An Empirical Assessment of Multiple Operationalizations of Involvement

Kapil Jain, University of Rhode Island
Narasimhan Srinivasan, University of Connecticut
ABSTRACT - Involvement is an important and much debated topic in consumer research. Rigorous measurements of the construct by Laurent and Kapferer and Zaichkowsky have led to scales which are finding increasing acceptance by several researchers. Building on recent studies, this paper makes an empirical comparison of the various scales in the literature :Involvement Profile (Laurent and Kapferer); Foote, Cone & Belding Planning Grid's involvement sub-scale (Vaughn); Personal Involvement Inventory (Zaichkowsky); and modifications of Zaichkowsky's scale. In so doing, the study brings further clarification and refinement to the involvement construct. A New Involvement Profile is distilled and offered for future research.
[ to cite ]:
Kapil Jain and Narasimhan Srinivasan (1990) ,"An Empirical Assessment of Multiple Operationalizations of Involvement", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 594-602.

Advances in Consumer Research Volume 17, 1990      Pages 594-602


Kapil Jain, University of Rhode Island

Narasimhan Srinivasan, University of Connecticut


Involvement is an important and much debated topic in consumer research. Rigorous measurements of the construct by Laurent and Kapferer and Zaichkowsky have led to scales which are finding increasing acceptance by several researchers. Building on recent studies, this paper makes an empirical comparison of the various scales in the literature :Involvement Profile (Laurent and Kapferer); Foote, Cone & Belding Planning Grid's involvement sub-scale (Vaughn); Personal Involvement Inventory (Zaichkowsky); and modifications of Zaichkowsky's scale. In so doing, the study brings further clarification and refinement to the involvement construct. A New Involvement Profile is distilled and offered for future research.


Involvement has been the focus of a large number of studies in the consumer research area in the past decade (Houston and Rothschild 1977, 1978; Kapferer and Laurent 1985, 1985/86; Laurent and Kapferer 1985; Richins and Bloch 1986; Vaughn 1980, 1986; Zaichkowsky 1985). A multitude of conceptualizations are available in the literature: product/brand involvement, enduring / situational involvement, cognitive / affective involvement, instrument / response involvement and so on and so forth. This diversity of views provides a very rich perspective of the construct. However, the development in conceptualization has far outpaced the developments in operationalizations, leading to a call for better measuring instruments and empirical testing (Rothschild 1984) and simpler middle-range theories that are parsimonious (Kassarjian 1978).

Toward this end, Zaichkowsky (1985) and Laurent and Kapferer (1985; Kapferer and Laurent 1985) have concentrated their efforts on rigorous scale development. This study seeks to build on their work and others' in the direction of further refinement of measures in capturing the complexity of the involvement construct.


Much of the diversity in definitions adopted by researchers in the involvement area begins at the conceptual level. Starting from Sherif and Cantril's (1947) ego-involvement to Beatty and Smith's (1987) consideration of the interactive nature of involvement, the complexity of the involvement domain has increased. In this study, we focus on just two parallel streams of theory development: Zaichkowsky's Personal Involvement Inventory (PII) and its modifications/extensions on one hand, and Laurent and Kapferer's Involvement Profile (IP) on the other - the remarkable similarity of the conceptual domains tested in these studies cries out for empirical validation.

Zaichkowsky (1985) defined involvement as "a person's perceived relevance of the object based on inherent needs, values and interests." Ratchford (1987) adopts a comparable perspective: involvement implies attention to something because it is somehow relevant or important.

However, a unidimensional approach to involvement is not sufficient to capture its complexity. Rothschild (1984) provided a broader conceptualization: "Involvement is an unobservable state of motivation, arousal or interest. It is evoked by a particular stimulus or situation and has drive properties. It's consequences are types of searching, information processing and decision making." The above definition has gained wide acceptance (Laurent and Kapferer 1985; McQuarrie and Munson 1987).

Laurent and Kapferer (1985) strongly urge that involvement should be thought of as a profile of several facets. They offered a 19-item profile of five dimensions: Importance, Pleasure, Sign-value, Risk Probability and Risk Importance. However, the five dimensions collapsed into four factors during empirical testing, with Importance and Risk Importance loading on the same dimension. In a revised 16-item profile, Kapferer and Laurent (1985, 1985/86) tapped five dimensions again, but replaced Importance with Interest. The five factor structure was confirmed, although only two scale items were available for assessing risk probability. In the latest update of the scale (Laurent and Kapferer 1989), the same five factor structure is retained, but with a change in the number of items in each subscale (Interest - 3 items; Sign - 3 items; Pleasure - 3 items; RiskImp - 3 items and RiskPro - 4 items). It is important to note that the five distinct dimensions of involvement are non-orthogonal. Strong correlational relationships were found for the following pairs of factors: Interest-Pleasure (0.55), Interest-RiskImp (0.50), and Pleasure-Sign (0.47). The empirical evidence supporting this multi-faceted conceptualization comes from large numbers of nonstudent subjects in France, over a number of years and spanning numerous products. However, since the scale was developed using French items and has not been published in its entirety, the potential for it's usage in the U.S. has been limited (Mittal and Lee 1988).

Coming to the other research stream of interest for this study, Zaichkowsky's empirical validation of the 20-item PII also covered numerous products. The PII appeared to capture one major factor - Relevance - across all product categories, and some minor factors, which were not considered further. The major factor accounted for approximately 70% of the variance. The PII is appealing in its simple structure (20 pairs of adjectives) and the single score used to represent the degree of involvement is useful for easy comparison of products along a continuum. In this fashion, it has been used by other researchers (Celsi and Olson 1988).

The scope of the PII was expanded by McQuarrie and Munson (1987), who concurred with Laurent and Kapferer's emphasis on a multidimensional involvement profile. McQuarrie and Munson felt that the PII reflected two dimensions of the Involvement Profile, thought to be Importance and Pleasure. So, in their revision of the PII (RPII hereafter), they sought to incorporate Sign and Risk components and dropped certain items that were considered causes of attitudinal contamination and redundancy. Their revised version did not distinguish between Risk Importance and Risk Probability as two separate dimensions but had only a single dimension to capture the Risk aspect. However, their attempt to capture the Sign sub-scale was not successful. They too found significant correlations among the sub-scales: Importance-Pleasure (0.60), Pleasure-Risk (0.49) and Importance-Risk (0.41).

Higie and Feick (1989) in measuring enduring involvement, borrowed items from Zaichkowsky's PII and Mcquarrie and Munson's RPII for their Hedonic (Pleasure) sub-scale and developed a reliable sub-scale for their Self-Expression (Sign) dimension. Therefore, the pool of items from these three studies (Zaichkowsky, McQuarrie and Munson, and Higie and Feick) collectively taps the same facets of involvement as does the involvement profile (IP) proposed by Laurent and Kapferer.

In summary, we have two streams of measure development that appear to share the same conceptual framework of a multi-faceted involvement construct. Or, are we really comparing apples with oranges? Speculations have been made about the overlap/congruency of the dimensions of involvement as used by earlier researchers (Ratchford 1987). However, empirical support for such inferences is not presently available. Hence, we decided to test the comparability of the two sets of items for domain congruency, potential differences in sub-scale reliabilities and relationship with the consequences of involvement.


The objectives of the present study included the following:

i) translate Laurent and Kapferer's Involvement Profile (French items into English),

ii) replicate scale structures (IP, PII and modifications),

iii) compare the factor structures of the two sets of items, which conceptually can be expected to be similar,

iv) assess correspondence/domain overlap between the various pooled items,

v) abstract a subset of items (three in each sub-scale) that best captures the various aspects of involvement as a potential refinement of either set. and

vi) test the performance of the abstracted scale.

Sources of scale items

We are grateful to Laurent and Kapferer for providing us with the original French items for the latest revision of their Involvement Profile, along with a tentative English translation. We used four other translators (two Americans and two Europeans) for independent translations and arrived at items 1-16 shown in Table 2. The other major sources were the published items of Zaichkowsky's PII, and McQuarrie and Munson's RPII, and Higie and Feick's study (items 2049 in Table 2). Lastly, the published involvement items (items 17-19 in Table 2) used in the FCB grid were added to the pool (Ratchford 1987). The measure of involvement in the FCB grid had only three items. Its parsimony and the advertising industry's acceptance made it attractive for inclusion. Besides, there was a need for some "better" risk measures. Thus, there were 49 items in total, half of which were reverse coded.

It is significant to note two possible sources of bias: one involving translation and the other methodological. Responses to the English items may not necessarily yield a pattern similar to that of the original French items, not only because of the translation (differing usage of idioms and expressive aspects of the language), but also because of the differences in the marketplace settings -- product offerings, consumer tastes, and culture. The methodological issue concerns the reformulation from the original likert-type format to a semantic differential format. The change in format was necessitated by the requirements for conformity with the rest of the items used in the study. In this respect, Jaccard, Weber, and Lundmark (1975) have shown that the two types of scales yield similar results, and hence this may not be a serious limitation.

Data Collection

Ten products were selected after a review of previous studies. Multiple products were chosen to retain the generalizability across product classes, similar to other researchers. The ten products were: alarm clock, batteries, calculator, chocolate, cologne/perfume, detergents, haircut/styled, music tapes/records, newspaper and radio. These products were chosen to reflect a spectrum of involvement profiles and in light of students' familiarity with the products. To reduce tediousness of the task and respondent fatigue and boredom, five pairs of products were established at the outset and each respondent completed the scales for only one pair of products.

A convenience sample of 375 student respondents, almost all of whom were undergraduates, was drawn from two major north eastern universities. Each respondent received two random orders of the pool of 49 semantic-differential items, one for each product in the pair. The product order in the pair was also randomized. A total of 735 usable responses, pooled across all products, were available for analysis.




Replication of Previous Scale Structures

Responses were pooled across all the ten products for the analysis, and oblique factor analysis was used. The replication analysis of the various scales yielded consistent and encouraging results, both in the factor structures and Cronbach alpha values of the sub-scales.

The IP items yielded four factors: Interest and Pleasure loading on the same dimension, Sign, Risk Probability and Risk Importance. Together, the extracted factors explained 61% of the variance (Replication of the IP is discussed later). The PII items yielded two factors, accounting for 61% of the variance (major factor = 48%; minor factor = 13%). The first factor reflected Relevance/Importance and the second factor reflected Pleasure, as was expected. The inter-factor correlation was 0.46. Three factors were recovered for McQuarrie and Munson's RPII, that together explained 66% of the variance (46%, 11% and 9% for the three factors respectively). The three factors were Pleasure, Importance and Risk. The inter factor correlations were as follows:

Pleasure-Importance = D.59, Pleasure-Risk = 0.35, and Importance-Risk = 0.29. The structure of the factor analysis reported by Higie and Feick was also recovered (77% of the variance was explained). The correlation between-the two dimensions, Pleasure and Sign, was 0.56. The three FCB items yielded a single factor, recovering 72% of the variance.

Table 1 shows the Cronbach alpha values obtained in the present study as well as those reported in the literature. It can be seen that there is a great deal of consistency. Similar to the finding by Kapferer and Laurent (1985), the internal consistency of the Risk Probability sub-scale was not high (0.54 vs 0.57). McQuarrie and Munson's Risk items yielded a slightly higher alpha value of 0.67. The inclusion of the FCB scale to bolster the assessment of Risk is, therefore, seen to be justified, and its alpha value was 0.81. All the other sub-scales had alpha values in the range of 0.67 to 0.94.

Comparison of The Two Sets of Involvement Profile Items

The two sets of items were (1) the English translation of Laurent and Kapferer's (1989) latest revision of the IP (study 1: items 1-16 in Table 2), and (2) the combination of items from the PII and its modifications (studies 3, 4 and 5: items 20 - 49 in Table 2). Each set of items was factor analyzed separately with data pooled across all the products.



Items 1-16, which formed the first set (taken from the Involvement eigenvalue criteria) that together explained 61% of the total variance (the varimax version showed the variance explained by Interest/Pleasure, RiskImp, Sign and RiskPro factors to be 36%, 12%, 7% and 6% respectively). Oblique rotation of the factors was used for interpretation. The Sign, RiskPro and RiskImp sub-scales were recovered as distinct dimensions. However, the subscales for Interest and Pleasure loaded on the same factor. One may recall that factors corresponding to these two sub-scales had the highest correlation amongst all the factors in the Kapferer and Laurent (1985) study. The inter-factor correlations of Interest/Pleasure with the Sign, RiskPro and RiskImp dimensions were 0.47, -0.08 and 0.46 respectively. Sign was correlated 0.02 and 0.45 with RiskPro and RiskImp respectively. The two dimensions of risk were correlated 0.11.

Items 20-49, which formed the second set were also factor analyzed. Again, using the scree test and eigenvalue criteria, four oblique factors were recovered, accounting for 64% of the variance. The dimensions which emerged were: Relevance, Pleasure, Sign and Risk (these accounted for 40%, 14%, 6% and 4% respectively in the varimax version). Relevance had a correlation of 0.44, 0.32 and 0.15 with Pleasure, Sign and Risk respectively. Pleasure was correlated 0.56 and 0.33 with Sign and Risk respectively. Sign and Risk had a correlation of 0.47.

It may be observed that the two sets, which were thought to be parallel versions, did not yield similar factor dimensions. Only Pleasure and Sign were common facets. Relevance emerged as a dominant dimension in the second set, but was absent from the first. The first set showed the two facets of risk, RiskImp and RiskPro, to be distinct, while all items related to risk loaded on the same factor in the second set. Hence, it may be expected, that were both sets to be combined and factor analyzed, five dimensions would result (Relevance, Interest/Pleasure, RiskImp, Sign and RiskPro).

Testlng Domain Overlap

Stewart (1981) recommends that a useful application of factor analysis is for testing the domain of various items which relate to the same construct. The present setting provides such an opportunity. Items 149 (Set 1, Set 2 and the FCB items) were subjected to an oblique factor analysis. As observed earlier, there is no theoretical justification for orthogonal dimensions. The scree test and eigenvalue criteria revealed five dimensions, as hypothesized. The emergence of the five dimensions from the increased set of variables indicates that the factors are robust.

Table 2 shows the factor loadings of the complete set of 49 items. The five factors are Relevance/Importance, Pleasure/Interest, Sign/Symbolic, Risk Importance and Risk Probability, explaining 59% of the total variance (in the varimax version, the factors accounted for 36%, 10%,-6%, 4% and 3% resp.). The pattern of the item loadings revealed that there were overlapping dimensions from the various scales. The first factor was the unique Relevance aspect of the PII. The second factor, Pleasure showed high loadings of items from the Pleasure/Interest subscale of the IP, and the Pleasure sub-scale of the RPII. The Sign factor had high loadings of items belonging to the Sign sub-scale of the IP, and the Self-Expression component of Higie and Feick's Enduring Involvement scale. The RiskImp factor had high loadings from items in the RiskImp subscale of the IP, the Risk sub-scale of the RPII, and all the three items from the FCB scale. Finally, the RiskPro factor reflected the RiskPro sub-scale items from the IP and the Risk items from the RPII.

Two salient findings emerge from Table 2. First, the Relevance component of involvement is totally untapped in the current battery of IP items. Interestingly, in Laurent and Kapferer's study reported in JMR (1985), the Importance items had loaded with Risk Importance and were replaced with Interest items in their later ACR and JAR studies (Kapferer and Laurent 1985, 1985/86). Perhaps, a different set of Importance items in the JMR study might have led to a distinct dimension, since Zaichkowsky's Relevance factor does have Importance related adjective pairs, such as "important/unimportant," "needed/not needed," and "essential/non-essential." Also noteworthy is that the RiskImp sub-scale remained as a distinct factor and did not load on the Relevance factor in the domain analysis.

The second salient finding relates to the mapping of Risk in the RPII on the Risk-related dimensions of IP. The three items of Risk load about equally on RiskImp and RiskPro factors. Hence, an argument can be made for retaining the two components of risk as distinct facets of IP, if more information is considered desirable. A counter argument, using a parsimony criterion, could be that the Risk sub-scale of the RPII may be substituted for the two facets of risk in the IP, since the significance of RiskImp may have been boosted by the inclusion of the FCB items in this analysis. However, dropping the three FCB items still yielded the same five factors as in Table 2, with RiskImp and RiskPro remaining distinct.

Table 3 shows the inter-factor correlations of the five dimensions. The correlations of Sign with Pleasure and RiskImp were both 0.58, which is the highest value in the table. Relevance correlated moderately with Pleasure (0.42), Sign (0.33) and RiskImp (0.35). Pleasure correlated strongly with RiskImp (0.49), but not with RiskPro (0.05). The remaining correlations for RiskPro were also low 0.21 with Sign and 0.23 with RiskImp. While Relevance, Pleasure, Sign and RiskImp appear to be related among themselves, RiskPro stands apart as a unique dimension by itself.

Abstracted Items for Sub-Scales and Performance Testing

Capitalizing on the strengths of different subscales offered by earlier researchers; we would now like to offer for further testing a concise set of items, reflecting a multi-faceted operationalization of involvement. The objectives of the pruning phase were the same as those of Laurent and Kapferer (1985/86): short scales (for convenience), multiproduct applicability (for generalizability) and single factored, reliable sub-scales (for psychometric purposes). The three items loading highest on each factor in Table 2 were chosen to yield a smaller set of 15 items for the New Involvement Profile (capitalized in Table 2). Particulars are shown below


The performance of the New Involvement Profile was compared to the other available scales in terms of ability to predict some of the consequences of involvement used in previous studies: greater information search, perception of differences among brands, and preference for a particular brand (Zaichkowsky 1985, McQuarrie and Munson 1987). Three items were used to capture the first proposition: "I would be interested in reading about this product," "I would pay attention to an ad for this product," and "I would compare product characteristics among brands for this product." The last two consequences were assessed using one item for each: "I think there are great differences among brands of this product" and "I have a most preferred brand of this product" respectively. All five statements were rated on a five-point scale from strongly disagree (1) to strongly agree (5). The three statements for greater information search were highly inter-correlated (alpha = 0.78) and were therefore combined into a single scale.





The results are summarized in Table 4. The predictive ability (adjusted R-squared) of the New Involvement Profile is seen to match the best of the earlier scales. The individual standardized regression coefficients for the five dimensions vary in significance across the five consequences of involvement used as the criterion variables. Every dimension of the New Involvement Profile has a significant impact on at least one consequence, thereby justifying retention of the five dimensions.


A multidimensional approach to measuring involvement was followed in this study. An English translation of Laurent and Kapferer's Involvement Profile and an alternative measure, derived from Zaichkowsky 's Personal Involvement Inventory and its modifications, were tested empirically using multiple products. As shown in Table 5, the range of products studied captured a spectrum of involvement scores as measured by Zaichkowsky's PII (1985) and Laurent and Kapferer's latest Involvement Profile (1989).



Empirical domain testing shows that Zaichkowsky's PII and Laurent and Kapferer's IP have unique components. A concise 15 item New Involvement Profile is abstracted from the sub-scales tested in this study, drawing on Zaichkowsky's PII measurement of the Relevance/Importance dimension, and the latest IP's contribution to the other four dimensions: Pleasure, Sign, Risk Importance, and Risk Probability. Obviously, this refinement process needs to be carried still further for understanding the intricacies of the Involvement construct. Specifically, the RiskPro subscale needs to be strengthened and the New Involvement Profile needs further validation on a fresh sample.


Beatty, S. E. and Scott M. Smith (1987), "External Search Effort: An Investigation Across Several Product Categories," Journal of Consumer Research, Vol. 14, (June), 83-95.

Bloch, P. H . (1983), "Involvement Beyond the Purchase Process: Conceptual Issues and Empirical Investigation," in Advances in Consumer Research, Vol. 9, A. A. Mitchell, ed., 413-417.

Bloch, P. H. and M. L. Richins (1983), "A Theoretical Model for the Study of Product Importance Perceptions," Journal of Marketing, 47, 3 (Summer), 69-81.

Cohen, J. B. (1983), "Involvement: Separating the State from Its Causes and Effects," Working Paper # 33, Center for Consumer Research, University of Florida.

Higie, R. A. and L. F. Feick (1989), "Enduring Involvement: Conceptual and Measurement Issues," in Advances in Consumer Research, Vol. 16 (forthcoming).

Houston, M. J. and M. L. Rothschild (1977), "A Paradigm for Research on Consumer Involvement," Working Paper # 11-7746, University of Wisconsin-Madison.

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

Jaccard, J, J. Weber and J. Lundmark (1975), "A Multitrait-Multimethod Analysis of Four Attitude Assessment Procedures, Journal of Experimental and Social Psychology, 11, 149-154.

Kapferer, J. N. and G. Laurent (1985), "Consumers' Involvement Profile: New Empirical Results," in Advances in Consumer Research. Vol. 12, E. C. Hirschman and M. B. Holbrook, eds., Association for Consumer Research, 290-295.

Kapferer, J. N. and G. Laurent (1985/1986), "Consumer Involvement Profiles: A New Practical Approach to Consumer Involvement," Journal of Advertising Research, Vol. 25, 6 (December), 4856.

Kassarjian, H. (1978), "Anthropomorphism And Parsimony," Presidential Address to the Association for Consumer Research, in Advances in Consumer Research, H. K. Hunt, ed., Vol. 5.

Laurent, G. and J. N. Kapferer (1985), "Measuring Consumer Involvement Profiles," Journal of Marketing Research, XXII (February), 41-53.

Laurent G. and J. N. Kapferer (1989), Personal Communication, (Fax), Feb.

McQuarrie, E. F. and J. M. Munson (1987), "The Zaichkowsky Personal Involvement Inventory: Modification and Extension," in Advances in Consumer Research, Vol. 14, P. Anderson and M. Wallendorf, eds., Association for Consumer Research, 36-40.

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

Mittal, B. and M. S. Lee (1988), "Separating Brand-Choice Involvement from Product Involvement Via Consumer Involvement Profiles," in Advances in Consumer Research, Vol. 15, M. Houston, ed., Association for Consumer Research, 43-49.

Muncy, J. A. and S. D. Hunt, "Consumer Involvement: Definitional Issues and Research Directions," in Advances in Consumer Research, Vol. 11, T. C. Kinnear, Ed., Association for Consumer Research, 193-196.

Ratchford, B. T. (1987), "New Insights About The FCB Grid," Journal of Advertising Research, August/September, 24-38.

Richins, M. L. and P. H. Block (1986), "After the New Wears Off: The Temporal Context of Product Involvement," Journal of Consumer Research, Vol. 13, (September), 280-285.

Rothschild, M. L. (1984), "Perspectives on Involvement: Current Problems and Future Directions," in Advances in Consumer Research, Vol. 11, T. C. Kinnear, ed., Association for Consumer Research, 216-217.

Sherif, M.- and H. Cantril (1947), The Psychology of Ego-Involvement. New York: John Wiley & Sons, Inc.

Stewart, D. W. (1981), "The Application and Misapplication of Factor Analysis in Marketing Research," Journal of Marketing Research, Vol. XVIII (February), 51-61.

Vaughn, R. (1980), "How Advertising Works: A Planning Model," Journal of Advertising Research, 20 (October), 27-33.

Vaughn, R. (1986), "How Advertising Works: A Planning Model Revisited," Journal of Advertising Research, February/March, 57-66.

Zaichkowsky, J. L. (1985), "Measuring the Involvement Construct," Journal of Consumer Research, Vol. 12 (December), 341-352.