of individual products. In both cases, only two percent of the variability in the discriminant space is attributable to group differences. Nevertheless, two personality characteristics are statistically significant discriminants in the case of the toothpastes, but not in the case of the air fresheners. This difference in itself might offer some insight as to the nature of these two criterion.

The toothpastes offered a clearly perceived choice: Ultra Brite with its emphasis on appearance ("sex-appeal") and Colgate's emphasis on the practical advantages of decay prevention (MFP Fluoride). The choice between air fresheners was less clear. Some women seemed to select the disinfectant for its practical disinfecting qualities, but some apparently selected it because of its milder aroma (a sense-oriented quality). In terms of the model, at least some S's selected the different alternatives for the same reason. When this proves to be the case it is highly likely that efforts at discriminating between selectors of one or the other Product will be most difficult.

Clearly, the nature of the criterion does make a difference. When clearly perceived product advantages can be linked directly to well defined lifestyles, personality variables may be used to advantage. Where product advantages are more ambiguously perceived, however, or cannot be unambiguously linked to a potential lifestyle, personality variables will be of considerably less value.

This difference may also be noted among the various product clusters (criterion III). Prediction is better vis-a-vis appearance-oriented product clusters. This is probably because "appearance" is a more narrowly defined dimension than is "performance" (which may include product advantages as varied as speed, cleanliness, efficiency and convenience).

Lastly, a Post-hoc remark with respect to the adequacy of the theoretical framework; there is strong reason to suspect that the results of this study were negatively affected by the increasingly important "consumerism" factor. From observation this appears to have been a rather salient dimension in the evaluation of the new products included in this study, especially those of a non-functional nature. Often, products that were once perceived and appreciated as colorful are now seen as "colorful but polluting". In other words, the paradigm may have failed to specify and measure an important dimension.

It is the more educated, higher income groups that have been more vocal in this regard. Inasmuch as they are the critical market segment that tends to adopt new products early and often, [Our sample was selected in part to reflect this fact.] this is likely to pose a major new challenge to the successful introduction of a large range of new products.

REFERENCES

Allport, G.W., Yernon, P.E. & Lindzey, G., Study of Values,(3rd ed.) Boston: Houghton Mifflin Co., 1960.

Berlyne, D.E., Novelty, Unpublished paper, Department of Psychology, University of Toronto. See also: Berlyne, D.E., Conflict, Arousal and Curiosity, New York: McGraw-Hill, 1960.

Berlyne, D.E., Structure and direction in thinking, New York: John Wiley & Sons, 1965.

Blake, B., Perloff, R. & Heslin, R., Dogmatism and acceptance of new products: Journal of Marketing Research, 1970, 7, 483-6.

Bourne, F.S., Group influence in marketing and public relations, in R. Likert and S. P. Hayes, Jr., (Eds.) Some applications of behavioral science research, Paris: UNESCO, 1957, 217-24.

Brody, R.P., & Cunningham, S.M., Personality variables and the consumer decision process, Journal of marketing research, 1968, 5, 73-7.

Bruner, J.S., Social psychology and perception, in E.E. Maccoby, T.M. Newman, and E.I. Hartley, (Eds.), Readings in social psychology, New York: Holt Rinehart & Winston, Inc., 1958, 85-94.

Campbell, D.T. & Fiske, D.W., Convergent and discriminant validation by the multitrait-multimethod matrix, Psychological Bulletin, 1959, 56, 81-105.

Cattell, R.B. & Eber, H.W., Handbook for the sixteen personality factor questionnaire, Champaign, Illinois: Institute for Personality and Ability Testing, 1957.

Cohen, J.B., The Role of Personality in Consumer Behavior, in H.H. Kassarjian & T. Robertson (Eds.) Perspectives in consumer behavior, Glenview, Illinois: Scott, Foresman and Company, 1968.

Cohen, J.B., An interpersonal orientation to the study of consumer behavior, Journal of marketing research, 1967, 4, 270-8.

Cunningham, S.M., The major dimensions of perceived risk, in D. Cox (Ed.) Risk taking and information handling in consumer behavior, Boston: Harvard University Press, 1967.

Edwards, A.L., Edwards Personal preference schedule manual, New York: Psychological Corporation, 1957.

Evans, F.B., Psychological and objective factors in the prediction of brand choice: Journal of Business, 1959, 32, 340-369.

Fishbein, M., The relationships between beliefs, attitudes and behavior, in Shel Feldman (Ed.), Cognitive consistency, New York: Academic Press, 1966, 199-223.

Goldberg, M.E., Product innovation: a study of personality factors in product perception, unpublished Ph.D. Dissertation, the University of Illinois, 1971.

Gough, H.E., Personal communication, April 15, 1970.

Gough, H.E., California psychological inventory manual, Palo Alto, California: Consulting Psychologists Press, 1957.

Gough, H.E., The adjective check list, Palo Alto, California: Consulting Psychologists Press, 1952.

Horney, K., The neurotic personality of our time, New York: W.W. Norton & Co., Inc., 1937.

Jacoby, J., Innovation proneness as a function of personality, Paper presented at the American Marketing Association's Fall Conference, Boston, 1970.

Kaponen, A., Personality characteristics of purchasers, Journal of Advertising Research, 1960, 1, 6-12.

Kassarjian, H.H., Personality and consumer behavior: a review. Paper presented at the 1970 American Marketing Association's Fall Conference, Boston, 1970.

Kassarjian, H.H., Social character and differential preference for mass communication, Journal of Marketing Research, 1965, 2, 146-53.

Kegerris, R.J., Engel, J.F. & Blackwell, R.D., Innovativeness and diffusiveness: a marketing view of the characteristics of earliest adopters, in D.T. Kollat, D. Blackwell and J.F. Engel, (Eds.), Research in Consumer Behavior, New York: Holt, Rinehart and Winston, Inc., 1970, 671-689.

Lewin, K., Field theory in social science; selected theoretical papers, D. Cartwright, (Ed.), New York: Harper, 1951.

Marlow, D., & Gergen, K., Personality and social interaction, in G. Lindzey & E. Aronson, (Eds.), The handbook of social psychology, (2nd ed.) 3, Reading, Massachusetts: Addison-Wesley Publishing Company, 1969, 590-665.

Mitchell, Jr., J.V., A comparison of the first and second order dimensions of the 16 PF and CPI inventories, The Journal of Social Psychology, 1963, 61, 151-166.

Massy, W.F., Frank, R.E. & Lodahl, T., Purchasing behavior and personal attributes, Philadelphia: University of Pennsylvania Press, 1968.

Morrison, D.G., On the interpretation of discriminant analysis, Journal of Marketing Research, 1969, 6, 156-163.

Nunnally, JOF Psychometric Theory, New York, McGraw-Hill, 1967.

Pessemier, E.A., Burger, P.C. & Tigert, D.J., Can new product buyers be identified?s Journal of Marketing Research, 1967, 4, 349-54.

Reisman, R., Glazer, N. & Denney, R., The Lonely Crowd, New Haven: Yale University Press, 1950. (Abridged edition, 1961).

Robertson, T., Consumer innovators: the key to new product success, California Management Review, 1967, 10, 361-70.

Robertson, T. & Kennedy, J.N., Prediction of consumer innovators: application of multiple discriminant analysis, Journal of Marketing Research, 1968, 5, 64-60.

Robertson, T. & Meyers, J.E., Personality correlates of opinion leadership and innovative buying behavior, Journal of Marketing Research, 1969, 6, 164-8.

Rogers, E.M., Diffusion of innovations, New York: The Free Press, 1962.

Tatsuoka, M.M., Discriminant analysis: the study of group differences, Selected topics in advanced statistics, an elementary approach. Champaign, Illinois: Institute for Personality and Ability Testing, No. 6, 1970.

Tolman, E.C., Behavior and psychological man, Berkeley: University of California Press, 1951.

Westfall, R., Psychological factors in predicting product choice, Journal of Marketing, 1962, 26, 34-40.

Wilding, J. and Bauer, R.A., Consumer goals and reactions to a communication source, Journal of Marketing Research, 1968, 5, 73-77.

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