A Comparison of Involvement Measures For the Purchase and Consumption of Pre-Recorded Music

ABSTRACT - Five measures of music involvement, based on work with both traditional products, music, and aesthetic activities, were compared for reliability and validity in relation to consumer decision-making variables. These measures tapped the importance of music to the individual, listening and purchase behavior, and the degree to which the music was experienced by respondents. The results showed that the experientially-based measures were superior to more traditional gauges of involvement. The implications of this finding, and its generalizability to other aesthetic activities, are discussed.



Citation:

Marya J. Pucely, Richard Mizerski, and Pamela Perrewe (1988) ,"A Comparison of Involvement Measures For the Purchase and Consumption of Pre-Recorded Music", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 37-42.

Advances in Consumer Research Volume 15, 1988      Pages 37-42

A COMPARISON OF INVOLVEMENT MEASURES FOR THE PURCHASE AND CONSUMPTION OF PRE-RECORDED MUSIC

Marya J. Pucely, Florida State University

Richard Mizerski, Florida State University

Pamela Perrewe, Florida State University

ABSTRACT -

Five measures of music involvement, based on work with both traditional products, music, and aesthetic activities, were compared for reliability and validity in relation to consumer decision-making variables. These measures tapped the importance of music to the individual, listening and purchase behavior, and the degree to which the music was experienced by respondents. The results showed that the experientially-based measures were superior to more traditional gauges of involvement. The implications of this finding, and its generalizability to other aesthetic activities, are discussed.

INTRODUCTION

The concept of involvement has received substantial attention in the social psychology and marketing literatures during the past 40 year_. Consumer involvement clearly has implications relative to the consumer decision-making process and marketing strategies developed based on this process. Empirical findings indicate that level of this construct has an impact on purchasing effort, cognitive and evaluative complexity, attitude-behavior relationships, effects of information on attitude formation, process of advertising evaluation, and recall. For an excellent review of research in this area, see Slama and Tashchian (1985a) and Zaichkowsky (1985). -

Recently, several attempts have been made to define and operationalize this construct with three separate measures of involvement published in the marketing literature during the past two years. These measures are primarily concerned with types of products emphasized in conventional consumer research. Holbrook and Hirschman (1982), however, have hypothesized that some products, such as leisure activities, aesthetic goods, sporting events, etc., may entail a different decision-making process from that suggested by the information-processing paradigm. These differences imply involvement relative to aesthetic goods may manifest itself in a manner that is different from involvement with tangible products and conventional purchasing activities. The purpose of this research is to develop involvement scales that are applicable to aesthetic products such as popular music.

RECENT MEASURES OF CONSUMER INVOLVEMENT

One operationalization of the involvement concept, developed by Slama and Tashchian (1985b), reflects consumers' involvement with purchasing activities in general. Purchasing involvement, defined as "a general measure of the self-relevance of purchasing activities to the individual" (p. 73), is based on Kassarjian's (1981) Consumer Trait Theory which suggests there are individual differences among people which make some more interested, concerned, and involved in the consumer decision process regardless of the specific product or situation. Thus, purchasing involvement deals with interest in shopping itself, which transcends the particular product being purchased or the specific situation.

A different conceptualization of involvement is related to Rothschild's (1979) notion of enduring involvement, which is the degree of interest in a particular product category that the individual brings into a particular situation. Zaichkowsky (1985) developed an involvement measure, the Personal Involvement Inventory (PII), to capture this aspect of involvement. While the PII was designed to measure involvement with a product class, Zaichkowsky presented evidence indicating the scale may be sensitive to different purchase situations and suggested it could be modified to apply to marketing communications.

The third involvement scale, developed by Laurent and Kapferer (1985), also reflects enduring involvement with a product category. This instrument measures four different facets of involvement: (1) perceived importance of the product and perceived importance of negative consequences of a mispurchase; (2) perceived probability of making a mispurchase; (3) hedonic value and/or emotional appeal of the product class; and, (4) perceived symbolic value of the product, its purchase, or its consumption.

INVOLVEMENT WITH AESTHETIC PRODUCTS

Holbrook and Hirschman (1982) have hypothesized that type of involvement (as opposed to level of involvement) may vary depending on the product category being considered. They suggest the information processing perspective, and conventional consumer research, ignores the experiential view which encompasses consumption phenomena such as leisure activities, aesthetics, symbolic meanings, emotions, play and artistic endeavors.

Both the goals of consumption and the decision-making process leading to purchase may vary between traditional and aesthetic products. According to Holbrook and Hirschman's (1982) experiential perspective, consumers may appreciate the product for its own sake apart from any utilitarian functions performed or tangible benefits gained through product use. The consequences of experiential consumption are fun, amusement, fantasy, arousal, sensory stimulation, and enjoyment. Furthermore, the experience of actual consumption may have a greater impact on consumer decision making for many aesthetic products than it does for conventional products. For example, an individual will typically "consume" a song (hearing it on the radio, perhaps) prior to making a purchase and that experience will influence the subsequent purchase decision. Thus, the events surrounding consumption may warrant greater attention when considering experiential products.

While the three involvement scales discussed above measure different aspects of consumer involvement, all implicitly deal with the types of purchasing activities and goods/services emphasized in the currently prevailing information processing perspective of consumer behavior. Zaichkowsky (1985) used instant coffee, laundry detergent, and color television sets to test the discriminating ability of the PII. Laurent and Kapferer (1985) validated the Consumer Involvement Profile using fourteen product types in the categories of household appliances, personal care products, food items, and clothing. The criterion validity of both of these scales was assessed by associating involvement scores for products with variables such as extensiveness of the pre-purchase decision process, scope of information search, and product differentiation within the category. While Slama and Tashchian's (1985) purchasing involvement scale transcends specific products and situations, the actual measure emphasizes traditional pre-purchase factors and intervening variables such as information acquisition, comparison shopping, price orientation, and risk perception.

Since type of involvement may varying depending on type of product (Holbrook and Hirschman 1982), these scales may not be applicable to aesthetic products. Recent research suggests purchase and consumption behavior for some product classes, including the performing arts, may differ substantially from behavior associated with traditional products such as package goods and consumer durables. Hirschman and Holbrook (1982) suggest these products prompt a hedonic response when used by consumers and require measures of the multisensory, fantasy, and emotive aspects of the consumption experience. In addition, measures of the experiences derived from product usage may be more relevant for studying decision-making relating to hedonic consumption situations than the traditional measures utilized. Thus, measures emphasizing emotional and experiential aspects of involvement may be required to adequately address the impact of this construct on consumer responses to hedonic products.

RESEARCH PURPOSE

The purpose of this research was to develop and test involvement measures relating to the purchase and consumption of popular music. Five involvement scales, described in the methodology section of this paper, were developed based on a review of both marketing and music psychology literatures. These involvement measures are then evaluated and compared in terms of the four components of construct validation discussed by Peter (1981): reliability, convergent validity, discriminant validity, and nomological validity.

METHODOLOGY

Music Involvement Measures

[In order to conserve space, the actual scales have not been included in this paper. For a copy of the measures, please contact the authors.]

Enduring Music Involvement. This measure, designed to assess the importance of music to an individual, is based on Rothschild's (1979) enduring involvement concept. Development of this scale began with three focus groups. Participants were undergraduate students who perceived themselves to be more interested in music than their peers. Discussion centered around how group members experience music, e.g., why is it important, what role does it play in your life, what does music do to and/or for you. Video tapes of the focus groups were reviewed to develop an initial battery of scale items. These items were refined and edited resulting in an instrument with 21 Likert-type statements. This scale was then pretested among 65 undergraduate students, and the internal consistency assessed using Cronbach's alpha. As a result, two items were eliminated and a Cronbach's coefficient alpha of .932 was obtained for the remaining 19 items. Since Nunnally (1967) stated coefficients of .80 or greater are sufficient for basic research, the reliability of the enduring music involvement scale was deemed adequate.

Subjects in the main experiment responded to each of these 19 items on a five-point scale anchored by "strongly agree" and "strongly disagree." Responses were summed to calculate an overall enduring music involvement score for each subject. [For all variables measured in this research, individual items were coded/recoded so higher numbers indicate "more of" the quality being measured.]

Behavioral Involvement. The music literature suggests a general measurement approach to establish music involvement that is tied to self-reports of music participation. Dixon (1980) developed a scale for measuring music involvement that gauges concert attendance, hours spent listening to various audio media, and purchases of music products such as pre-recorded tapes and records. This measure reflects a behavioral manifestation of involvement much like that suggested for more traditional products by Engel and Blackwell (1982).

Two facets of behavioral involvement were measured, active and passive. Active involvement was based on subjects' reported attendance at non-dance musical concerts and purchase of music records, prerecorded tapes, and discs. Passive involvement refers to the amount of time subjects indicated they spend listening to music on radio, television, and records/tapes/discs. Two versions of each question were developed varying the time period covered. One form was administered during a pre-experiment survey and the other approximately three weeks later so that test/retest reliabilities could be ascertained.

Experiential Involvement. A more recent perspective deals with experiential aspects of product consumption (Hirschman 1984). This experiential view suggests the objective of some consumption situations may be to stimulate sensory activity, to stimulate cognitive activity, or some combination of both. Two experientially-based involvement measures were generated to capture both facets of this perspective. These scales are song-specific and related to Rothschild's (1979) notion of situation involvement. Scale items, suggested in work by Swanson (1978), were modified based on comments of focus group participants. The first of these measures, called sensory involvement, probes the sensory relationship between the user and a piece of music. The second experiential measure, analytic involvement, assesses the degree to which a song stimulates cognitive activity within the listener.

To assess experiential involvement, subjects responded to seven statements on a seven-point scale anchored by "strongly agree" and "strongly disagree" after listening to each test song.

Additional Evaluative Measures

Distractibility. A three-item scale was developed to ascertain the degree to which subjects are distracted by music while they are performing other tasks such as studying. Ridgeway (1976) suggested that individuals who are highly involved with music maintain a greater degree of absorption in the music during usual listening experiences. Since people have limited processing capacity, it follows that highly involved individuals would devote a greater proportion of this capacity to background music and a smaller proportion to the task which should lead to a higher perception of distraction.

Attention. The amount of attention subjects pay to music was assessed relative to both music in general and to each of two test songs. Dixon (1980) suggested more highly involved individuals pay more attention to music than do less highly involved individuals. In addition, Greenwald and Leavitt (1984) maintain that audience involvement in a communication is related to the degree of attention paid to and the amount of processing capacity allocated to that communication by the individual. Thus, individuals who are more highly involved with music are expected to pay more attention to music compositions.

To determine the amount of attention paid to music in general, respondents were asked, "Of the total time you spend listening to music, what percent is spent 'really listening'?" (Dixon 1980). This question was asked on two occasions--during the pre-experiment survey and after the experiment was completedCin order to determine test/retest reliability of the measure. Subjects also responded to the statement, "I paid a lot of attention to this song," on a seven-point scale anchored by "strongly agree" and "strongly disagree" for each of the two test songs.

Affect/Liking. Affect was measured with Huber and Holbrook's (1979) Index of Global Evaluation. This metric was computed by summing responses to a list of nine bipolar adjectives separated by a six-point scale.

Purchase Intent. Subjects used a seven-point strongly agree/strongly disagree scale to respond to the statement, "I would purchase this song the next time I shop for music."

Subjects

The sample consisted of 110 volunteer undergraduate business students attending a large southeastern university. Subjects were enrolled in an introductory management class and received extra credit for their participation in the study. None of the focus group participants or enduring involvement measure pretest subjects were included in this sample.

Music Selection

The choice of the four songs used in the main experiment was based on the goal of having music that was unfamiliar to the subjects and of good professional quality. After a pool of rock-oriented songs was initially chosen by the experimenters, the songs were pretested on several groups of subjects with the same background as the subjects used in the main experiment. This reduced the total to eight songs that were judged approximately average on a seven-point like/dislike affect scale. This was done to eliminate "floor" and "ceiling" effects on the evaluative measures. None of the songs were correctly identified by these subjects.

Four of the original group of eight songs were randomly chosen to be the "pre-exposed" group. These four were recorded on a tape in a randomly determined order. Each selection was followed by an announcement of the song title and artist in a manner similar to that of a disc jockey. This entire sequence of songs was recorded twice resulting in a tape approximately 28 minutes in length. The two test songs were randomly chosen from the four in the pre-exposure group. An additional set of tapes consisting of these two songs was produced for the evaluation portion of the experiment. The order of the songs was varied on these tapes to eliminate the impact of order effects.

Procedure

Approximately three weeks prior to the main portion of the experiment, the enduring music involvement questionnaire was administered to the subjects during their normal class period by a faculty member not associated with this research. Subjects were then recruited for the main portion of the experiment by one of the authors at the end of class. The subjects were not told the purpose of the experiment nor its connection with the questionnaire they had filled out earlier. This procedure was followed to minimize the association between the music questionnaire and the subsequent experiment.

Four subjects were scheduled for each experimental session. In order to present the music in a relatively realistic environment, subjects were assigned tasks to perform after arriving at the test facility. Individuals in the "pre-exposure" sessions listened to the pre-recorded tape of four songs played twice; individuals in the control sessions worked on the tasks for the same amount of time, but did not listen to the music. After working on the tasks, subjects filled out a questionnaire eliciting information about the task to disguise the fact that the music was the focus of the research. Since neither exposure to the music while performing the task nor the various task variables had an impact on the proposed relationships, these variables were dropped from the subsequent analyses.

After completing the first portion of the study, subjects were taken to another room to complete the experiment. Each of the four subjects was seated in an individual carrel to avoid contamination of the results due to respondent interaction. The experimenter then played the first test song in its entirety. After listening to the song, subjects completed the affect, attention, experiential involvement and purchase intent scales for that song. This procedure was repeated for the second test song. These two songs are referred to as Song 1 and Song 2 in the subsequent analyses. Finally, subjects answered questions eliciting behavioral involvement.

RESULTS

Reliability Of the Involvement Measures

Table 1 present Cronbach's coefficient alphas, item-to-total correlation ranges, and test/retest correlations for the various music involvement measures. These statistics will be used to assess the reliability of the measures.

Internal consistency was estimated using Cronbach's coefficient alpha and item-to-total correlations for each scale. A large coefficient alpha indicates the set of items performs adequately in capturing the construct which motivated the measure, and relatively large correlations of each item to the total score for the measure indicate each item contributes equally to the common core of the construct (Churchill 1979).

It is apparent from Table 1 that the enduring music involvement and experiential involvement measures meet these criteria to a greater extent than do the behavioral involvement measures. Cronbach's coefficient alpha, where applicable, is substantially higher for the non-behavioral involvement measures than for the behavioral scales (.885, .932 and .887 versus .666 and .619). Furthermore, the song-specific measures exhibit more consistent and generally larger item-to-total correlation ranges than do the behavioral measures. While the item-to-total correlations for enduring music involvement are not as high as those for the sensory and analytic involvement scales, a coefficient alpha of .885 is considered acceptable for a scale in its developmental stages (Nunnally 1967).

TABLE 1

MUSIC INVOLVEMENT RELIABILITY SCORES

Test/retest correlations for the behavioral involvement measures, while significant, are relatively small (r=.129, p<.10 and r=.236, p<.01). These low correlations question the stability of the behavioral scales.

Relationships Among Measures

As noted earlier, in addition to demonstrating internal consistency, an appropriate measure of a construct should (1) provide evidence of convergent and discriminant validity and (2) behave as expected relative to other measures (Churchill 1979). Tables 2 through 5 exhibit correlations among the involvement measures and additional evaluative variables.

Table 2 presents correlations among the involvement measures. Correlations between song-specific scales and the enduring and behavioral measures was small and nonsignificant in all cases. This provides evidence that involvement with music may have both enduring and situational components. While enduring music involvement is significantly correlated with the pre-experimental survey active and passive behavioral measures (r=.351 and r=.203, p< 01), it is not correlated with the post-experiment measures (r=.019 and r=.097). In addition, the behavioral measures are more highly correlated within survey instrument than across questionnaires (r=.202 and r=.136 versus r = .036 and r=.046). These results suggest subjects may have been responding to the behavioral questions with a consistency bias.

In terms of song-specific measures, sensory involvement is significantly correlated with analytic involvement within songs, but not across songs. In addition, the correlation between the sensory measures across songs is small and not significant (r=.076) while the analytic measures are highly correlated across songs (r=.456, p<.01). Apparently, experiential involvement has both sensory and analytic aspects, however, the sensory component performs better in discriminating between songs.

TABLE 2

CORRELATIONS AMONG MUSIC INVOLVEMENT MEASURES

It was hypothesized that individuals who are highly involved with music would spend a larger proportion of time "really listening" and would be more easily distracted by music while performing tasks. Table 3 presents correlations between the universal music involvement measures and the non-song-specific attention and distractibility scales. As expected, enduring music involvement is positively correlated with attention (r=.357 and r=.267, p<.01) and negatively correlated with distractibility (r=-.205, p<.01). The relationships between these scales and the behavioral measures are significant (p<.05) in only five of twelve correlations. The significant relationships are in the anticipated direction, however. Overall, these results suggest music involvement is positively related to attention and negatively related to distractibility as expected, and that the enduring music involvement scale is a better measure of the construct than the behavioral involvement scales.

Table 4 presents correlations between the two experiential involvement measures and song-specific attention and affect. Sensory involvement is significantly correlated (p<.01) with attention and affect within a song, but shows small and nonsignificant correlations with these scales across songs. This provides evidence of convergent and discriminant validity for the sensory involvement measure. While analytic involvement is significantly correlated (p<.01) with attention and affect within song, in one instance a significant correlation (p<.05) is exhibited across song measures (analytic involvement Song 1 with attention Song 2). This provides further evidence that the analytic involvement measure lacks discriminating power.

Table 5 presents correlations between the involvement measures and intent to purchase each of the two experimental songs. One would expect an involvement measure to bear a strong relationship to purchase intent. Both of the experiential involvement measures were significantly associated with purchase intent for the corresponding song (p<.01), while the universal involvement measures exhibited significant relationships in only three of ten comparisons. In addition, these three correlations were relatively small (r=.180, r=.211, and r=.239). These findings suggest music involvement has both stimulus-specific and enduring components, and that the song-specific measures may be more appropriate than the general measures in determining consumer responses to new music.

TABLE 3

CORRELATIONS WITH UNIVERSAL MEASURES

TABLE 4

CORRELATIONS WITH PRODUCT-SPECIFIC MEASURES

TABLE 5

CORRELATIONS WITH PURCHASE INTENT

SUMMARY AND CONCLUSIONS

Five music involvement measures were developed based on previous work in music and consumer research concerning aesthetic products. Three of these measures were designed to ascertain the importance of music in an individual's life. These five measures were then experimentally tested to evaluate relative reliability and validity against consumer decision-making variables.

The two behaviorally-based involvement measures were shown to have relatively poor internal reliability and small test/retest correlations. Two factors may account for the instability of these measures. First, the test/retest items were not identical but varied in terms of the time period covered and a potential history effect may have occurred. For example, one format asked for record purchases within the past year while the other requested this information for the past month. Situations such as purchase of a new stereo system, receipt of extra money from parents, sales at record stores, unusually high expenses, etc., could serve to inflate or deflate expenditures on records for a particular month. On the other hand, self-reports based on a year-long period are clearly suspect in terms of accuracy. The second factor is the previously noted potential consistency bias in the responses. The behavioral data recorded in the pre-experiment survey, which also collected enduring music involvement information, was significantly correlated with that measure. In addition, the information collected within a survey instrument was more strongly correlated than the data gathered between instruments. The literature on self-report inaccuracy (e.g., see Nisbett and Wilson 1977) is clearly relevant.

Along with demonstrating greater internal consistency, the enduring music involvement measure performed better than the behavioral measures in terms of. its relationship to other variables. Enduring music involvement was found to be positively associated with the proportion of time an individual spends really listening to music and negatively related to propensity to be distracted by music while doing other tasks. The behavioral measures exhibited inconsistent relationships with these two variables.

The song-specific experientially-based involvement measures had a significant and strong relationship to evaluative and purchase intent measures for the corresponding test songs. However, sensory involvement appears to perform better at discriminating between songs than does analytic involvement. Hirschman (1984) suggests that overall capacity and/or desire to obtain sensory and or cognitive experiences may be genetically determined and these inherent differences may result in determining individual capacity for experiential consumption activities. Future research should assess this phenomenon in relation to music. This study was conducted with a sample of students pursuing degrees in business. Perhaps individuals with other interests, e.g., art or theater students, would respond to these questions differently.

Unlike more traditional products, music involvement appears to be largely an experientially-based phenomenon. This finding may be similar to involvement with other aesthetic activities such movies and sports, although the actual scales may not be applicable. On the other hand, the difference between music and other products may be the result of the relative influence of the consumption experience on consumer decision-making in music. Future research is now underway to better understand the generalizability of these findings.

REFERENCES

Churchill, G. A., Jr. (1979), "A paradigm for developing better measures of marketing constructs," Journal of Marketing Research, 16(Feb), 64-73.

Dixon, R. D. (1980), "Suggested scales for the measurement of musical involvement and genre tastes," Popular Music and Society, 7(4), 223-44.

Engel, J. F. and R. D. Blackwell (1982), Consumer Behavior, CBS Publishing.

Greenwald, A. G. and C. Leavitt (1984), "Audience involvement in advertising: four levels," Journal of Consumer Research, 11(June), 581-92.

Hirschman, E. C. (1984), "Experience seeking: a subjectivist perspective of consumption," Journal of Business Research, 12, 115-36.

Hirschman, E. C. and M. B. Holbrook (1982), "Hedonic consumption: emerging concepts, methods and propositions," Journal of Marketing, 46(Summer), 92-101.

Holbrook, M. B. and E. C. Hirschman (1982), 'The experiential aspects of consumption: consumer fantasies, feelings, and fun," Journal of Consumer Research, 9(Sept), 132-40.

Kassarjian, H. H. (1981), "Low involvement--a second look," in Advances in Consumer Research, Vol. 8, K. B. Monroe, Ed., Ann Arbor Association for Consumer Research, 31-4.

Laurent, G. and J. Kapferer (1985), "Measuring consumer involvement profiles," Journal of Marketing Research, 22(Feb), 41-53.

Nisbett, R. E. and T. D. Wilson (1977), "Telling more than we can know: verbal reports on metal processes," Psychological Review, 84(May), 231-59.

Nunnally, 1. C. (1967), Psychometric Theory, McGraw-Hill Book Company.

Peter, J. P. (1981), "Construct validity: a review of basic issues and marketing practices," Journal of Marketing Research, 28(May), 133-45.

Ridgeway, C. L. (1976), "Affective interaction as a determinant of musical involvement," Sociological Quarterly, 17(Summer), 414-28.

Rothschild, M. L. (1979), "Advertising strategies for high and low involvement situations," in Attitudes Research Plays for High Stakes, John C. Maloney and Bernary Silverman, eds., Chicago: American Marketing Association, 74-93.

Slama, M. E. and A. Tashchian (1985a), "Involvement as a consumer behavior construct a critical review,' unpublished paper, The Florida State University.

Slama, M. E. (1985b), "Selected socioeconomic and demographic characteristics associated with purchasing involvement," Journal of Marketing, 49(Winter), 72-82.

Swanson, G. E. (1978), "Travels through inner space: family structure and openness to absorbing experiences," American Journal of Sociology, 83(4), 890-919.

Zaichkowsky, J. L. (1985), "Measuring the involvement construct," Journal of Consumer Research, 12(Dec), 341 -52.

Zaichkowsky, J. L. (1986), "Conceptualizing involvement," Journal of Advertising, 15(2), 4.

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

Authors

Marya J. Pucely, Florida State University
Richard Mizerski, Florida State University
Pamela Perrewe, Florida State University



Volume

NA - Advances in Consumer Research Volume 15 | 1988



Share Proceeding

Featured papers

See More

Featured

“But, will you think it's important to use mouthwash?” How Visual Communication of a Set Impacts Perceived Set Completeness and Item Importance

Miaolei (Liam) Jia, University of Warwick, United Kingdom
Xiuping Li, National University of Singapore, Singapore
aradhna krishna, University of Michigan, USA

Read More

Featured

Understanding the Framing of Recommendations

Jia Gai, Erasmus University Rotterdam, The Netherlands
Anne-Kathrin Klesse, Erasmus University Rotterdam, The Netherlands

Read More

Featured

F11. Anti-Consumption for Sustainability: The Environmental Impact of Anti-Consumption Lifestyles, Environmentally Concerned Individuals and Ethical Consumers

Laurie Touchette, HEC Montreal, Canada
Marcelo Vinhal Nepomuceno, HEC Montreal, Canada

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.