An Exploration Into the Scaling of Consumers' Involvement With a Product Class

Peter H. Bloch, Portland State University
ABSTRACT - Product involvement as an explanatory or moderating variable with respect to consumer behavior has recently become a topic of significant interest to marketers. However, the usefulness of the construct in consumer research is yet to be fully realized due to a lack of agreement and rigor in operationalizing product involvement. This paper addresses the need for valid measures of individual differences in involvement for particular product classes and reports on the psychometric development of a prototype scale to assess involvement with automobiles. Findings provide preliminary support as to the scale's reliability and validity.
[ to cite ]:
Peter H. Bloch (1981) ,"An Exploration Into the Scaling of Consumers' Involvement With a Product Class", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 61-65.

Advances in Consumer Research Volume 8, 1981      Pages 61-65

AN EXPLORATION INTO THE SCALING OF CONSUMERS' INVOLVEMENT WITH A PRODUCT CLASS

Peter H. Bloch, Portland State University

[The author thanks Marsha L. Richins and Robert A. Peterson for their comments and contributions to this work.]

ABSTRACT -

Product involvement as an explanatory or moderating variable with respect to consumer behavior has recently become a topic of significant interest to marketers. However, the usefulness of the construct in consumer research is yet to be fully realized due to a lack of agreement and rigor in operationalizing product involvement. This paper addresses the need for valid measures of individual differences in involvement for particular product classes and reports on the psychometric development of a prototype scale to assess involvement with automobiles. Findings provide preliminary support as to the scale's reliability and validity.

INTRODUCTION

The ways in which consumers are involved with the products they purchase and use has, in recent years, become a subject of increasing interest to consumer researchers. This interest is evidenced by the special sessions on the topic of involvement at the American Marketing Association's 1977 Attitude Research Conference and at the 1978 Conference of the Association for Consumer Research. In addition, journal articles dealing with some aspect of involvement have become increasingly common (e.g., Rothschild 1979b, Tyebjee 1979b, Webb 1979). While in these and other studies, product involvement has been defined in various ways, the definition proposed by Day (1970) is characteristic of several in the consumer literature. Day defined involvement as "the general level of interest in the object or the centrality of the object to the person's ego-structure" (p. 45). Several writers (Bogart 1967, Mitchell 1979) have agreed that involvement reflects the amount of interest evoked by the product while, Day's centrality notion has found support from Tyebjee (1979a) and others (DeBruicker 1979, Houston and Rothschild 1978. Lastovicka and Gardner 1979) who suggested that involvement occurs when a product is related to important values, needs or the self-concept.

To date, involvement researchers generally agree that for consumers as a group, products differ in their tendency to arouse involvement. For example, Robertson (1976) has emphasized the low level of involvement generated by many consumer products. In addition to differences across products, several writers (Houston and Rothschild 1978, Lastovicka and Gardner 1979, Tyebjee 1979a) have suggested that for any particular product class, levels of involvement will differ across consumers. The focus of this research is on this between-individuals perspective.

Reviewing the research on individual differences leads to the conclusion that highly involved individuals engage in more complex purchase decision making (Houston and Rothschild 1978), are more brand loyal (Jacoby 1971), are more likely to be opinion leaders (Corey 1971) and are more likely to generate negative cognitive responses to product-related messages (Houston and Rothschild 1978, Wright 1973) than do persons with low involvement. Given this research stream, product involvement appears to be not only a potentially useful moderating or explanatory variable with respect to consumer behavior, but also a possible basis for segmenting markets.

Despite the promising nature of involvement in consumer research, several authors have noted that its explanatory potential is, as yet, unrealized due to a lack of consensus and rigor in operationalizing the construct (Mitchell 1979, Ray 1979, Tyebjee 1979b). Although many measures of individual differences in involvement have been proposed since the concept's introduction to the marketing literature, few have been subjected to standard reliability and validity testing and at the present time there is no single, well-accepted involvement measure. [For a fuller treatment of issues and problems relating to the measurement of involvement, see Mitchell 1979, Houston and Rothschild 1978, and Ray 1979.]

In light of the prominent measurement difficulties relating to product involvement research, Mitchell (1979) has argued that the "first priority" in future field research in involvement is the development of a scale to measure consumers' involvement with a particular product class or brand (p. 194). He went on to say that only when such a scale is developed and subjected to traditional psychometric testing can product involvement be fruitfully related to other aspects of consumer behavior.

Given this call for action, the purpose of this paper is to report on an attempt to scale consumers' involvement with a particular product class and to provide a prototype involvement scale as one step in the further refinement of the involvement construct. This paper next reports on the development of a scale to measure consumer involvement with automobiles.

CHOICE OF PRODUCT TO BE STUDIED

In developing an involvement scale, the first critical decision that a researcher faces is whether the measure should be applicable to several products or to a single product. As part of their work on the dimensionality of a product involvement, Lastovicka and Gardner (1979) devised a scale able to tap individual differences in involvement for any product class. The benefits of their measure's generalizability appear to be balanced, however, by the imprecise character of its constituent items. Many of the Lastovicka and Gardner scale items appear ambiguous and may not capture how a consumer really feels about any one product. Examples include: "I can make many connections or associations between experiences in my life and this product'' and "use of this product helps me behave in the manner that I would like to behave" (p. 63). When scale items are specific to a single product class, they would appear to be more relevant and meaningful.

In a critique of involvement measures, Ray (1979) posited that such measures should be developed "in individual consumer research application situations" (p. 198). He cited the involvement measure developed by Rothschild (1974) in a study of political races and that used by Webb (1979) in research on a television clutter as particularly effective in that they were tailored to specific research situations. In a similar vein, Mitchell (1979) suggested that future involvement research might be based on the use of a scale(s) measuring the degree of interest or arousal evoked by a particular product class or brand.

A further argument for building a product-specific measure rests on the state of current understanding of the construct of product involvement. As noted by several authors (Lastovicka and Gardner 1979, Houston and Rothschild 1978, Mitchell 1979, Tyebjee 1979b), the literature on involvement is fragmented and for the most part based on measures of dubious quality. Therefore, because involvement theory remains in a developing state, the search for relationships between involvement and other aspects of consumer behavior appears to be worthwhile even if in a product-specific context.

Once the product-specific focus was chosen, the next decision concerned the choice of the product to be studied. After reviewing several possible options, automobiles were selected as the product class of interest due to the observation that car owners exhibit wide variation in involvement. For instance, a person who has minimal involvement with cars may drive a plain model and treat the vehicle essentially as an appliance, whereas another person with high involvement (such as in the case of car enthusiasts) may spend his weekends tinkering with cars and subscribe to several automotive magazines. This potential for variance and the familiarity of the product class to most consumers made automobiles a reasonable choice for scale development efforts.

PSYCHOMETRIC DEVELOPMENT

The literature on scaling presents a bewildering array of possible approaches to constructing a multi-item scale. The particular approach employed in this project is that presented by Churchill (1979) in an article on the development of measures for marketing constructs. The sections which follow discuss the steps used in specifying the domain of the construct, generating scale items, measure purification procedures, and the analysis of the scale's structure, as well as the reliability and validity of the final involvement scale.

Specifying the Domain of the Construct

The first step in developing a scale is to specify the domain of the construct under study, based on a review of relevant literature (Churchill 1979). A review of conceptual work on product involvement (e.g., Bogart 1967, bay 1970, Houston and Rothschild 1978, Lastovicka and Gardner 1979, Tyebjee 1979a), indicated the construct has generally been conceptualized in one of two ways. First, involvement has been treated as a temporary concern with high risk products which occurs during the purchase process (Bowes and Charles 197&, Houston and Rothschild 1978). Product involvement has also been viewed as a long-term interest in a product which is based on the centrality of the product to important values, needs, or the self-concept (Day 1970, DeBruicker 1979, Houston and Rothschild 1978, Tyebjee 1979a). According to Houston and Rothschild (1978) this second form of involvement is primarily a function of individual differences and as such, is appropriate for present scaling efforts. The temporary or situational form of involvement will not be studied here because it appears to offer little beyond another perspective on perceived risk.

As used in this research, product involvement is a construct which affects consumer behavior on an ongoing basis. Further, involvement varies across individuals, ranging from minimal levels to the extremely high levels exhibited by consumers such as car enthusiasts, wine connoisseurs, or camera buffs. Procedures used to generate scale items appropriate to the above conceptualization of product involvement are discussed in the following section.

Generating Scale Items

In order to achieve content validity for the scale, the item development process vas also based on the review of the product involvement literature. In addition, interviews were held with automobile dealers and sports car enthusiasts, persons presumed to be highly interested and involved in cars and therefore, able to shed light on the nature of the construct. As a final source of scale item ideas, advertising and editorial content of various automotive magazines were reviewed.

This exploratory work resulted in the construction of 66 statements which captured several aspects of involvement noted in the literature. Items dealt with (1) interest in cars (Bogart 1967, Day 1970, Mitchell 1979) and accompanying readiness to talk about cars (Bogart 1967, Corey 1971), (2) the relatedness of cars to important needs or values (Houston and Rothschild 1978, Tyebjee 1979a), and (3) use of one's car to express the self-concept (DeBruicker 1979, Lastovicka and Gardner 1979). Other items which were developed from the experience survey tapped highly product-specific domains such as willingness to perform car maintenance and devotion to the automobile in light of energy and ecological pressures.

The next step in item development involved editing the raw statement set. A group of six judges, consisting of both faculty and doctoral students from the areas of consumer behavior and social psychology, evaluated the 66 items for content validity (i.e., whether or not the items appeared to tap an abiding interest and concern with automobiles). Items which were deemed irrelevant were eliminated along with those which did not produce a consensus as to their direction of scoring. A 44-item preliminary scale emerged from the item development stage.

The scale consisted of items scored according to a six-point Likert-type format. This is the type of scale recommended by Houston and Rothschild (1978) for use in assessing individual differences in product involvement. Responses are scored so that low scores represent high involvement with negatively stated items to be reverse-scored prior to analysis. An individual's scale score is obtained by summing across the items.

This preliminary version of the involvement scale was administered to a group of 381 marketing students in large classroom settings. Classroom administrations were utilized for reasons of convenience and to obtain a nearly perfect response rate. This response rate benefit becomes salient in the case of research on involvement with a single product in that involvement would appear to be a determinant of response. For example, it seems reasonable that individuals highly involved with a product are more likely to complete a mail survey or personal interview that deals with that product than are subjects with low involvement. The responses to the 44-item scale were used in the next phase of scale development, measure purification.

Measure Purification

The major thrust in purifying a multi-item scale is to reduce, as needed, the number of scale items, with the goal of producing an internally-consistent instrument with highly intercorrelated items. This argument is based on the domain sampling model of measurement (Nunnally 1967) which "holds that the purpose of any particular measurement is to estimate the score that would be obtained if all the items in the domain were used" (Churchill 1979, p. 68). When all the items of a scale are highly intercorrelated, it may be concluded that they are drawn from the domain of a single construct .'

The most widely accepted indicator of s scale's internal consistency is Cronbach's (1951) coefficient alpha (a) which represents the average of all possible split-half reliability coefficients. Alpha will tend to be large when scale items are highly intercorrelated and thus appear to be tapping the same construct. [For a fuller discussion of Cronbach's alpha, see Peter 1979.]

In this phase of scale development, an iterative procedure was followed where alpha was calculated, items with low item-total correlations were discarded, followed by a subsequent alpha calculation. The 44-item preliminary scale produced an alpha of .82. Although this alpha greatly exceeds Nunnally's (1967) criterion for acceptable internal consistency of .50 to .60, further analysis revealed that several scale items had low item-total correlations. A series of alpha calculation iterations produced a set of 17 items which attained an alpha of .83. While this appears to be only a marginal increase in internal consistency, it should be noted that alpha is affected by scale length with more items tending to increase alpha levels (Cronbach 1951, p. 313). For purposes of comparison, the alpha level of the 17-item set was corrected to make it equivalent in length to the 44 item set. Using the formula below, if the 17 item set was lengthened to 44 items while maintaining the same internal consistency, alpha would be .93 (Nunnally 1967, p. 223). This corrected alpha indicates that the

EQUATION    (1)

where:

ac = corrected for scale length

k = number of times the scale is to be lengthened

a = uncorrected alpha

measure purification process not only produced a more parsimonious set of items, but also one which possesses a significantly higher level of internal consistency. The 17 items along with their item-total correlations are displayed in Table 1.

Scale Structure

Responses to the reduced set of items shown in Table 1 were next used in a principal components analysis to investigate the dimensionality of the scale and shed further insight on the nature of the involvement construct. Although many scale development articles perform principal components or factor analyses at the outset and prior to measure refinement work, Churchill (1979) noted that in such instances, a large number of difficult to define dimensions tend to emerge due to the presence of irrelevant item. It is Churchill's position that principal components analyses are more profitably employed after an internally consistent set of items is obtained via the assessment of coefficient alpha.

In performing the principal components analysis, the eigenvalue > 1.00 criterion was used to limit the number of factors extracted. Table 2 presents the varimax rotated factor loadings matrix. After examining the content of those items with the highest loadings, the six factors which emerged may be interpreted as follows:

Factor 1: Enjoyment of driving and usage of cars

Factor 2: Readiness to talk to others about cars

Factor 3: Interest in car racing activities

Factor 4: Self-expression through one's car

Factor 5: Attachment to one's car

Factor 6: Interest in cars

These dimensions, although specific to automobiles, appear to capture various aspects of product involvement well established in the consumer behavior literature. Factors 1, 4, and 5 capture the centrality orientation of involvement, while Factors 2, 3, and 6 concern product interest. The dimensionality of product involvement originally used to generate the scale items appears to have reasonable empirical support.

TABLE 1

17 ITEM INVOLVEMENT SCALE

Reliability Assessment

Although the preceding analyses produced a scale with high internal consistency, Churchill recommends that alpha be calculated on a new sample to assess the scale's internal consistency reliability. Using a new sample of 57 students, responses to the 17-item scale produced an uncorrected alpha of .79. This figure is slightly less than that calculated with the larger sample; however, its magnitude is still well above the .50-.60 level of acceptability, particularly in light of scale length.

Internal consistency reliability, estimated by coefficient alpha, assesses the measurement error which is attributable to scale content. Alpha does not estimate measurement error or unreliability caused by factors which are external to the scale, such as differences in administration situations (Churchill 1979). As a further test of scale reliability the instrument was readministered to the sample of 57 students after a two week interval. Although the assessment of test-retest reliability is characterized by serious problems brought about by memory and history effects (Nunnally 1967, Peter 1979), this test was used to provide at least some indication of the scale's stability over time.

TABLE 2

VARIMAX ROTATED FACTOR MATRIX

These two administrations produced a Pearson product moment correlation of . 78 (p < .001) for summed scale scores. A correlation of this magnitude is evidence of satisfactory test-retest reliability.

Scale Validity

The final stage of the scale development process concerned the initial validation of the involvement scale. Validity deals with the extent to which a scale is measuring what it purports to measure. As Allison (1978) noted,

...for psychological traits, validity is a matter of degree rather than an all-or-none characteristic. Moreover, it must be evaluated over time and various situations, especially in the case of construct validity of a newly developed test (p. 571).

Keeping this in mind, the analyses which follow can only be regarded as providing a beginning in the validation process for the proposed involvement scale.

To assess its criterion validity, the involvement measure was administered to a group of 90 students along with a set of demographic questions and measures of reported behavior. The behavioral measures tapped the frequency of participation in a variety of activities presumed to be associated with high involvement with automobiles. The behavioral items were developed from a review of automotive magazines and depth interviews with members of sports car clubs. Frequency of participation was measured on a six-point scale, ranging from "always" to "never," with lower scores indicative of higher involvement. Frequency of pleasure driving was also investigated using a multichotomous question with six choices ranging from "more than once a week" to "never or almost never." Table 3 presents the correlations of the scale scores and the reported behavioral items. The proposed involvement scale correlated significantly in the hypothesized direction with all of the reported behavioral measures providing evidence of the scale' s criterion validity.

TABLE 3

CORRELATIONS BETWEEN THE INVOLVEMENT SCALE AND REPORTED BEHAVIORAL MEASURES

Several researchers have suggested that one measure of a scale's construct validity is whether it can differentiate the positions of groups known to differ on the construct of interest (Churchill 1979, Lundstrom and Lamont 1976). To further validate the scale and determine if it indeed behaves as expected, such an analysis was undertaken.

In order to perform this analysis, the involvement scale was administered to 52 members of sports car clubs. Members of such clubs typically attend ten or more automotive events (e.g., races, rallys, car shows) per year, subscribe to several automotive magazines, and have a considerable financial investment in the ownership of several automobiles. Thus, one can assume these individuals have high enduring involvement with automobiles. The scale was administered at the regular monthly meetings of the car clubs. Scores obtained from the car club members were contrasted with those of the 90 students used in the previous analysis. The student sample was assumed to possess a moderate degree of interest and involvement in cars relative to the car club members. The mean score for the student sample was 59.01 contrasted with 37.39 (a higher level of involvement) for the car club group. Using a t-test, the difference between the two means was significant at p < .001. In addition, significant differences were also found for all the reported behavioral items.

CONCLUSIONS

The results presented above indicate that the construct of product involvement is scalable for a particular product class using conventional procedures. Furthermore, the scale presented is a prototype for others which may be constructed for different research applications and it appears to stand up well to traditional psychometric resigns procedures. Although the scale presented is applicable only to the product class of automobiles, the product-specificity should not limit its usefulness in efforts to further refine the construct of involvement. The scale would appear to have application to a number of research situations either as a predictor variable or as a covariate. For example, involvement may be related to other consumer behavior concepts such as opinion leadership, perceived risk, innovativeness, brand loyalty, or information processing.

While the above analyses provide preliminary support for the worthiness of the proposed involvement scale, one caution is in order. This scale was developed using student samples, and further testing with other populations and survey modes is needed to further establish the scale's validity.

In summary, this paper has presented a scale to assess individual differences in consumers' involvement with the product class of automobiles as a first attempt to directly scale involvement in a particular product class. Psychometric tests of the measure's reliability and validity seem to provide adequate support that use of the scale in future research deserves consideration.

REFERENCES

Allison, Neil K. (1978), "The Psychometric Development of a Test for Consumer Alienation from the Marketplace," Journal of Marketing Research, 15, 565-575.

Bogart, Leo (1967), Strategy in Advertising, New York: Harcourt, Brace Jovanovich.

Bowen, Lawrence and Chaffee, Steven H. (1974), "Product Involvement and Pertinent Advertising Appeals," Journalism Quarterly, 51, 613-621.

Churchill, Gilbert A. (1979), "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research, 16, 64-73.

Corey, Lawrence G. (1971), "People Who Claim To Be Opinion Leaders: Identifying Their Characteristics by Self-Report," Journal of .Marketing, 35, 48-53.

Cronbach, L. J. (1951), "Coefficient Alpha and The Internal Structure of Tests," Psychometrika, 16, 297-334.

Day, George S. (1970), Buyer Attitudes and Brand Choice, New York: Free Press.

DeBruicker, F. Stewart (1979), "An Appraisal of Low-Involvement Consumer Information Processing," in Attitude Research Plays for High Stakes, eds. J. C. Maloney and B. Silverman, Chicago: American Marketing Association, 112-132.

Houston, Michael J. and Rothschild, Michael L. (1978), "Conceptual and Methodological Perspectives on Involvement" in 1978 Educators' Proceedings, ed. S. C. Jain, Chicago: American Marketing Association, 184-187.

Jacoby, Jacob (1971), "A Model of Multi-Brand Loyalty," Journal of Advertising Research, 11, 25-31.

Lastovicka, John L. and Gardner, David M. (1979), "Components of Involvement," in Attitude Research Plays for High Stakes, eds. J. C. Maloney and B. Silverman, Chicago: American Marketing Association, 53-73.

Lundstrom, William J. and Lamont, Lawrence M. (1976), "The Development of a Scale to Measure Consumer Discontent," Journal of Marketing Research, 12, 373-381.

Mitchell, Andrew A. (1979), "Involvement: A Potentially Important Mediator of Consumer Behavior," in Advances in Consumer Research, Vol. 6, ed. W. L. Wilkie, Ann Arbor: Association for Consumer Research, 191-196.

Nunnally, J. (1967), Psychometric Methods, New York: McGraw Hill.

Peter, J. Paul (1979), "Reliability: A Review of Psychometric Basics and Recent Marketing Practices ", Journal of Marketing Research, 16, 64-73.

Ray, Michael L. (1979), "Involvement and Other Variables Mediating Communication Effects as Opposed to Explaining All Consumer Behavior," in Advances in Consumer Research, Vol. 6, ed. W. L. Wilkie, Ann Arbor: Association for Consumer Research, 1970, 99.

Robertson, Thomas S. (1976), "Low Commitment Consumer Behavior,'' Journal of Advertising Research, 16, 19-24.

Rothschild, Michael L. (1974), "The Effects of Political Advertising on Voting Behavior of a Low-Involvement Electorate," unpublished Ph.D. dissertation, Graduate School of Business, Stanford University.

Rothschild, Michael L. (1979a), "Advertising Strategies for High and Low Involvement Situations," in Attitude Research Plays for High Stakes, eds. J. C. Maloney and B. Silverman, Chicago: American Marketing Association, 74-93.

Rothschild, Michael L. (1979b), "Marketing Communications in Nonbusiness Situations," Journal of Marketing, 43, 11-20.

Tyebjee, Tyzoon T. (1979a), "Refinement of the Involvement Concept: An Advertising Planning Point of View," in Attitude Research Plays for High Stakes, eds. J. C. Maloney and B. Silverman, Chicago: American Marketing Association, 94-111.

Tyebjee, Tyzoon T. (1979b), "Response Time, Conflict and Involvement in Brand Choice," Journal of Consumer Research, 6, 295-304.

Webb, Peter H. (1979), "Consumer Initial Processing in a Difficult Media Environment," Journal of Consumer Research, 6, 225-236.

Wright, Peter L. (1973), "The Cognitive Processes Mediating Acceptance of Advertising," Journal of Marketing Research, 10, 53-62.

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