The Zaichkowsky Personal Involvement Inventory: Modification and Extension

Edward F. McQuarrie, Santa Clara University
J. Michael Munson, Santa Clara University
ABSTRACT - A revised version of Zaichkowsky's (1985) Personal Involvement Inventory (PII) was developed and tested. Termed the RPII, the revision attempts to incorporate the multifaceted perspective on involvement developed by Laurent and Kapferer (1985), and also to purge the PII of some potentially problematic scale items. Findings from 136 students who rated 12 products showed the RPII to be successful.
[ to cite ]:
Edward F. McQuarrie and J. Michael Munson (1987) ,"The Zaichkowsky Personal Involvement Inventory: Modification and Extension", in NA - Advances in Consumer Research Volume 14, eds. Melanie Wallendorf and Paul Anderson, Provo, UT : Association for Consumer Research, Pages: 36-40.

Advances in Consumer Research Volume 14, 1987      Pages 36-40


Edward F. McQuarrie, Santa Clara University

J. Michael Munson, Santa Clara University


A revised version of Zaichkowsky's (1985) Personal Involvement Inventory (PII) was developed and tested. Termed the RPII, the revision attempts to incorporate the multifaceted perspective on involvement developed by Laurent and Kapferer (1985), and also to purge the PII of some potentially problematic scale items. Findings from 136 students who rated 12 products showed the RPII to be successful.


The construct of involvement has been a central concern in consumer research over the past decade. Early work focused on a dichotomy of high and low involvement products, with the latter demanding a different model of how consumers process information and make choices (Kassarjian and Kassarjian 1979; Robertson 1976). Later efforts attempted to further differentiate the concept of involvement. Thus, Houston and Rothschild (1978) distinguished situational, enduring and response involvement, and Bloch and Richins (1983), writing on product importance, distinguished instrumental from enduring importance. Over time, definitions and distinctions proliferated, to the distress of some scholars. Cohen (1983) attempted to bring order by insisting that the antecedents and consequents of involvement be considered separately from the state itself. Rothschild (1984) declared that the conceptual elaboration of the involvement construct had reached a point of diminishing returns. He argued that a consensus had formed around a definition of involvement as "a state of arousal, interest or motivation," and that the new priority should be data collection and not further conceptualization.

During 1985, two milestones were reached in the effort to ground the involvement construct. Zaichkowsky (1985), in the Journal of Consumer Research, and Laurent and Kapferer (1985), in the Journal of Marketing Research, reported the development of methodologically sound measures of involvement. These authors were careful to measure the "state" of involvement, rather than relying on indicants associated with the antecedents and consequents of this state. She result in each case is a "multi-item" scale (i.e., inventory) which survived multiple tests of validity, and which is claimed to be of general applicability across product categories. These two inventories promise to be a significant contribution.

Therefore, one notes with consternation that these separate efforts have produced two very different inventories. The Personal Involvement Inventory (PII) of Zaichkowsky treats involvement as a unidimensional construct; its 20 items are summed to produce a single score. Whereas, Laurent and Kapferer are adamant that involvement is multi- faceted, and claim that an Involvement Profile (IP) is required. They argue that a consumer's involvement cannot be expressed in a single score, because the type of involvement is as important as its level. Their 20 item scale (1985) taps four facets of involvement: perceived importance, decision risk (probability of making a mistake), sign value (whether a product reveals the consumer to other people), and a pleasure component. Only the first, and to some extent the last of these facets is represented among the items comprising Zaichkowsky's PII. While the two inventory development efforts did use different types of items (semantic differential in the PII and Likert in the IP), and different populations, the high standard of rigor adhered to in both efforts makes it difficult to explain away their divergent results on methodological grounds.

The problem is conceptual: Is involvement with a product category one thing, or many? We find Laurent and Kapferer's (1985) arguments for their IP persuasive. They point first to the tendency of researchers and managers to use involvement in association with various qualifiers: situational or enduring, personal or emotional, and so forth. Second, each of their four facets can be convincingly related to arousal, which Cohen (1983) has argued is the fundamental constituent of the state of involvement. Perceived importance, decision risk, psychosocial risk (sign value), and pleasure are all plausible sources of a greater or lesser degree of arousal. Third, their analyses demonstrate both that individual products will be ranked differently on the four facets.

Despite these good conceptual arguments for the use of the IP rather than Zaichkowsky's PII in studies of involvement, there remain two problems: (1) the full IP has never been published; (2) while the text of the measure could doubtless be obtained from the authors, there is no guarantee that translations of the 20 Likert statements into English will yield the same item structure as the French originals. Given that additional work would in any case be required before the Involvement Profile could be widely used in this country, it seems worthwhile to ask whether Zaichkowsky's Personal Involvement Inventory could not instead be adapted to reflect a more multi-dimensional perspective.

The primary goal of the current study is thus to modify the PII so as to produce a measure that will incorporate risk and sign-value components, as well as perceived importance and pleasure. Our objective is to maintain as much continuity as possible with the existing measure, while still incorporating the necessary additional material. A we did not want the final revised version to exceed the length of the present PII, it was necessary to determine candidates for elimination from among the PII's original items. A conceptual critique of the PII was developed to guide these choices; we were mindful of the strictures placed by Zaichkowsky on investigators seeking to shorten or otherwise revise the scale (footnote 1, p. 344).


Examination of the PII's scale items alerted us to the possibility of interpretational confounding. In general terms, interpretational confounding "occurs as the assignment of empirical meaning to an unobservable variable which is other than the meaning assigned to it by an individual a priori to estimating unknown parameters" (Burt 1976, p. 4). The PII seems to include two distinct groups of adjectives. One group contains items that would possess high face validity as indicants of involvement, prior to any empirical validation work (e.g., "interesting-boring"). But the other group contains terms with quite different connotations (e.g., "beneficial-not beneficial," "valuable-worthless"). In other words, the PII contains some adjectives classically associated with a state of involvement, and others normally associated with the measurement of attitude. Empirically, Zaichkowsky (1985) found all these terms to be highly consistent (coefficient alpha ranged as high as .97); but conceptually, they represent two different constructs. Thus, we suspect that interpretational confounding may be present in the form of attitudinal contamination. While a measure of involvement should be distinct from a measure of attitude, it is not clear that the PII satisfies this test (Peters and Churchill 1986).

If interpretational confounding is present in the form of attitudinal contamination, then the PII can be expected to overestimate consumers' involvement with certain types of products, particularly those that can be liked or endorsed without being experienced as arousing or interesting. Accordingly, this overestimation may be most marked for what might be termed "humble products;" i.e., everyday items which, while indispensable, are of no great importance in themselves. Conversely, it may not occur, or be noticeable, for those products which are not used by everyone, nor considered necessities; i.e., the special interest category of Lastovicka and Gardner (1979). Here attitudinal terms like "needed" or "essential" may be expected to reflect involvement equally well as arousal-based terms such as "interesting" or "exciting."

There is suggestive evidence that interpretational confounding does occur when the PII is used to scale the involvement level of various product categories. Zaichkowsky found laundry detergent to be the third most involving product among the 14 she examined. She defends these results in terms of the presumed central role of detergent in the lives of a "relatively homogeneous group of middle-aged females" (p. 346), contrasting this with the peripheral role played by a technical appliance such as a color television. Her point has merit, but remains troubling, given the exactly opposite results reported by Kapferer and Laurent (1984) and Laurent and Kapferer (1985), in two separate, large studies of French housewives. In the latter one, detergent was the lowest-scoring product on all four facets of involvement; in the former, it was among the lowest; and in both studies, detergent scored much lower than television.

While our major criticisms of Zaichkowsky's (1985) PII concern the absence of a multi-dimensional approach and the danger of attitudinal contamination, we would also note two other potential problems. First, one might question the appropriateness of terms such as "superfluous" and "mundane" if the measure were to be applied outside a University setting. In general, the syllable count for the PII seems uncomfortably high. Second, some terms in the PII appear to be redundant (e.g., "interesting-boring" and "interested-uninterested" are both included).

Despite these criticisms, the extensive and laudable validation work performed by Zaichkowsky argues for a revision of the PII rather than its abandonment. We sought both to keep the best of her item pool, and to replace potentially problematic terms.


Development of-the Instrument

Four of the 20 item pairs in the PII were discarded on priori grounds: "superfluous-vital," "mundane-fascinating," "significant-insignificant," and "fundamental-trivial." All seemed inappropriate for use with a non-college educated population. Eight new item pairs were devised. Since the PII already had many items reflective of perceived importance of hedonic value, the new items were designed to reflect the facets of decision risk or sign value (the one exception was "fun-not fun"). The 16 old and 8 new item pairs were arranged as shown in the Exhibit.

Unfortunately, preliminary analyses indicated that the four item pairs designed to measure sign value tit not cohere as a single factor. The problem lay with the third and fourth pairs listed (i.e., "heed others' wishes" and "my own business"), which were relatively uncorrelated with any of the other 22 item pairs. Perhaps they tap a joint decision-making dimension, rather than sign value. The two remaining pairs ("says something about me" and "tells me about a person"), taken directly from Kapferer and Laurent (1984), were found to have high loadings on the pleasure factor, and did not constitute a separate factor by the minimum eigenvalue test (i.e., X > 1.0). This is not surprising, since the sign and pleasure facets had the highest inter-correlation in Laurent and Kapferer (1985). These latter two items were therefore incorporated into the pleasure sub-scale in the present analysis. In the discussion below, "RPII" and "OPII" refer to two overlapping subsets of the 22 remaining scale items, as shown in the Exhibit.

Subjects and Procedure

The semantic differential instrument shown in the Exhibit was completed by 80 undergraduate and 56 MBA students. [Although each student rated a total of 12 stimulus objects, two separate lists were used, so that in all 24 objects were rated. Some of these ratings (e.g., of issues--cf. Hupfer and Gardner 1971) were taken for a different purpose and are not discussed here. As a result, the number of valid cases (students X products minus missing cases) is 899 for the analyses reported in Tables 1 and 2 and Figure 1, and 449 for those reported in Table 3.] After one week, 67 of the 80 undergraduates completed a form which measured the consequences of involvement (described below). Lastly, four weeks after the initial administration of the semantic differential instrument, 52 undergraduates completed the 22 scale items in the Exhibit a second time for automobiles and toothpaste. Thus, 104 cases were available for testing the stability of the RPII and OPII over time.

Criterion measures

Five criterion measures, modeled after those used by both Zaichkowsky (1985) and Laurent and Kapferer (1985), were developed to assess various consequences of involvement. As noted by these authors, a higher degree of involvement should produce: l) greater information search; 2) greater complexity of choice processes; 3) greater brand commitment; and 4) greater differentiation of brands. The first of these was assessed with the following two items, each measured on a five-point Likert scale: "I would be interested in reading about this product," and "I would pay attention to an ad for this product." The second consequence was measured using a single five-point Likert item: "I would compare product characteristics among brands." Inasmuch as all three of these items were found to be highly correlated (.66, .53 and .55, respectively), they were summed together (coefficient alpha s .81). Brand commitment was measured as the degree of liking for the brand now owned or regularly used (seven-point semantic differential, anchored by "like" and "dislike"). Brand differentiation was measured as the number of acceptable brands in the product category, with five response options: mine only, a couple, several, many, all. This measure was reverse-scored: the fewer the acceptable brands, the greater the brand differentiation.



Sable l shows internal consistency estimates for the inventories and sub-scales, and also their inter-correlations. Consistent with Zaichkowsky (1985), the OPII was found to exhibit a very high degree of internal consistency (.95). But, despite being composed of 3 sub-scales, the RPII also exhibits more than satisfactory internal consistency (.93). It should be noted, however, that this high alpha coefficient does not necessarily imply that the RPII (or OPII for that matter) is unidimensional. It is possible to have a high alpha for a scale which has two or more dimensions. Therefore, a subsequent factor analysis (discussed below) was used to assess dimensionality.

Within the RPII, the importance and pleasure sub-scales are very consistent, while the risk scale is somewhat less so. This same pattern was observed by Laurent and Kapferer (1985). Additionally, the inter-correlation of the importance and pleasure sub-scales is similar to that found by Laurent and Kapferer (1985); however, our measure of decision risk is more closely related to the other two sub-scales than was the case in their studies. The intercorrelation of the OPII and RPII inventories is quite high (r - .87), as would be expected given the substantial scale overlap between the two. This degree of association shows that we were successful in maintaining comparabilito with the OPII.



She test-retest results were more disappointing. However, the correlation between the two administrations of the RPII (.80) was higher than that of the OPII (.69).

Test-retest correlations for individual items ranged from .20 to .74 for the OPII, and from .20 to .75 for the RPII. These numbers are lower than the .88 to .93 range reported by Zaichkowsky for the total 20 item inventory, and the .60 to .93 range she reported for individual items. Several plausible explanations for the lower numbers in the current study exist: the longer time frame between test and retest; greater heterogeneity in the circumstances of inventory administration (i.e., completed outside the classroom); fewer people and fewer products tested; differences in the set of products that was examined; and, of course, the fact that only 16 of Zaichkowsky's 20 items were included in the OPII.

On the basis of the internal psychometric criteria discussed above, we can conclude that the RPII is on a par with the OPII. In addition, it appears that our importance, pleasure and risk sub-scales are on a par with those developed by Laurent and Kapferer.

Factor Structure of the RPII and OPII

A principal components analysis with varimax rotation of all factors with eigenvalues greater than one was performed for each involvement inventory separately (Table 2). Regarding the OPII, we found, as did Zaichkowsky, one major and one minor factor, which here accounted for 60% and 15% of the total explained variance. However, it is clear that the major factor reflects attitude more than involvement; the highest loadings are for the scale items "needed," "essential," and "beneficial." Items such as "interesting" load mostly on the second factor. To the extent that a simple sum reflects primarily the first factor, and to the extent that terms typically associated with involvement do not load heavily on that first factor, the OPII may not truly reflect the construct of involvement. The question of whether this is a practical or only a theoretical problem is deferred until the section on scaling of products below.



The results for the RPII are encouraging. Three factors were retained, accounting for 54%, 13% and 10%, or 77% in total, of the explained variance. The rotated loadings show that simple structure obtains, and that each item loads on the expected factor. These loadings are both more homogeneous, and somewhat higher than those reported in Laurent and Kapferer (1985). To test the generalizability of these results, we separated the sample into undergraduate and MBA groups, and redid the factor analysis of the RPII for each group separately. The MBA students were enrolled in an evening program, and compared to the undergraduates, were generally older, employed full-time, and more likely to be married. A comparison of the factor structures for the two groups showed both the same simple structure, and also very little difference in the loadings for individual scale items. As was the case with the reliability analyses, these factor analyses add to our confidence that the RPII holds promise as a replacement for the OPII.

Scaling of Product Involvement

If the OPII does suffer from attitudinal contamination, then this should be manifest as an overestimate of the involvement associated with certain product categories. Figure l shows the 12 products investigated, ordered from lowest to highest, based on their involvement score on the RPII. These scores are plotted by the solid line. The dashed line plots the scores of each product on the OPII. Deviations between the two lines indicate products which are scaled differently by the two measures. Visual examination of Figure l provides suggestive evidence of interpretational confounding in the OPII. Overestimation, relative to the RPII, occurs in exactly those cases --detergent, motor oil and toothpaste--where it was predicted: "humble but useful products" that everyone consumes. However, the RPII and OPII are in substantial agreement regarding most of the products examined. For instance, they are equally effective in determining that business suits are more involving than soft drinks (Figure l). What emerges is that the OPII may systematically err in estimating the degree of involvement for certain products vis-a-vis others. This finding, of course, holds true only to the extent that the ordering of products due to the RPII does not also, in the reader's eyes, suffer from interpretational confounding.



Prediction of the Consequences of Involvement

The final comparison of the OPII and RPII concerns their ability to predict some of the consequences which are supposed to follow from involvement: greater brand commitment, greater brand differentiation, and more information search and choice complexity. Regression analyses were used to compare the degree of association between the OPII and the consequences of involvement with that between the three sub-scales of the RPII and the consequences of involvement. If involvement is multi-faceted, as Laurent and Kapferer claim, then inclusion of each of the subscales in the regression analyses should explain more variance than when the OPII is entered as sole predictor. On the other hand, if Zaichkowsky is correct and involvement is unidimensional, then the more psychometrically powerful OPII measure, consisting of 16 highly consistent items, should be more successful than the three "fragments" of involvement represented by the RPII sub-scales.

Three regressions were run for each consequence of involvement (Table 3): l) one where the OPII was the only predictor variable (left most column); 2) one where the three RPII sub-scales were the only predictors (column 2); and 3) one where the three sub-scales were added to an equation containing the OPII, and the increment in explained variance determined (in-text tabulation). Looking at the first and second columns, in every case the three sub-scales of the RPII are equal to or superior to the OPII in explaining variation in the consequences of involvement. Since a comparison of single and multiple correlations is inherently unfair, it is important to note that even when the OPII is entered first into the regression equation, the sub-scales are able to explain significant additional variation when they are added (R2 increments - .023, .023 lp < . 01] and .060 [p < . 001], respectively, for the three criteria . This is the more remarkable because of the substantial overlap in item content. The inference must be either that involvement is best treated as multi-faceted, or that the OPII contains items which contribute noise rather than signal. Probably both are true.

As predicted by Laurent and Kapferer, the beta coefficients for the individual sub-scales vary across the consequences of involvement. Importance (Table 3, column 3) is strongly predictive of brand commitment and brand differentiation, while pleasure (column 4) is the best predictor of information search and choice complexity. Decision risk (column 5) is the weakest of the three subscales, a finding also reported by Laurent and Kapferer (1985). However, it does show a significant inverse relation to brand commitment, and a significant direct relation to information search and choice complexity. (Because multicollinearity among the predictors can make for unstable beta coefficients, we split the sample in half and recomputed these regressions. The relations just described continued to hold for both sub-samples.) Allowing for some differences in the way the consequences of involvement were measured, both the pattern and the level of the associations between the sub-scales and the measures of consequences are comparable to those reported in Laurent and Kapferer (1985). The same holds true for the size of the associations involving the OPII, in comparison to those reported by Zaichkowsky (1985).


Investigators who require a measure of enduring involvement with a product category are encouraged to consider this revised version of the Personal Involvement Inventory (RPII), as a viable alternative to Zaichkowsky's (1985) original version. The high correlation between the revised and the original version gives confidence that the extensive validation work performed by Zaichkowsky supports the revised as well as the original version. At the same time, the sub-scale structure of the RPII allows one to incorporate the perspective of Laurent and Kapferer (1985), and to benefit from their validation efforts. In addition, the RPII appears to offer several practical advantages over Zaichkowsky's PII. Its substantially shorter length (14 vs. 20 items) should yield some cost economies due to reductions in questionnaire length, interview time, and respondent fatigue.





However, a few cautionary notes appear in order. Although the RPII receives support, since it was not compared directly against the original it would be premature to declare it better than the OPII. The extent to which any inventory is "better" than another is dependent upon the specific needs of the researcher. In the current study, Zaichkowsky's origin-l inventory was shortened for two primary reasons: l) the necessity to reduce respondent fatigue when collecting involvement ratings on numerous stimuli from the same individual; and 2) our expectation that many non-college educated respondents might not understand the meaning of specific scale items. This expectation, however, remains an empirical question; future research might compare the suitability of the PII and the RPII on populations with diverse demographic characteristics.

Future research might also be devoted to the development of standardized measures of the consequences of involvement. We found it disconcerting to have to judge the merits of two multi-item scales--the RPII and OPII-using single-item measures of the consequences of involvement. Further refinement of measures of the involvement construct itself may not be possible until better measurement procedures for the antecedents and consequences of involvement are developed.


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

Burt, Ronald S. (1976), "Interpretational Confounding of Unobserved Variables in Structural Equation Models," Sociological Methods and Research, 5, 3-52.

Cohen, Joel B. (1983), "Involvement and You: 1000 Great Ideas," in Advances in Consumer Research, Vol. 10, eds. Richard Bagozzi and Alice Tybout, Ann Arbor, MI: Association for Consumer Research, 325-328.

Houston, Michael J. and Michael L. Rothschild (1978), "Conceptual and Methodological Perspectives on Involvement," in Research Frontiers in Marketing: Dialogues and Directions, ed. Subhash C. Jain, Chicago: American Marketing Association, 184-187.

Hupfer, Nancy and David Gardner (1971), "Differential Involvement with Products and Issues: An Exploratory Study," in Proceedings: Association for Consumer Research, 262-269.

Kapferer, Jean-Noel and Gilles Laurent (1984), "Consumers' Involvement Profile: New Empirical Results," in Advances in Consumer Research, Vol. 12, eds. Elizabeth C. Hirschman and Morris B. Holbrook, Provo, UT: Association for Consumer Research, 290-295.

Kassarjian, Harold H. and Waltraud H. Kassarjian (1979), "Attitudes Under Low Commitment Conditions," in Attitude Research Plays for High Stakes, eds. John C. Maloney and bernard Silverman, Chicago: American Marketing Association. 3-15.

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

Laurent, Gilles and Jean-Noel Kapferer (1985), "Measuring Consumer Involvement Profiles," Journal of Marketing Research, 22 (February), 41-53.

Peters J. Paul and Gilbert A. Churchill (1986), "Relationships Among Research Design Choices and Psychometric Properties of Rating Scales: A Meta-Analysis," Journal of Marketing Research, 23 (February), 1-10.

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

Rothschild, Michael L. (1984), "Perspectives on Involvement: Current Problems and Future Directions," in Advances in Consumer Research, Vol. 11, ed. Thomas Kinnear, Ann Arbor, MI: Association for Consumer Research.

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