Health Promotion Services Consumption: Involvement and Program Choice

Nora J. Rifon, Michigan State University
Brian E. Mavis, Michigan State University
Elizabeth Tucker, Michigan State University
Bertram E. St÷ffelmayr, Michigan State University
ABSTRACT - This study examines consumer involvement as a predictor of employee decisions to enroll in worksite health promotion programs. The research extends the application of the involvement construct to the area of the consumption of services, and contributes to the validation of Zaichkowsky's (1985) Personal Involvement Inventory. A short form of the PII is developed specifically for use in adult, non-student populations, and is used to effectively differentiate among employee health promotion program choices.
[ to cite ]:
Nora J. Rifon, Brian E. Mavis, Elizabeth Tucker, and Bertram E. St÷ffelmayr (1992) ,"Health Promotion Services Consumption: Involvement and Program Choice", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 679-687.

Advances in Consumer Research Volume 19, 1992      Pages 679-687


Nora J. Rifon, Michigan State University

Brian E. Mavis, Michigan State University

Elizabeth Tucker, Michigan State University

Bertram E. St÷ffelmayr, Michigan State University


This study examines consumer involvement as a predictor of employee decisions to enroll in worksite health promotion programs. The research extends the application of the involvement construct to the area of the consumption of services, and contributes to the validation of Zaichkowsky's (1985) Personal Involvement Inventory. A short form of the PII is developed specifically for use in adult, non-student populations, and is used to effectively differentiate among employee health promotion program choices.


The conceptualization and scaling of involvement has been of interest to consumer researchers since the concept's introduction to consumer behavior almost three decades ago (Krugman 1965). As the concept and its measurement have evolved, issues related to the dimensionality of the construct and appropriate strategies for the measurement of involvement have been debated in the literature (Higie & Feick, 1989; Jain & Srinivasan, 1990; Laurent & Kapferer, 1985). Nevertheless, there is a consensus among many researchers that the essence of involvement is perceived personal relevance (Celsi & Olson, 1988; Higie & Feick, 1989; Krugman, 1967; Petty & Cacioppo, 1979; Zaichkowsky, 1985).

Zaichkowsky (1985) has developed a measure of involvement based on its definition as perceived personal relevance. The Personal Involvement Inventory (PII) was designed to be generalizable across people, objects and situations. The research reported here illustrates and further validates the use of Zaichkowsky's (1985) PII for the measurement of an individual's involvement with a service category in a non-traditional consumer situation. In addition, this paper describes the development of a shortened version of the PII for use in adult, non-student samples. Further, as researchers and providers in the area of worksite health promotion, we examined involvement's capacity to predict an individual's participation behavior in one of four worksite health promotion programs (smoking cessation, weight/nutrition management, stress management and exercise/fitness).


Zaichkowsky (1985) addressed the conceptualization and measurement of involvement, independent of its behavioral consequences. She defined involvement as "a person's perceived relevance of the object based on inherent needs, values and interests" (p. 342). Celsi and Olson (1988) further explicate involvement as essentially perceived personal relevance. Perceived relevance is a function of a consumer's perceptions of his/her needs, goals and values and their congruence with the consumer's knowledge of the product category. "More specifically, the personal relevance of a product is represented by the perceived linkage between an individual's needs, goals and values (self-knowledge) and their product knowledge (attributes and benefits)" (p. 211). The greater the perceived linkage, the stronger the feelings of personal relevance of the product category.

It has been suggested that an individual's level of involvement affects product-related information searching and decision-making (Engel, Kollat & Blackwell, 1978; Bettman, 1979), as well as his/her processing of persuasive communications and resultant attitude change (Petty & Cacioppo, 1986).

Using involvement to explore employee relationships with worksite health promotion programs will advance the study of employee participation in several ways. It will provide practical information for program providers as well as basic information for theory development. A primary goal of health promoters is to increase participation in worksite health promotion programs. Studying involvement will increase the health promoter's understanding of the factors influencing the decision-making process of those individuals choosing to participate or not participate in the programs. Ultimately, involvement could be used as a segmentation variable, particularly for the development of appropriate promotional messages for the varied target audiences.

In addition, using involvement to profile the target audience's perceived relevance of health promotion programs would add a valuable consumer dimension to the study of participation in health promotion activities. Research on participation in worksite health promotion programs has provided little insight regarding an individual's decision to participate in programs, and has been dominated by a medical perspective for profiling and targeting audiences.

In a review of the literature on worksite health promotion programs, Conrad (1987) concluded that "participants seem to be a self-selected, somewhat healthier group of employees" (p.318). A more recent, but unpublished review of the participation literature suggests that there are few variables consistently found to be related to participation. From among 40 variables reported in 21 studies, only self-efficacy, smoking status and education level consistently distinguished between participants and non-participants (Mavis, Stachnik, Gibson & St÷ffelmayr, unpublished manuscript). The self-selection process has been troublesome to health promoters, since individuals who participate are not the individuals who exhibit the most health-related need.

This research investigates an individual's perceived relationship with worksite health promotion services by measuring the individual's involvement with the specific programs offered. In contrast to the traditional approach of health risk profiling, we hypothesize that involvement levels with the health promotion programs will differentiate audiences for a variety of programs.

Measuring Involvement

Involvement was defined as an individual's "perceived relevance of the object based on inherent needs, values, and interests" (Zaichkowsky 1985, p. 342). The scale used to measure involvement was Zaichkowsky's (1985) 20-item semantic differential Personal Involvement Inventory (PII), which has been shown to have reliability and validity.

Unlike previous studies which have used the PII to study involvement using primarily student-based samples (Celsi & Olson, 1988; Jensen, Carlson & Tripp, 1989; McQuarrie & Munson, 1987; Zaichkowsky, 1985; 1987), this research examined involvement levels in one adult sample of employees, and one adult sample with some graduate students. Since we were primarily interested in the adult sample, we were concerned about the appropriateness of the PII when used with groups other than students. We decided to systematically modify the scale for future use in an adult, non-student population.

These concerns appeared to be well-founded. McQuarrie and Munson (1987) commented on the syllable count of some of the words used in the scale, noting "in general, the syllable count for the PII seems uncomfortably high" (p. 37). In addition, when the present researchers pretested the scale, feedback from participants indicated that the length of the scale would prohibit its usage more than once in a sitting. In fact, some respondents reported that one presentation of the scale was "annoying". We thought it necessary to shorten the PII by eliminating problematic items, while maintaining the scale's integrity.

Research Objectives and Contributions

The initial objective of this research was to examine the utility of involvement for the study of employee participation in worksite health promotion programs. However, given the previously mentioned considerations, the development of a short version of the PII for use in non-student samples became another objective of the research.

The results reported in this paper focus on the measurement of involvement. The study tests the factor structure of Zaichkowsky's (1985) 20-item scale with an adult, non-student sample, in response to a service rather than a product. In addition, the short version of the scale was developed based on the data from one sample of subjects and then cross-validated in another sample to assess its criterion-related validity against a behavioral outcome measure obtained two months after the scale was administered.


The rationale for the modification of the PII was based on meeting the following objectives:

1. the development of a 10-item version of the PII to minimize respondent annoyance and fatigue when the scale is used in a repeated measure situation;

2. the identification and elimination of scale items which may prove problematic to some subjects as a function of their level of education;

3. the minimization of redundant items (McQuarrie & Munson, 1987); and

4. the maintenance of the factor structure established by Zaichkowsky (1985).

Problematic items were identified using missing value data and reading levels associated with the words used in the scale. One sample of subjects was used to generate missing value data and the factor structure of the full PII scale. A second sample was used to evaluate the shortened version.

The Health Promotion Services Package

The services subjects rated were a package of health promotion programs available through the university. Four behavioral modification programs, stress management, exercise/ fitness, weight control and smoking cessation were available to units requesting the service. Employees could participate in programs free of charge, however, the unit administrators incurred fees to cover the costs of programs. The programs consisted of supervised, group, weekly meetings and continued over the course of fifteen weeks.

Missing Value Data: Sample I

All of the employees with offices located in the administration building of a major midwestern university were used as subjects in this part of the study. Since the health promotion programs were offered to the employees of an entire building or "worksite" at one time, the design required that all employees within the building be eligible for participation in the research. At the time of this study, the administration building had been targeted for the next series of health promotion activities.

As a part of a larger research project, the 20-item involvement scale was included in a survey distributed to the sample via intracampus mail. The survey was a pre-intervention measure, used to obtain baseline measures of health-related psychographic variables, as well as demographic profiles. A total of 287 useable surveys were returned; this represents a 57.3% response rate. The characteristics of the respondents are listed in Table 1.

Each participant was asked to rate one of four health promotion programs using the 20-item involvement scale. The four programs were weight management, exercise, smoking cessation, and stress management. Pre-testing the survey led us to believe that asking subjects to rate more than one program would severely reduce our response rate, and possibly create respondent fatigue, boredom and response bias.





Respondents were categorized into three employee groups: faculty, supervisors, and clerical-technical support staff. The frequency of missing values for each of the 20 items were tabulated separately for each employee group. Table 2 lists the items in ranked order based on the total missing values across all three employee groups. The number next to each word pair indicates the order of presentation of the word pair within the scale. Figure 1 graphically depicts the pattern of missing values for each employee group and for the total sample.

Certain word pairs were unanswered more often than others; faculty had the lowest rate of missing data across all items compared to the two other employee groups. The pattern of missing data suggests that the education level of the respondent was related to successful completion of the Personal Involvement Inventory.



Another possible contributing factor is subject fatigue. The data in Figure 1 do not suggest any overall fatigue in the response pattern, although a possible interaction with level of education is evident. Little or no evidence of a fatigue effect, as indicated by an increased frequency of missing was found for faculty. The frequency of missed items tended to increase among the responses for supervisory staff, whereas the clerical-technical staff patterns show the clearest evidence of possible fatigue.

Grade Level for Personal Involvement Items

The American Heritage Word Frequency Book reports the results of a survey of over 2,000 public, private and parochial schools, 10,000 texts and 5 million words, for the frequency of word usage by grade level. The information provided in the report indicates the modal grade level at which the words are presented to students. In order to assess the reading level of the involvement scale, the grade assignments provided by the word frequency book were determined for the word pairs comprising the scale.

Table 3 displays the grade levels associated with the words in each word pair for the 20-item involvement scale. The word pairs are listed in the same order as in Table 2, representing their ranking based on missing values. The average grade level associated with the scale is 6.13, the median grade level is 7, and the grade level ranges from second to twelfth grade.

Factor Analysis

The data used in the factor analysis included all 287 subjects' responding to the four different health promotion programs. The results of the factor analysis for the 20 items comprising the scale, using varimax rotation, are reported in Table 4.

The first factor explained 63.6% of the total variance; the amount of variance explained by the second factor was 7.0%. All of the items loaded positively on the first factor. The residual factor consisted of four items with higher loadings on the second than the first factor. The word pairs comprising the residual factor were boring/interesting, unexciting/exciting, mundane/fascinating, and appealing/unappealing. The two-factor solution was comparable to that reported by Zaichkowsky (1987), with the same items loading on the residual factor. The data validate the scale's factor structure, even when used to measure involvement with a service category not traditionally considered in consumer behavior.


A rational-empirical approach to scaling (Jackson 1970) was used to combine the information on the average reading levels of the word pairs, their missing value frequencies, and the factor analysis to select 10 items for deletion from the scale. Of the ten items with the highest missing value frequencies, eight were dropped. This effectively reduced most of the redundancy in the scale, and eliminated items with high grade levels such as vital/superfluous and mundane/ fascinating. Of the items with the lowest missing value frequencies, two were dropped. The six items with highest missing values were eliminated. Of those six, three belonged to the second factor; consequently, appealing/unappealing (the remaining item on the second factor) was retained.



With six items eliminated, four items from the first factor were then identified for elimination. Congruent with McQuarrie and Munson's (1987) concerns that four of the items "seemed inappropriate for use with a non-college educated population", trivial/fundamental and significant/ insignificant were eliminated. Mundane/fascinating and vital/ superfluous had been already eliminated based on their overall missing values and reading levels. Significant/insignificant exhibited relatively high missing values and reading levels. Despite the fact that the overall missing value for trivial/fundamental was not in the top ten rank, it was more highly ranked for the occupational group of staff people, who were likely to have lower levels of education than the other two occupational groups.

Using both the missing values and reading levels, wanted/unwanted and beneficial/not beneficial were the last two items eliminated. Wanted/unwanted had the highest missing values of the items remaining (not including essential/nonessential) and beneficial/not beneficial had the highest reading levels. (Essential/nonessential had been retained since other items with similar meanings, such as not needed/needed, and trivial/fundamental, had already been eliminated.)

The final shortened version retains one item from the residual factor (appealing/unappealing) and nine items from the main factor. The items in the shortened version of the PII are uninterested/ interested, essential/nonessential, appealing/ unappealing, matters/doesn't matter, irrelevant/ relevant, no concern/of concern, useless/useful, means a lot/means nothing, important/unimportant, and valuable/invaluable.

The reading level information on the original and short forms of the scale are presented in Table 5. The average reading level of the shorter version is less than that of the original; this change is not statistically significant (t(58) = .502). However, the highest grade level associated with the scale has been reduced from twelfth grade to ninth grade.

The data from the original sample were rescored based on the new 10-item short form of the scale. The correlation between the original and short forms of the scale was .98. The internal consistency estimates for each version of the scale, using Cronbach's alpha, were comparable. Alpha was .968 for the full scale and .946 for the short scale. A comparison of the distribution of the scale scores for the administration building sample are provided in Table 6.

Sample II

The short version of the PII was administered to a second sample of subjects. The 268 subjects were faculty, staff and graduate students in two academic campus buildings where health promotion programs were being offered. Subjects were asked to provide ratings of personal involvement for each of three programs: weight management, stress management and exercise programs. The demographic profile of the this second sample is displayed in Table 7. In contrast to the characteristics of Sample I, Sample II consisted of more males than females, was somewhat younger, and contained graduate students.

Internal Consistency

Since subjects' responded on the short form of the PII for all three health promotion programs, Cronbach's alpha was calculated across subjects' responding for each program. The scale exhibited high internal consistency for all three programs; the short form PII alphas were .946 for the weight management program, .945 for the exercise program and .952 for the stress management program.









Criterion-Related Validity

Six weeks after the short form PII scores were collected, program enrollment began. One might hypothesize that individuals who viewed health promotion programs as more personally relevant would be more likely to use the programs than individuals who viewed the programs as less personally relevant. In fact, individuals viewing a specific program, such as weight management, as more personally relevant than the other programs, would be more likely to use the weight management program rather than the stress or exercise programs.

Based on these speculations the following hypotheses were developed:

1. Program participants will have scored higher on the short PII in response to all programs as compared to non-participants.

2. Participants in the weight management program will have scored highest on the short PII in response to the weight management program as compared to their scores in response to the exercise and the stress programs.

3. Participants in the stress management program will have scored highest on the short PII in response to the stress management program as compared to the exercise and weight programs.

4. Participants in the exercise program will have scored highest on the short PII in response to the exercise program as compared to the weight and stress programs.

To test these hypotheses, the short form PII scores of program participants and non-participants were compared using a four group ANOVA with a repeated measure. A significant main effect of program enrollment (see Tables 8 and 9) confirms the first hypothesis; program participants exhibited higher short form PII scores than non-participants. Hypotheses 2, 3 and 4 were confirmed by the pattern of short form PII scores and the significant interaction (see Tables 7 and 8).

Involvement and Program Choice

Other research on worksite health promotion program participation has indicated that self-efficacy, smoking status and education are related to participation. The findings on the relationship between smoking status and participation are somewhat tautological: smokers join programs on smoking cessation and non-smokers do not. Additional analyses were performed to determine the potentially confounding effect of education on program choice (self-efficacy was not measured in this study).

Education was not directly measured in this study so job level was used as a surrogate measure for education. A three group ANOVA (Faculty, Staff, Graduate Student) with a repeated measure (PII scores for the weight, stress, and exercise programs) produced no significant main effect of job level (F(2,237)=1.17, p=.311).


The data presented in this paper support the generalizability of Zaichkowsky's PII across situations and objects. The full form PII exhibited the same factor structure as reported by Zaichkowsky in 1985. However, the data also suggest that the PII may not be as generalizable across people. The education level of the subjects appears to be an important factor in assessing the applicability of the PII for measuring involvement in adult, non-student populations. The subjects in our study represented a broader range of adults than usually seen as respondents in studies on involvement; the supervisory and the clerical-technical staff exhibited a greater number of item omissions on the full form PII than the faculty in our sample. Indeed, the supervisory and clerical-technical staff exhibited a tendency towards response fatigue which may have indicated a greater degree of difficulty with the PII than experienced by faculty.





The short form PII provides an alternative measurement tool when education levels are an issue in the sample being studied. In addition, when more than one administration of the PII is desirable, the short form PII would lessen the likelihood of respondent fatigue. When involvement is measured in applied settings, rather than the laboratory, scale length is crucial particularly when it is included as one of a series of measures. In contrast to Zaichowsky's claim that a reduction in scale length would not affect ease of administration (1985, note 1), the results of this research suggest that scale length is important.

The results of this study provide a new direction for the study of health promotion. The relationship between the short form PII scores and enrollment behavior was especially encouraging, and the six week time delay between the collection of the PII scores and subjects' enrollment choices strongly supports the observed relationship. This finding may be useful to worksite health promoters for the planning and implementation of communication strategies. Individuals in this study enrolled in programs which appeared personally relevant to them. The traditional medical model approach to health promotion program enrollment ignores the perceived needs and desires of the employees and uses risk factors to assess the individual's need for a particular program.

Additional issues of scale validation arise regarding the generalizability of the finding that involvement is related to program choices. The results may be generalizable to other worksite health promotion situations, however, how do they relate to other types of brand choices? The specific worksite health promotion program offerings studied in this research have been viewed by the researchers as varieties of a brand. For some time the researchers have become acutely aware of the need to establish "corporate" image, brand name recognition and brand identity on campus. Marketing and communication efforts have been incorporated into the structure of the overall health promotion plan; even a logo has been developed to enhance image and recognition. Nonetheless, worksite health promotion services constitute a wide product category and the domains of brands versus product categories remains unclear.

For instance, an alternative perspective is to view different programs as representative of different product categories. A weight management program can be viewed as a product category within which several brands are offered, including private labels such as Weight Watchers, and public sector brands, such as the one in this study.

The three programs offered in this study are all based on the same behavior modification theory and are structured in similar ways. In fact, one person with a need to lose weight could satisfy that need in an exercise program and one person with a need to reduce stress could do so in the exercise programs, as well. The different programs may satisfy the same need, but cater to different tastes for behavioral change. It is difficult to know how to classify the programs from a brand or category perspective, however, consumer perceptions may provide the necessary information for clarifying this point. Clearly, additional research is need to further validate the use of the shortened version of the PII.

The area of worksite health promotion is increasing in its importance and individuals are coming in contact with such services on a more regular basis. Indeed, the commercialization of health promotion has dramatically grown in the past decade as evidenced by the increased number of advertisements promoting home exercise equipment, weight control programs, and diet/nutritional supplements. As the focus of health shifts from an illness perspective to a preventative and health maintenance perspective, these issues will become a more important part of the consumer's daily living and health services decision-making activities.


Bettman, J. R. (1979), An Information Processing Theory of Consumer Choice, Chicago: Addison-Wesley Publishing Co.

Celsi, R. L. and J.C. Olson (1988), "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15, 210-224.

Carroll, J., P. Davies and B. Richman (1971), The American Heritage Word Frequency Book, Boston: Houghton-Mifflin Publishing.

Conrad, P. (1987), "Who Comes to Worksite Wellness Programs? A Preliminary Review," Journal of Occupational Medicine, 29 (4), 317-320.

Engel, H. F., D. Kollat and R.D. Blackwell (1979), Consumer Behavior (4th ed.) Chicago: The Dryden Press.

Higie, R. A. and F.F. Feick (1989), "Enduring Involvement: Conceptual and Measurement Issues," Advances in Consumer Research, 16, 690-695.

Jackson, D.N. (1970), "A Sequential Strategy for Personality Scale Development," in Current Topics in Clinical and Community Psychology Vol. 2, ed. C. Spielberger, New York: Academic Press, 62-96.

Jain, K. and N. Srinivasan (1990), "An Empirical Assessment of Multiple Operationalizations of Involvement," Advances in Consumer Research, 17, 594-602.

Jensen, T. D., L. Carlson, and C. Tripp (1989), "The Dimensionality of Involvement: An Empirical Test," Advances in Consumer Research, 16, 680-686.

Krugman, H. E. (1965), "The Impact of Television Advertising: Learning Without Involvement," Public Opinion Quarterly, 29 (Fall), 349-356.

Krugman, H.E. (1967), "The Measurement of Advertising Involvement," Public Opinion Quarterly 30 (Winter), 583-596.

Laurent, G. R.and J. Kapferer (1985), "Measuring Consumer Involvement Profiles," Journal of Marketing Research, 17, 41-53.

Mavis, B.E., T.J. Stachnik, C.A. Gibson, and B.E. Stoffelmayr (unpublished), "Characteristics of Participants and Non-Participants in Worksite Health Promotion".

McQuarrie, E. F. and J.M. Munson (1987), "The Zaichkowsky Personal Involvement Inventory: Modification and Extension," Advances in Consumer Research, 14, 36-42.

Mitchell, A. A. (1981) "The Dimensions of Advertising Involvement," Advances in Consumer Research, 8, 25-29.

Petty, R. E. and J.T. Cacioppo (1986), "The Elaboration Likelihood Model of Persuasion," Advances in Experimental Social Psychology, 19, 125-205.

Petty, R. E. and J.T. Cacioppo (1979), "Issue Involvement Can Increase or Decrease Message Relevant Cognitive Responses," Journal of Personality and Social Psychology, 37 (Oct), 1915-1926.

Zaichkowsky, J. L. (1987), "Familiarity: Product Use, Involvement or Expertise?," Advances in Consumer Research, 14, 296-299.

Zaichkowsky, J. L. (1985), "Measuring the Involvement Construct," Journal of Consumer Research, 12, 341-354.