You are on page 1of 12
Reference Group Influence on Product and Brand Purchase Decisions WILLIAM O. BEARDEN MICHAEL J. ETZEL* Consumer perceptions of reference group influence on product and brand deci sions were examined using 645 members of a consumer panel and 151 respon: denis in & followup study. Differences for 16 products in informational, value ex: pressive, and util n influence wer investigated in a nested repeated ‘measures design. The resuits support hypothesized citferences in reference group influence between publicly and privately consumed products and luxuries and necessities, ‘or some time, social scientists have recognized group membership as a determinant of behavior. The fact that people act in accordance with a frame of reference produced by the groups to which they belong is a long-accepted and sound premise (Merton and Rossi 1949). However, even casual observation revealed perplexing contradictions be- tween group membership and behavior. Many individuals simply did not behave like the majority of people in their recognized groups (e.g., social class or educational level). ‘Though much of this uncharacteristic behavior was explain- able by constructs other than group membership, the ap- parent instability of group influence created confusion. A Partial solution was found in the concept of “reference groups,"’ which recognizes that people frequently orient themaselves to other than membership groups in shaping their behavior and evaluations and that reference groups can; perform a diversity of functions (Merton and Rossi 1949). Marketers have generally accepted the reference ‘group construct as important in at least some types of con- sumer decision making. The present study investigated ref- ‘erence group influence on product and brand purchase de- cisions by examining the interrelationships among two forms of product use conspicuousness (Bourne 1957) and three types of reference group influence (Park and Lessig 197). REFERENCE GROUP CONSTRUCT ‘The opetationalization of reference groups is actually rel- atively recent. Hyman (1942) coined the term in a study of “Willian O. Bearden is Associate Professor of Marketing, College of Business Administration, University of South Carling, Columbia, SC 29208, and Michael J. Etzel is Professor of Marketing and Chairman, College of Busines, University of Notre Dame, Note Dame, IN 46556. ‘Theielpful suggestions of tree anonymous reviewers are pratefully ac Knowledged. social status when he asked respondents with which indi- Viduals or groups they compared themselves. This initial characterization was followed by additional research (New- comb 1943; Sherif 1948) and numerous refinements (Camp- bell et al. 1960; French and Raven 1959; Merton 1957; ‘Sherif and Sherif 1964; Shibutani 1955; Smith, Bruner, and White 1956; Tumer 1955) that clarified and expanded the meaning of the concept. For example, Kelley (1947) dis- tinguished between reference groups used as standards of comparison for self-appraisal (comparative) and those used as a source of personal norms, attitudes, and values (nor- mative). This developing body of literature provided a basis for a series of applications undertaken in a number of fields. For example, Hyman and Singer (1968, p. 7) note that the concept has been applied in studies of farmers, scientists, alcoholics, newspaper people, the mentally ill, consumers, voters, juvenile delinquents, and opinion leaders. To that list could be added steel distributors (Kreisberg 1955). phy- sicians (Coleman, Katz, and Menze! 1966), auto owners (Grubb and Stem 1971), cosmetic users (Moschis 1976), and students and housewives (Park and Lessig 1977). In addition to investigating the presence of reference ‘group influence within identifiable groups, there have been a series of consumer research studies into specific aspects of reference group influence. Venkatesan (1966) attempted to test the differential effects of compliance (Kelman 1961) and reactance (Brehm 1966) in an expe subjects selecting the “best” from among under different forms of group pressure. He was able to establish the influence of voiced group sentiment on a sub- Ject (compliance); however, the method used in operation alizing reactance has been questioned (Clee and Wicklund 1980). Burnkrant and Cousineau (1975) demonstrated that People use others’ product evaluations as a source of in- formation about products. In study of consumer brand choice, Witt (1969) con- (© JOURNAL OF CONSUMER RESEARCH # Vol. 9 @ Sepember 1982 183 Copyright © 2001 All Rights Reserved 184 firmed earlier nonmarketing studies which indicated that group cohesiveness influences behavior. The far-teaching influence of groups was suggested by Cocanongher and Bruce (1971), who found that socially distant reference groups can influence consumers if consumers hold favor- able attitudes toward the members or activities of that group. Taking a somewhat different approach, Witt and Bruce (1970) found the operation of group influence related to the extent of social involvement associated with the prod- uct being investigated. Stafford (1966) found individual brand choice was affected by group influence. ‘The concept of group influence in consumer research has ‘been further refined through studies of various aspects of the social influence process. Witt and Bruce (1972) sug- gested the existence of at least seven different determinants of influence including perceived risk, expertise of the re- ferent, and the individual's need for social approval. Mos- chis (1976) found that consumers use both reflective and comparative appraisal (Jones and Gerard 1967) in product choices. That is, they engage in direct, verbal interaction to determine the reference group's evaluation as well as ‘observing the behavior of reference group members in re- gard to the decision under consideration. Park and Lessig (1977) investigated reference group influence and found students more susceptible than housewives to group influ- ‘ence for a variety of products. ‘The construct is commonly used by marketing pract tioners. Reference group concepts have been used by ad- vertisers in their efforts to persuade consumers to purchase products and brands. Portraying products being consumed in socially pleasant situations, the use of prominentiattrac- tive people endorsing products, and the use of obvious group members as spokespersons in advertisements (Kotler 1980) are all evidence that marketers and advertisers make substantial use of potential reference group influence on consumer behavior in the development of their communi- cations. Alluding to reference groups in persuasive atempts to market products and brands demonstrates the belief that reference groups expose people to behavior and lifestyles. influence self-concept development, contribute to the for- ‘mation of values and attitudes, and generate pressure for conformity to group norms. RESEARCH HYPOTHESES A reference group is a person or group of people that significantly influences an individual's behavior. Within this general framework, several types of influence have been identified. Based on the work of Deutsch and Gerard (1955) and Kelman (1961), information, utilitarian, and vvalue-expressive influences have been identified (Park and Lessig 1977). Informational influence is based on the desire to make informed decisions. Faced with uncertainty, an individual will seek information. From the many sources available, the most likely to be accepted are those viewed as credible. Referents with high credibility include those with presumed expertise or significant others. Utilitarian reference group influence is reflected in at- irae ‘THE JOURNAL OF CONSUMER RESEARCH tempts to comply with the wishes of others to achieve re- ‘wards or avoid punishments. If an individual feels that cer- tain types of behavior will result in rewards or punishments from others and these outcomes are viewed as important, hhe or she will find it useful to meet the expectations of these significant others. ‘A third type of influence, value-expressive, is character- ized by the need for psychological association with a person for group and is reflected in the acceptance of positions expressed by others. This association can take two forms. ‘One form is an attempt to resemble or be like the reference ‘group. The second type of value-expressive influence flows from an attachment or liking for the group. The individual is responsive to the reference group out of a feeling for it, not because of a desire to be associated with it. The occurrence of all ofthese forms of influence requires the opportunity for social interaction or public scrutiny of behavior. Seeking information, complying with the pref erence of others, and adopting values of others all involve some form of communication or observation of decisions, opinions, or behavior. In a purchase context, this implies ‘products that will be seen by others. Besides the opportunity for observation, it is important to consider what elements Of an item will be noticeable. Certainly one is the product, itself, and another is the brand of the product. This ap- s presented in what is probably the most recognized ‘marketing discussion of reference group influence (Bourne 1957), in which the distinction between product and brand decisions is stressed. Bourne (1957, p. 218) originally proposed that reference ‘group influence on product and brand decisions isa function ‘of two forms of “‘conspicuousness."* The first condition, affecting product decisions, is that the item must be ““ex- clusive™ in some way. No matter how visible a product is, if virtually everyone owas it, it is not conspicuous in this sense. This is operationalized here as the distinction be- tween luxuries and necessities. By definition, necessities are possessed by virtually everyone, while luxuries have a degree of exclusivity. Second, for reference group influence to affect brand decisions, the item must be “'seen or iden- tified by others."” This can be operationalized in terms of where an item is consumed. Publicly consumed products are seen by others, while privately consumed products are riot. That is, those brand decisions involving products, can be noticed and identified are more susceptible to reference group influence. ‘Combining the concepts of public-private consumption and luxury-necessity items produces the following four conditions: (1) publicly consumed luxuries, (2) publicly, ‘consumed necessities, (3) privately consumed luxuries, and (4) privately consumed necessities. When applied to prod- uct and brand decisions, these conditions create a total of eight relationships that are the basis underlying this study: 1. Publicly consumed luxury (PUL}—a product consumed in public view and not commonly owned or used (€.8., {golf clubs). In this ease, whether or not the product is med and also what brand is purchased is likely 10 be influenced by others (Bourne 1957, p. 219) Mghts Moservec REFERENCE GROUP INFLUENCE 8s EXHIBIT A ‘COMBINING PUBLIC-PRIVATE AND LUXURY-NECESSITY DIMENSIONS WITH PRODUCT AND BRAND PURCHASE DECISIONS Publ Product ‘Weak reference Strong releronce Brand up influence (—) | group inttuence (+) Public necessiies | Public luxuries ‘Strong reference Influence: Weak Influence: Strong ‘group influence product and ‘product and Gy strong brand brand Examples: Examples: Golf Wristwatch, ‘clubs, snow skis, automobile, sailboat man's suit Necessity tumury Private necessives | Private lxuries ‘Weak reference Influence: Weak lnfivence: Strong ‘group influence product and product and ) brand ‘Weak brand Examples: Examples: TV Mattress, ‘game, trash floor lamp, ‘compactor, tetigerator icemaker Private Relationships with reference group influence: 22, Because itis a luxury, influence forthe product should be sirone. 'b, Because it will be seen by others, influence for the brand of the product should be strong. 2. Privately consumed lurury (PRL}—a product consumed ‘out of public view and not commonly owned or used (e.g, rash compactor). In many cases, the brand is not conspicuous or socially important and is a matter of in- dividual choice, but ownership of the product does convey ‘a message about the owner (Bourne 1957, p. 220) Relationships with reference group influence: a, Because it is a luxury, influence for the product should be strong. bb, Because it will not be seen by others, influence for the ‘brand of the product should be weak. 3. Publicly consumed necessity (PUN)—a product consumed i public view that virwally everyone owns (e.g... wiist- ‘Walch). This group is made up of products that essentially all_people or a high proportion of people use, although differing as to type of brand (Boume 1957, p. 220). Relationships with reference group influence: 8, Because it is a necessity. influence for the product should be weak. bb, Because it will be seen by others, influence for the ‘brand of the product should be strong. 4. Privately consumed necessity (PRN}—a product con- sumed out of public view that virally everyone owns (e-g., mattress). Purchasing behavior is largely governed by product atibutes rather than by the influences of oth- 75. In this group. neither products nor brands tend 10 be Socially conspicuous and are owned by nearly all con- suimers (Boure 1957, p. 221), Relationships to reference group influence: a. Because it is a necessity, influence for the product should be weak: '. Because it will not be seen by others influence for the brand of the product should be weal. ‘These relationships are summarized in a modification of the Bourne framework in Exhibit A and presented in the form of the following 12 hypotheses for testing. The six product decision hypotheses (H,,—Hg,) reflect the proposi- tion that a Juxury-necessity main effect exists, while no such effect is postulated for public-private influences. Con- sequently, for product decision comparisons, reference {group influence is hypothesized to adhere to the following, pattern: VinAVay z B In contrast, forthe six brand decision hypotheses, a sig ificant public-private main effect is hypothesized, while no such effect is postulated forthe Iuxury-necessity factor. Reference group influence is expected then to follow the patter embodied in these remaining six hypotheses: PUL = PUN PUL > PRL PUL > PRN PUN > PRL PUN > PRN PRL = PRN Copyright © 2001 All Rights Reserved METHOD Preliminary Procedures To select products for testing, @ preiiminary list of 81 products was developed. This list consisted of approxi- mately 20 items that the authors felt fell into cach of the four conditions described previously. A questionnaire was constructed to assess perceptions of the individual products as publicly or privately consumed and as luxuries or ne- cessities. ‘A convenience sample of nonstudent adults participated in this preliminary phase. The questionnaire was adminis- tered door-to-door in a middle-class residential area of Co- Jlumbia, South Carolina in the spring of 1979. Adults agree- ing to cooperate were given a copy of the survey, and completed responses were picked up the following day. ‘One hundred households were initially contacted and a total of 57 useable responses were collected. Nonrespondents included those who did not complete the questionnaires and people not at home when the retum visit was made. De- ‘mographic characteristics of the sample were: median ed- tucation, 15 years; median age category, 31 to 35 years: and ‘median family income category, $25,000 to $30,000. Respondents were requested to indicate their perceptions about whether the 81 products were, frst, luxury or neces- sity items, and then, publicly or privately used, on a series, of six-point scales. ‘The instructions to the instrument de- scribed luxuries as not needed for ordinary, day-to-day liv- ing. Necessities were described as being necessary for ot- inary, day-to-day living. Response categories were labeled and scored as follows: (1) a luxury for everyone, luxury for almost all people, (3) a luxury for the majority of people, (4) a necessity for the majority of people. (5) a necessity for almost all people, and (6) necessity for everyone. ‘The same 81 items were assessed by the respondents as being publicly or privately consumed. The fotlowing defi nitions were provided: © A public product is one that other people are aware you possess and use. If they want 10, others can identify the brand of the produet with little of no difficulty. © A private product is one used at home or in private at some location, Except for your immediate family. people ‘would be unaware that you own or use the product. ‘The six-item scales were labeled: (1) a public product for everyone, (2) a public product for almost all people, (3) a public product for the majority of people, (4) a private product for the majority of people, (5) a private product for almost all people, and (6) a private product for everyone. These two dimensions, i.e., public-private and luxury-necessity, represent the underlying determinants of conspicuousness, which Boume assumed to be “the most general attribute bearing on a product's susceptibility to reference group influence’” (1957, p. 218). First. the prod- ‘uct must be conspicuous in the sense that it can be seen and identified (it is, consequently, presumed to be subject to disapproval). Second, the product must be conspicuous in ‘THE JOURNAL OF CONSUMER RESEARCH TABLET PRETEST MEAN SCORES FOR SIXTEEN PRODUCTS EXAMINED" ‘Seale comparison (valve) Public (1}~ Luxury (1)— Product catgory Private (6) Necessity (6) Public cry (PUL) ‘Golf clubs 274 1.86 ‘Sailboat 261 146 Show skis 265 17 ‘Tennis racket 268 228 Public necessity (PUN) Man's suit 205 439 Woman's dress 283 486 ‘Automobile 164 468 Wristwatch 265 396 Private kxury (PRL) Pool table 412 457 “Trash compactor 429 474 ‘Automatic losmaker 428 173 ‘Videogame 421 ast Private necessity (PRN) Retnigerator 4.02 5.35 Blanket 458 551 Matvess 433 519 Lame 405 519 ‘rte ete bad on commnencn sample of ST ronsudht ass Sexes war ‘boar amend pupae ar ay-neesSy Soares ange tom 16 the sense of exclusivity because it is not owned by everyone (Bourne 1957). Based on the distribution of the resulting mean scores. four products were selected as representing each of the product categories: public-luxury (PUL), public-necessity (PUN). private-luxury (PRL), and private-necessity (PRN). The 16 products selected and their mean scores are presented in Table 1 Survey Design The research design called for separate reference group influence evaluations of product and brand decisions for 16 products, for a total of 32 evaluations. To have a manage- able questionnaire, it was decided that an individual re- spondent should be required to deal with a total of four evaluations. Thus, eight different versions of the survey instrument were constructed. Four versions contained only product decisions, and four contained brand decisions. One product was selected from each of the four categories— PUL, PUN, PRL. and PRN—to make up each of the ver- sions, Thus an individual respondent received a packet con- taining a series of reference group influence questions re- garding either product or brand decisions for four different products representing each ofthe four categories (.c., pub lic luxury through private necessity). Order of the product categories was randomized across the eight questionnaire versions to avoid order bias. The questionnaire and sample configurations are illustrated in Exhibit B. This design re- ITEMS CSO REFERENCE GROUP INFLUENCE exw RESEARCH DESIGN LAYOUT touy Newey PUL PUN caps Pablo Gotta) rata (n = 88) PRL PRN mate (poise) __telterton PUL PUN oe ee 5 PAL PRN Produc Bate Vane) oor ane) Bes Sens pu PUN rapa Pie (ens) asada f=) PAL PRN Private. (icemaker) (mattress) PUL PUN cups PARC __(sebea)—_womara ess (m= 78) PRL PRN Priale (wash compactor) (dare PUL PUN cops PH oo) (rae (n= Bey PRL PRN Pvt (potable) __(lagurton PUL PUN nowy Pele Camera) (wit (n= 75) PRL ‘PRN brand pve _(rVgune)__(oriamo) ear = PUL PUN coup Pubke__(enbeskis) (avanti Ko = 80) PRL PRN Private: (icemaker) (mattress) PUL PUN tome Pate etsy oman nn {n= 86) PRL PRN Private (rash compactor) __ (blanket) sults in decisions (product versus brand) and the product configurations (e.g. mattress, automobile, golf clubs) serv- ing as between-subjects factors. The luxury versus necessity atid public versus private dimensions represent withi subjects factors. Reference group influence was assessed using 13 of the 14 individual items developed by Park and Lessig (1977. 105). These items were designed to reflect informa- "One of the eight versions was protested on a convenience sample of 20 nonstndent adults for ease of understanding and completion. Based on the resis of this pretest, several items inthe reference group sale were tlighay modified to improve ela. 187 tional, value-expressive, and utilitarian reference group in- fluences. Informational reference group influence occurs when a person actively seeks information from people viewed as knowledgeable or observes the behavior of ac- knowledged experts. It is based on the concept of compar- ative influence suggested by Deutsch and Gerard (1955). \Value-expressive reference group influence is characterized by a person behaving in a manner that will improve his or her self-image or create the impression of attachment to the group (Kelman 1961). Utilitarian reference group influence is reflected in compliance to group norms or standards to gain rewards or avoid punishments that may be forthcoming from the group (Asch 1952). ‘The individual items were operationalized as six-point bipolar agree(6)-disagree(1) statements. Scales were scored so that higher values represented greater influence percep- tion. This is in contrast to the four-place “not relevant”’ to “uahy relevant” scales that provided the option of three ve and one negative position used by Park and Lessig. G57? A balanced sicpoint seal offered respondents & more complete range of alternatives. The sampling proce- dure used by Park and Lessig (1977) in studying house- wives and students involved telephone and mail surveys. This is somewhat different from the panel mailing used here, but is similar to the followup study described tater. Respondents were instructed to indicate their degree of ‘agreement with each item as it applied to product or brand selection decisions. Example items regarding product de- cisions for each of the three reference group subscales fol- low: (© Informational: An individual would seeic informa about poo! tables from fellow workers who are familiar ‘with them, (© Value-expressive: An individval would probably feel that purchasing a pool table would enhance his or her image among other people. (© Usiltarian: An individual's decision about whether or not to buy a poo! table would be influenced by the expectations of family members. ‘The three variations of group influence were represented 1s a summed composite of four informational, five value~ expressive, and four utilitarian items. Since the items were designed to reflect three types of reference group influence, the individual items were combined to form measures of informational, value-expressive, and utilitarian reference group influence. Combining items into separate measures of three constructs was supported across brand and product, decisions by coefficient alpha internal consistency esti- mates.” An informational influence item referring to inde- Average ternal consistency estimates (coeticient alpha) forthe prod- ‘uct decisions were 0.63, 0.86, and 0.71 for the informational, valve: ‘xpessve, and uilitariansubseler, respectively. For the brand decisions, the respective internal consistency estimates averaged 0.70, 0.80, and 0.71. Test-retest reliability was assessed using two convenience samples of 78 and 40 students fr one version ofthe product questionnaire and one ‘version ofthe brand questionnaire in a three-week test-retest administra tion, All the test-retest correlations were significant (p< 0.01): the average values ranged from 0.53 0 0.68. Copyright © 2001 All Rights Reserved ‘THE JOURNAL OF CONSUMER RESEARCH ‘TABLE 2 (OVERALL RESULTS OF MANOVA ANALYSES Informational Value-expressive Utitarian Fivalve dt Probabilty Favalue df Probability Fivalue df Probability Primary study (n= 645) ‘Between-subjects factors ‘Decision: Product vs. brand (PB) 18 4 o2 sa 4 45 7 1 ‘Grouping: Product configuration (PC) 3 24 s7 3 00 841300 Interaction: PB x PC 3 16 3% 3 at 9% 3 | (40 within-subjects factors. Lusasy vs. necessity (LN) 580981 90 1498 1 00 162201 Publi: vs. private (PP) 885 10 02761221 00 0 a4g0 100 Interaction: LN x PP 1498 ot 67391 00 7081 1 0 Fotowup study (n = 151) ‘Between-subjects factors (Grouping: Product configuration (PC)* oo st 744 4 ot aor 4 05 within-subjects factors (Decision: Produet vs. brand (PB) 347 1 96 ast 1 00 28431 0 Lunury vs. necessity (LN) ost 1 00 598 1 02 1821 18 Public vs. private (PP) 7058 1 00124231 00 3136 1 = 00 PB x LN 1838 1 00 7092 1 00 2515 1 0 PB x PP 803 1 00 Bt a7 24 72 LN x PP 1684 5 00 m4 39 256 1 4 PB x LN x PP 5521 02 125042 Yes 120 “Duleences in eyes rod er aos conigvaton ada vate owe of ny eo pdt conbnatont flowin pendent testing agencies was omitted because it appeared awkward for some products included in the study. Data were collected from 2 mailing to 800 members of a statewide consumer pane! during the summer of 1979. Panel households are selected to be representative of urban and rural residents with annual incomes above $5,000. However, the panel is somewhat upscale as a whole in terms of education and income when compared to Bureau of Census averages for the area ‘The total sample of 800 was divided into eight subsam- ples of 100. Each subsample was sent one of the eight versions of the questionnaire (see Exhibit B). Followup mailings resulted in 645 completed responses (80 percent useable response rate). The remaining 20 percent were equally divided between nonrespondents and incomplete or tunuseable replies. Cell sizes ranged from 71 to 88 for the four product decisions survey, and from 75 to 86 for the four brand decision evaluations. Comparisons of the de- rmographic characteristics of the original 800 with the 645, respondents did not suggest significant differences between Using the reference group subse and the product calegores (e-2., PUL) 28 tails in an analysis similar to that reported by Park and Lessig (1977). the reference group subscales were further examined using the multirai-mukimethod procedure sug~ tested by Campbell and Fiske (1959). The resokng corelatons between fnetsures of the same tal as evidence of convergence suggest thatthe informational measures are distinct from the value-expressve and wil tarian scales, Evidence of discrimination as demonstrated when corels- tions between a measure and another measie onthe same alae grealer ‘than correlations between measures having neither trait nor method in ‘common was provided for the product and brand analyses 66 of 72 and (65 of 72 times, respectively, respondents and nonrespondents or unuseable responses. The median family income category and average education for the respondents were $18,000 to $24,000 and 14.4 years, respectively. This is slightly lower than the profile for respondents to the pretest questionnaire. The instructions introducing the questionnaire carefully distinguished between product and brand decisions. This ‘was followed by an example in which product decisions were described as those involving a decision whether or not to buy, for example. a color television set. Choosing a color ‘TV from among Magnavox, RCA, Zenith, and others was used to exemplify a brand decision. Instructions were then provided for responding to 2 six-point agree-disagree scale. Depending on whether it was a product or brand decision questionnaire, a sample product or brand question about 10-speed bicycles was presented, along with a scale to rein- force the type of decision the respondent was to make. At the beginning of each section of the questionnaire, the words “‘product decisions” or “brand decisions" were printed and underlined, RESULTS ‘The overall results of the nested design were first ex: amined using repeated measures multivariate analysis of variance. These results—along with the results of a fol- lowup study described later—are presented in Table 2. The type of decision (product versus brand) approached signif- icance only for the informational reference group influence. Variations in the product configurations (the products in- ‘cluded in each questionnaire) were significant for the value- expressive and utilitarian influence evaluations, Seve REFERENCE GROUP INFLUENCE ‘The within-subjects factors (i.e., luxury versus necessity, public versus private) and the interactions were all signif icant across the three reference group subscales, but there were differences in the pattern of results. The luxury-necessity dimension of conspicuousness appeared particulatly sensitive regarding respondents” perceptions of informational reference group influence (F = 550.98, p < 0.01), In contrast, these overall results suggest that the public-private dimension affected value-expressive (F = 761.22, p < 0.01) and utilitarian (F = 214.30, p < 0.01) perceptions to a greater degree. Product Category Comparisons Differences in responses to the reference group influence scales across the four product categories (e.g., PUL, PUN, PRL. PRN) were examined separately for the product and brand decisions. The results of the repeated measures anal- ysis of Variance are presented in Table 3. In 23 of the 24 ‘cases (8 product configurations X 3 influence dimensions), the individual analysis of variance F-values were significant (@ < 0.001). These results suggest substantial differences. in consumer perceptions of reference group influence across. the four product categories represented by the specific prod- ucts used! in this study. Given these overall differences, individual paired- comparison tests were run for each of the possible pairs of product categories by each type of reference group influ ence to fest the research hypotheses. These results, which are analyzed separately for the product and brand decisions, are provided in Table 4, along with the directional hy- potheses., One-tailed tests were used when direction of in- fluence was hypothesized. A conservative significance level, (P< 0.001) was used to account for the increased proba- bility of finding differences with the large number of in- ividual comparisons For the nonequal hypotheses, 21 of 24 comparisons were significantly different. Eighteen of these differences were in the hypothesized direction. The three italicized pairs of values in Table 4 represent nonequal hypotheses that were found to be significantly different but counter to the ex- pected direction. These three significant differences re- flected public necessity-private luxury comparisons. Eight of the 12 equal hypothesis comparisons were found to be significantly different. This is not particularly dis- couraging because hypothesizing that two measures arc equal for different product categories is a fairly stringent assumption. In sum, 22 of the 36 comparisons were consistent with the hypotheses. For the product analyses, these consistent findings were clustered (5 of 6) in the informational refer- “in an ¢ffon wo simplify presentation of the results. the paired-compar- lson hypatbesis tests are based on averages across tbe product configu ‘ations for the brand and product decisions. Paired caests were also run fo product by product basis. These results were consistent with the average spore reslis summarized in Table 4. 189 cence group influences; in the brand analyses, they were ‘concentrated in the value-expressive (5 of 6) and utilitarian 6 of 6) reference group influences. The fact that three types of reference group influences ‘were measured does not imply that all three should be pres- cent or absent in an individual case. In fac, it would seem reasonable to find one type of influence operating and the ‘others absent in a particular situation. For example, in the purchase of @ man’s suit, value-expressive influence might play a much larger role than either informational or utili- tarian influence. Thus it is reasonable to consider the hy- potheses from the point of view of the presence or absence of any type of reference group influence. From that per- spective, all six relationships for product decisions and five out of six relationships for brand decisions are supported by the results of the panel study. Additional Findings A followup study was conducted in an attempt to partially replicate and extend the findings. The respondents in the previously described panel study were exposed to questions dealing with either product or brand decisions (i.e. the type of decision was a between-subjects factor). By means of the instructions and examples described earlier, care was taken to stress the distinction between brand and product decisions. The question is raised, however, whether the responses reflected differences in the two types of deci- sions. In this followup study. each respondent was asked first about reference group influence on product decisions, then about brand decisions. To verify the earlier product selection procedures, manipulation checks were included at the end of each survey regarding the luxury-necessity ‘and public-private dimensions. Respondents. Three hundred households selected ran- domly from the telephone directory of a medium-sized 'SMSA were contacted by telephone and asked to participate in a study of consumer purchase decisions; 270 agreed to respond to a mail questionnaire. The original phone re- spondents who were willing to participate were randomly assigned to one of two questionnaire versions. Larger sub- samples than those used in the panel study were felt to be justified, given the nature of the two samples (ie., random phone selection versus a panel survey) and the lower antic ‘pated response rate in the followup study. This followup ‘was based on 151 completed responses (50 percent response rate based on the original telphone sample of 300). Data were collected from the adult in each houschold agreeing to participate. Survey Design. Two of the four product combinations | used in the panel study were selected for this analysis. Two ‘questionnaires were constructed, each containing the ref- ference group influence questions regarding product and brand decisions for four different products representing the four categories (e.g., PUL). In both questionnaires, product, design questions preceded brand decision items. Again, the Copyright © 2001 All Rights Reserved 190 ‘THE JOURNAL OF CONSUMER RESEARCH TABLES MEAN SCORES AND SUBGROUP ANOVA RESULTS" Product decisions Group! _ letluence N PUL PUN PRL PRN Fevalue Group + 28 Cubs Suit Poo! tabie Rettigerator Informational 1988, 14.68 1302 1839 55.20 Value-expressive 1594 18.60 1351 849 85.05 Utltarian 11.98 1482 27 953 31.16 Group 2 85 Racket Watch TV game Lamp Informational 2001 17.01 1982 1652 37.93 Value-expressive 1429 1134 10.40 Bet 30.63 Unitarian 10.75 9108 10.48 925 7.30 Group 3 m Skis Automobile loomaker Mattress Informational 1920 187 1824 1473 38.78 Value-expressive 13.44 1489 9:19 810 45.49 Ubitarian 1049 1213 957 934 13.96 Group 4 78 Sailboat Dress “Trash compactor Blanket ‘informational 21.22 1513 20:33 15.50 6797 Value-expressive 1829 1895, 932 21 123.13 Unitarian 1268 1401 378 Baa 4278 Brand decisions PUN PRL PRN Fevalue Group 5 82 Suit Pool table Informational 14.35 19.04 18.00 58.02 Value-expressive 1631 1420 10.235, 3473 Unitarian 1278 1199 953, 1385 Group 6 18 Wateh ‘Wogame Lamp {aformational 17.63 1997 17.28 Value-expressive 1343 1088 1472 Ubitaran 10.98 11.01 9196 Group 7 80 ‘Automobile lcomaker Mattress Informational 20.27 1929 17.88 a7 Value-expressive 1349 967 B51 2853 Usilitarian 44 1087 882 3978 Group 8 86 Dress. ‘Trash compactor Blanket ‘formational 1676 20.77 16.75 4750 Value-expressive 1872 10.68 974 e162 Ubltarian 1428 97 929 1429 “At Fle eat one (Ge 6itarar were agar < 0001) order of the product categories was randomized between the two survey versions. ‘The reference group influence statements were identical to those used in the panel study. Each version was mailed to half of the adults agreeing to parcitipate as a result of the telephone interview. The completed responses were al- most equally divided (75 and 76) between the two survey versions. Reliability and Manipulation Check Estimates. Internal consistency was again estimated using coefficient alpha “The estimates averaged 0.64, 0.84, and 0.70 forthe infor- rational, value-expressive, and utilitarian subscales, re- spectively. Scaled statements similar to those used in the product selection procedure for public-private and luxury-necessity dimensions were included at the end of each questionnaire as manipulation checks. These tests for significance regarding the public-private and luxury—necessity dimensions were, for each possible comparison, consistent with the preliminary product selection procedures, The av- erage r-values were 4.62 for the public-private comparisons and 18.53 for luxury-necessity comparisons. Specifically, those products selected as representing privately consumed goods differed significantly from the products perceived as being publicly consumed, and the products selected as lux- tries were perceived differently from those selected as ne- cessities. Results. The followup study data were also examined in an overall analysis using repeated measures multivariate TGS Reserves REFERENCE GROUP INFLUENCE TABLE 191 4 PAIRED-COMPARISON HYPOTHESES AND RESULTS* Influence Comparison Hypotheses Informational Value-expressive Usitarian Products (h = 322) PUL-PUN > 2002 re21% 1552 15.85, 1181 PUL-PAL = 20.02 19.39", 1552 10.70" 1151 PUL-PRN, > 20.02 16.29% 1552 asst W351 PUN-PRL < ye21 sass 18.85 10.78 1242 PUN-PRN = 1621 1629" 15.85 ‘38 1242 9.16" PAL-PAN > 19.33 1629 10.79 838" 10.70 one Brands (n|= 323) PUL-PUN - 20.17 16.99¢, 1472 1842" 1183 sar PUL-PRL > 2017 1957 1472 1126 1183, 10.73% PUL-PRN > 20.17 7.19 1472 10.00% 1183 ost PUL-PAL, > 16.99 1957 1542 1126 1227 sores PUN-PRN > 16.99 1719, 15842 10.00" 1227 eat PRL-PRN = 1857 719 11.28 10.00" 10.74 9st ‘rigun ofess aan mean wares combined aces ain. ais iecata pars recon eneq hyper egy tr courte rpc rain, Tie reas teary aero p 20.00) ona tre wre wed wen Grecion hades Plvmahon and waa aes ac posite anes ot 424 vate yom vne bod a passe ange of 51 30 analysis of variance. These results are presented in the lower half of Table 2. As in the panel study, differences in the product configurations used to form the question- naires apparently did affect respondents” perceptions of value-expressive (F = 7.44, p < 0.01) and utilitarian (F = 4.01, p < 0.05) influence. However, unlike the panel study, the decision factor (product versus brand) was mar- ginally Significant for the informational (F = 3.47, p < 0,06) and significant for the value-expressive (F = 28.31, p< 0.01) and utilitarian (F = 28.43, p < 0.01) dimen- sions. It appears that when subjects responded to reference group influence questions for both product and brand de- cisions, the distinction affected responses. With the excep- tion of the luxury-necessity factor for utilitarian influence. the two onspicuousness dimensions hypothesized to affect, reference group influence perceptions were again signifi- cant. Differences in responses to the reference group influence scales across the four product categories (c.g.. PUL, PUN, PRL, PRN) were again examined separately for the two decisions. In 10 of the 12 analyses, the F-values were sig- nificant|(p < 0.001), again suggesting substantial differ- ences in Consumer perceptions of reference group influence actoss the four product categories. ‘The tesults of the paired-comparison tests for the 12 hy- potheses are shown in Table 5 for both the product and brand decisions. Based on the followup data, 25 of the 36 ‘comparisons were consistent with the hypotheses. For the nonequal hypotheses, only one significant comparison was ‘counter to the hypothesized direction. Fifteen of the 25 paired-comparison tests that were consistent with the ref- erence proup influence hypotheses involved the brand de- cision arialyses. DISCUSSION Summary ‘The present effort investigated three t ‘group influence on product and brand decisions across four product categories delineated by variations in product con- spicuousness. The interaction of public-private consump- tion and luxury-necessity dimensions resulted in four dif- ferent producvbrand combinations. Bourne's (1957) original framework hypothesized strong reference group in- fluence for public-loxury product and brand decisions and negligible influence on private-necessity product and brand decisions. Differential influence was hypothesized for public-necessity and private—luxury items. ‘This study is limited by the normal problems associated with a mail survey design and the use of projective re- sponses. A caveat is also appropriate regarding the impli- cations of omitted variables, such as perceived risk and product familiarity, as well as other confounding effects, For example, without a more complex design it is impos- sible to tell whether the significant effects for the decision factor (product versus brand) in the followup study are due to demand effects or whether the inclusion of both deci for each respondent resulted in the presence of meaningful differences. Regarding the possibility of confounding ef- fects due to omitted variables, similar efforts in the future should consider covariance analysis in attempting to control for other influences. This is particularly noteworthy because the distinction between luxuries and necessities implies varying costs, and hence risk. Given these limitations, what was learned about refer- ence group influence? When respondents were faced with Copyright © 2001 All Rights Reserved 192 ‘THE JOURNAL OF CONSUMER RESEARCH TABLE 5 PAIRED-COMPARISON HYPOTHESES AND RESULTS: FOLLOWUP STUDY* Influence Comparison Hypotheses Informational Value-expressive Unitarian Products PUL-PUN, 20.16 reas 1837 yaar 11.08 10.49, PUL-PRL 20.16 1932 1837 1120 1408 1026" PUL-PRN 20.16 16.19 1837 rare 11.08 87a PUN-PAL 1885 10.32 1321 m2 10.49, 10.28 PUN-PAN 1885 16:19" 1321 are 1048 873" PRL-PRN > 19.32 16.10 1121 ara 10.28 a7 . 1991 19.28" 15.19 15.98" 1163 1231 PUL-PRL > 1991 1924 15:19 11.08 11.63 1047 PUL-PRN > 1991 as 15:19 reors 1163 1078 PUN-PAL > 1928 1924 18.95 i108 123 1047" PUN-PRN > 1928, i785" 15.95 s2or* 1231 10.75" PRL-PRN * 1926 1788 11.08 s207 1047 1075" ‘Fpees preset sate mean scares conned sss siboupe.aleuad pa ol mau roprvats rovaqul hyphens agar deren but cout expec Src, ‘Pared nas wre scary ere © — 0.01) onetalg Iss wae ured whe econ hypotwseablomatoa! a ara vals Ra parte range 4 2m cree vos had a pss range 51030 "Eau bt song irene group iwreehpabwsced ‘Kasra tweak eltone up unc Nodeszee ‘Corte nth hpaestes uence a single decision type (e.g.. product versus brand), the de- cision factor was only significant for informational group influence. However, in both studies the absolute values were consistently greater for brand choices as opposed to product choice decisions. This suggests a greater role for appeals based on reference groups in stimulating selective demand. Variations in the sets of products selected affected perceptions of value-expressive and utilitarian reference ‘group influence. The absence of significant informational reference group effects suggests consistent information seeking by individuals across similar types of products. Consistent with Bourne's (1957) framework, the lux- ury-necessity and public-private dimensions were consis- tently significant as within-subject factors in both studies. This finding was reflected in substantial differences across the four categories—public luxuries, public necessities, pri- vate luxuries, and private necessities. Nineteen of 36 paired-comparison hypotheses were supported in both the panel and followup study.® Further, if these results are tem- ered by the fact that all three types of reference group influence would not be expected to be operative in all pur- chase situations, the results provide fairly strong evidence for the need to consider differential effects of reference group influence across purchase situations. Reexamination on a product-by-product basis of both the hypotheses tests and the reliability estimates did not reveal any noticeable patterns regarding products with confirmed hypotheses versus products with disconfirmed hypotheses. However, three observations are noteworthy. When con- Sin tual, 24 of the mean pairs in terms of direction and significance were replicated by the followup study. sidered individually, the two clothing items included as, public necessities were involved in slightly more of the ddisconfirmed hypotheses for product decisions. In general, the hypotheses were more often supported for the brand decisions. Based on the results presented in Tables 4 and 5, four of the directional hypotheses were disconfirme both studies. Three of these four were significant and dealt with public necessity and private Iuxury comparisons. For product decisions, public necessities were perceived as in- volving more value-expressive and utilitarian influence than private luxuries. This may reflect fear of embarrassment from not owning products which many feel are required for normal living. Also unexpected from the hypotheses, brand, decisions for public necessities involved less informational influences than private luxuries. This may be attributable to the fact that because necessity ownership is so common, Jess information seeking is necessary Unresolved Issues ‘The processes through which reference group influences operate and affect information processing, evaluation of alternatives, and eventual decision making are in need of study. The extended behavioral intention model (Fishbein, and Ajzen 1975) provides a logical framework for exam- ining many of these interactions. The model is capable of handling various types of decisions (e.g., brands versus products), incorporating different levels of specificity (e.g... situational factors and product differences), and is amenable to the use of experimental manipulations. Further, through ‘examining salient reference groups at the individual level, the differential role that varying groups may have on an individual's product and brand decisions can be explored. a REFERENCE GROUP INFLUENCE Situational variations and their impact on the complexity of studying reference group influences on purchase deci- sions alsa need to be addressed. For example, the purchase of a product such as beer may be viewed by others, but consumption may occur in private. Or visitors to the home may well have the opportunity to identify the brands of private luxuries, hypothesized here to involve weak refer- ence group influence. ‘As alluded to earlier, future efforts should consider in- cluding other socioeconomic and attitudinal variables, cither in covariance designs or as moderator variables. This premise is substantiated by the differential effects of ref- erence group influence between housewives and students found by Park and Lessig (1977). Similarly, perceived risk should increase susceptibility to reference group influence in many instances. Prior experience and knowledge have been found to affect product attribute versus brand pro- cessing (Bettman and Park 1980). Likewise, product class familiarity may be hypothesized to affect reference group influences. Generalized self-confidence and product- specific éelf-confidence found relevant in previous studies, con information seeking may also inhibit or encourage com- ‘munication about products and brands among reference ‘group members (Locander and Hermann 1979). Finally, the potential for changes in the perceptions of products among consumers and the pervasiveness of prod- Luct ownership on reference group influence need to be ac- knawledged. Through promotion, itis possible to associate certain images with products that might bring reference group influence into play (Bourne 1957; Lessig and Park 1978) under conditions (e.g., private necessities) that might not otherwise be expected. In contrast, product diffusion may shift products over time from exclusive to common ‘ownership, and hence reduce the significance of reference group influence. [Received May 1981. Revised February 1982.) REFERENCES Asch, S, (1952), Social Psychology. Englewood Cliffs, NJ: Pren- tice; Hall Beltman, James R. and C. Whan Park (1980), “Effects of Prior ‘Knowledge and Experience and Phase of the Choice Process ‘on Consumer Decision Processes: A Protocol Analysis. Journal of Consumer Research, 7 (December), 243-248, Bourne, Francis S. (1957), ""Group Influence in Marketing and Public Relations," in Some Applications of Behavioral Re- seatch, eds. R. Likert and S, P. Hayes, Basil, Switzerland: UNESCO. Brehm, Jack W. (1966), A Theory of Psychological Reactance, New York: Academic Press. [Burkrant, Robert E. and Alain Cousineau (1975), ““Informational ‘and Normative Social Influence in Buyer Bebiavior."* Journal ‘of Consumer Research, 2. (December), 206-215. Campbell, Angus. Philip E. Converse, Warren E, Miller, and Donald Stokes (1960), The American Voter, New York: John Wiley. 193 ‘Campbell, Donald T. and Donald W. Fiske (1959), "Convergent ‘and Discriminant Validation by the Multtrait-Multimethod Matrix,” Psychological Bulletin, 56, 81-105. Clee, Mona and Robert A. Wicklund (1980), “Consumer Behav- ior and Psychological Reactance,"” Journal of Consumer Re- search, 4, 389-408. Cocanongher. A. Benton and Grady D. Bruce (1971), “Socially Distant Reference Groups and Consumer Aspirations,"* Jour: nal of Marketing Research, 8, 379-381 Coleman, James S., Elihu Katz, and Herbert Menzel (1966). ‘Medical Innovation: A Diffusion Study, Indianapolis: Bobbs- Merrill. Deutsch, M. and Harold B. Gerard (1955), "“A Study of Nor- ‘mative and Informational Social Influences Upon Individust udgment,” Journal of Abnormal and Social Psychology. 51. 624-636, Ford, Jeffry D., and Elwood A. Ellis (1980), “A Reexamination ‘of Group Influence on Member Brand Preference,"" Journal of Marketing Research, 17, 125-132. Fishbein, Martin and leck Ajzen (1975), Belief. Attitude, Inten- tion, and Behavior, Reading, MA: Addison-Wesley French, John R. P., Jr. and Bertram Raven (1959), "*The Bases of Social Power," in Studies in Social Power, ed. Dorwin Cartwright, Ann Arbor: University of Michigan Press, 150-167 Grubb. Edwaid L. and Bruce L. Stern (1971), “Self-Concept and Significant Others,"" Journal of Marketing Research, 8. 382-385. Hawkins, Del 1, Kenneth A. Coney. and Roger J. Best (1980) Consumer Behavior, New York: John Wiley Hyman, Herbert H. (1942), “The Psychology of Status." Ar- chives of Psychology. 269. 94-102. ‘and Eleanor Singer (1968), Readings in Reference Group Theory and Research, New York: The Free Press. Jones, Edward E. and Harold B. Gerard (1967), Social Psychol ogy, New York: John Wiley. Kelley, Harold H. (1947), “*Two Functions of Reference Groups." in Readings in Social Psychology, eds. Guy E. Swanson, Theodore M. Newcomb, and E. L. Hartley. New ‘York: Holt, Rinchart & Winston, 410-414. Kelman, Herbert C. (1961), “Processes of Opinion Change. Public Opinion Quarterly, 28. 37-78. Kotler, Philip (1980), Marketing Management: Analsis. Plan- ring. and Control, Englewood Cliffs. NJ: Prentice-Hall, Kreisberg, Louis (1955), “Occupational Controls Among Stce! Distributors." American Journal of Sociology, $6. 203-212. Lessig, V. Parker and C. Whan Patk (1978), “Promotional Per- spectives of Reference Group Influence’ Advertising Impli cations.”* Journal of Advertising, 7. 41-87 Locander. William B. and Peter W. Hermann (1979), "The Effect ‘of Self-Confidence and Anxiety on Information Secking in ‘Consumer Risk Reduction,” Journal of Marketing Research. T6(May). 268-274. McNeal, James U. (1973), An Introduction 10 Consumer Beha- for, New York: Jobn Wiley. Merton, Robert K. (1957), ““Continuities in the Theory of Ref- ference Groups and Social Structure," in Social Theory and Social Seucture, ed. Robert K. Merton, New York: The Free Press, 281-368. ‘and Alice Kitt Rossi (1949), "Contributions to the Theory ‘of Reference Group Behavior.” in Social Theary and Social ‘Siructure, ed. Robert K. Merton, New York: The Free Press 225-215. Moschis, George P. (1976), "*Social Comparison and Informal Copyright © 2001 All Rights Reserved 194 Group Influencs 237-244, Newcomb, Theodore M. (1943), Personality and Social Change, ‘New York: Dryden, 374-386. Newcomb. Theodore M. (1950), Social Psychology, New York: Holt, Rinehart & Winston, 225. Park, C. Whan and V. Parker Lessig (1977), “Students and Housewives: Differences in Susceptibility to Reference Group Influences,"* Journal of Consumer Research, 4. 102-110. Sherif, Muzafer (1948), An Outline of Social Psychology. New ‘York: Harper & Row. (1953), “The Concept of Reference Groups in Human Relations,” in Group Relations at the Crossroads, eds. Mu- zafer Shetif and M. O. Wilson, New York: Harper & Row. and Carolyn Sherif (1964), Reference Groups. New York Harper & Row. Shibutani, Tamotsu (1955), “Reference Groups as Perspectives.”* ‘American Journal of Sociology, 60, 862-569. Journal of Marketing Research, 13, ‘THE JOURNAL OF CONSUMER RESEARCH ‘Smith, M. Brewster, Jerome Bruner, and Robert W. White (1956), Opinions and Personaliry, New York: John Wiley. Stafford, James E. (1966), “Effects of Group Influence on Con- sumer Brand Choice Preference,” Journal of Marketing Re- search, 3, 68-15. ‘Tumer, Ralph H. (1955), “Reference Groups and Future-Oriented Men," Social Forces, 34, 130-136. Venkatesan, M. (1966), "Experimental Study of Consumer Be- havior Conformity and Independence,"* Journal of Marketing Research, 3, 384-387. ‘Witt, Robert E. (1969), “Informal Social Group Influence on ‘Consumer Brand Choice,"” Journal of Marketing Research, 6, 473-471 and Grady D. Bruce (1970), “*Purchase Decisions and Group Influence."" Journal of Marketing Research, 7. 533-535, ‘and Grady D. Bruce (1972), “Group Influence and Brand ‘Choice Congruence,"* Journal of Marketing Research, 9, 440-443, a Copano ezO0T All Rights Reserved

You might also like