Dyadic Perceptions in Personal Source Information Search
ELSEV|ER
Laura J. Yale FORTLEWISCOLLEGE Mary C. Gilly UNIVERSITYOFCALIFORNIA
The vast majority of personal source information search research has ignored the dyadic relationship of information seeker and source, concentrating on one or the other. In this study, a method is used that retains the dyadic nature of personal information search, incorporating perceptions of opinion leadership, expertise, and influence from both members of the dyad. The results cast doubt on traditional measures of opinion leadership and some of the other implicit assumptions underlying personal source information search theory, j BUSNRES 1995. 32.225-237
he elevated role of personal source information is one of the cornerstones of communication, consumer behavior, and innovation diffusion theories. Five decades of research have provided volumes of theory and empirical evidence regarding the provision and use of information from personal sources, especially regarding the existence, characteristics, and influence of individuals who serve as providers of information for other groups members-individuals often termed "opinion leaders" within the literature (Rogers, 1983). These information providers have received a great deal of research attention within the fields of marketing and consumer behavior. Scholarly articles have detailed the correlates of these influentials, the relative importance of the information they provide, and the measurement techniques to be used in determining which individuals within a group are indeed influencers of other group members. However, the important dyadic component of personal source information search has been underresearched. The use of noncommercial interpersonal sources of information is a unique form of consumer information gathering. The word-ofmouth (WOM) channel is directly interactive as opposed to other channels of information provision (save the commercial channel of personal selling). Both the receiver and the source of the WOM perceive the communication process, and both can react to each other's input and presence. Perceptions in the WOM channel flow two ways; the source perceives a variety of attributes about the receiver and the communication situation, and the receiver perceives a variety of attributes about
T
Address correspondence to Laura J. Yale, Assistant Professor, School of Business Administration, Fort Lewis College, Durango, CO 81301. Journal of Business Research 32, 225-237 (1995) © 1995 Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010
the source and the situation. The interacting pair bring a number of existing relational characteristics and perceptions of each other to the information episode. Most previous research on WOM communications has separated the interacting pair and focused on only one or the other members of the dyad. The study described here was designed to retain the dyadic nature of the source-receiver relationship and to provide empirical results concerning theoretically assumed/proposed, but previously untested, relationships between WOM source and WOM receiver. Specifically, this investigation attempts to study the WOM dyad as a mutually perceiving pair of communicators. Three personal source perceptions are measured in this article: (1) source and receiver perceptions of "opinion leadership" in order to compare identification methods of opinion leadership, (2) source and receiver perceptions of product knowledge and experience, and (3) source and receiver perceptions of personal source influence in the receiver's product decision-making process. The next section provides a brief overview of research concerning personal sources, i.e., influentials, with special attention paid to measurement techniques used to identify influentials, the traditional "opinion leader" in particular. Following this critique, hypotheses are developed from existing theory with regard to source/receiver perceptions. The unique method used to test these hypotheses is then described, followed by the results and a discussion.
Identifying Personal Sources Persons who serve as information sources for others, that is influentials, have been assigned various terms within innovation, marketing, and consumer behavior literature. The most longstanding term has been "opinion leader" (see Rogers, 1983 for a nearly exhaustive review). First developed by Lazarsfeld, Berelson, and Gaudet (1948), opinion leaders are defined as individuals who are knowledgeable about a product category, and who are frequently able to influence others' attitudes or behaviors (Solomon, 1992). In 1955, Katz and Lazarsfeld were the first to do an in-depth consumer-related study to investigate the frequency, level of influence, and demographic characteristics of "opinion leaders."
ISSN 0148-2963/95/$9.50 SSDI 0148-2963(94)00048-J
226
J Busn Res 1995:32:225-237
Various other terms have been assigned subsequently to personal sources of information, based primarily on somewhat different views of the influential's characteristics and behaviors. The "innovator" and "market maven" are arguably the two terms most frequently applied to other types of influentials. The innovator (or early purchaser, early adopter) is termed such because of his/her propensity to adopt a new product/idea prior to others within the social group and without the benefit of word-of-mouth recommendations and positive experiences of others (Midgley, 1977; Midgley and Dowling, 1978). These innovators then become both visual and oral sources of information regarding the innovation. More recently, Feick and Price (1987) identified a different type of influential whom they term the "market maven." This personal source is distinct from both the traditional opinion leader and the innovator in that market mavens "possess more general marketplace information spanning product categories" (Price, Feick, and Higie, 1987, p. 333). The market maven is a willing provider of general "retail" information rather than an expert source of product-category-specific information. These three types of influentials are identified using a number of different methods. Rogers (1983) lists four methods for identifying and measuring opinion leadership: (1) sociometric, (2) key informant, (3) self-designation, and (4) observation. Utilizing the sociometric method, the researcher asks group members to identify to whom they go for advice and information about an idea. In the informant mode, the researcher chooses several key individuals from the social system who in turn indicate which members are opinion leaders for the group as a whole. The self-designation method consists of the administration of a scale, set of scales, or other question(s) to a sample of social system members. Opinion leaders are then identified by analysis of responses to the item(s). This method is the most frequently used of the four in marketing and consumer behavior studies (cf. Childers, 1986; Feick and Price, 1987; King and Summers, 1967; Richins and Root-Schaffer, 1988). Finally, in the observation method, the researcher directly observes interactions of group members and records instances of communication, inferring opinion leadership. Innovators have been identified using somewhat similar methods. Midgley and Dowling (1978) state that three methods have dominated the innovator identification research: (1) time of adoption of a single idea/product, (2) time of adoption of a cross-section of ideas/products, and (3) self-designation via a score compiled from items measuring innovativeness as a trait. Market mavens have so far been solely identified utilizing a selfdesignation questionnaire (see Feick and Price, 1987 for specific items). Each of these methods of identifying influentials has its advantages and disadvantages (Rogers, 1983). Sociometric methods rate well in terms of validity, but analysis of the data can be complex and the technique is applicable only to closed systems in which all members can be identified (cf. Ronchetto, Hurt, and Reingen, 1989). The method also entails interview-
L.J. Yale and M. C. Gilly
ing a large number of respondents in order to identify a relatively small number of influentials. The informant method is much easier and less time consuming to perform than the sociometric method but validity is dependent on the knowledge and insight of the informants used. The observation method is the most "valid" in that the researcher directly observes behavior and identifies influentials by observed instances of interpersonal influence. However, the method requires great patience and observation skills, is only practical with a small group, and assumes all interpersonal influence takes place where it can be observed. The self-designation method has been used to identify influentials. But the self-designation method is less valid than the sociometric or observation methods because the individual's perception of his/her role as an influential is measured, as opposed to any actual influencing behavior (Brooker and Houston, 1976). The method therefore depends on the accuracy with which the respondent knows his/her own behavior and reports it without exaggeration or bias. Kohn andJacoby (1973) found that self-designation of innovativeness did not strongly correlate with the nonsubjective identification method of time of adoption, whereas Jacoby (1972), in an opinion leadership study, did find that three identification methods, i.e., sociometric, self-designation, and key-informant, correlated significantly. An additional problem with all of the identification methods described previously is that none measures influence from the perspectives of both interacting parties. Measured via the sociometric and informant methods, influence is identified by the sole perceptions of the receiver. It is therefore implicitly defined as a characteristic bestowed upon the source by the receiver. The self-report/self-designation methods identify influencers by the sole perceptions of the influencers themselves. In this case, influence is defined as a characteristic self-assigned; respondents indicate whether they perceive themselves to be influentials. The observation technique is the measure most closely related to a dyadic measure in that both sources and receivers are observed simultaneously: However, the technique cannot determine influence perceptions on the part of either member of the dyad. Given the dyadic nature of interpersonal influence, the perceptions of both the giver and receiver are important in determining the level of influence manifested in the WOM episode. Perceptions of the communicating parties may be congruent or different with regard to any number of variables within the WOM episode (e.g., knowledge level of the interacting members, influence on the decision, etc.). The body of single perspective studies has revealed a great deal about the characteristics of influencers such as opinion leaders, innovators, and market mavens, along with information concerning those who seek WOM from others, and, to a more limited extent, added to knowledge regarding the WOM process itself. In addition, scholars have occasionally studied the personal influence process by incorporating both members of the communication episode, notably Arndt (1967a, 1967b, 1968a, 1968b) and Brown and Reingen (1987).
Dyadic Perceptions in Information Search
J Busn Res
227
199s:32:22s-237
The heart of the criticism is the assumption that a commonality of interpersonal perceptions between WOM source and receiver exists but this congruity has not been empirically tested. To test the commonality of perceptions between WOM receivers and their respective sources, we decided to focus upon the most frequently researched and traditional "opinion leader" sourcereceiver pair. Therefore, in the next section, several hypotheses are offered that are implied or assumed within the literature regarding the dyadic perceptions of source and receiver, specifically with regard to opinion leadership identification, product knowledge and experience, and level of influence.
leader to leader. This latter finding is consistent with the communication/sociology theory of homophily (Lazarsfeld and Merton, 1954; Rogers and Bhowmik, 1971) that most human communication will occur between a source and a receiver who are alike, i.e., homophilous, with regard to certain attributes, such as demographic variables, beliefs, values, etc. The Arndt study included both passive WOM receipt and active WOM seeking. In the case of active information seeking, WOM participants' perceptions with regard to opinion leadership of potential sources would be expected to be more conscious. The hypothesis suggested by the opinion leadership literature is: Hla: Information seekers will select interpersonal sources
Hypothesized Relationships Opinion Leadership Perceptions Opinion leadership has primarily been identified and investigated in marketing and consumer behavior through self-report on the part of the interpersonal source of information. These studies tell us about the perceptions of individuals concerning their own influencing behaviors, but they provide no insight into the WOM receiver's viewpoint. The receiver's perspective, it can be argued, is the more important perspective in cases of active information seeking, the information transmission most likely to result in significant personal influence (Engel, Blackwell, and Miniard, 1990). In the next discussion we utilize the term WOM seeker (in place of the more general WOM receiver) to emphasize that our study focuses on active seeking of information from a WOM source. Opinion leadership theory suggests that consumers recognize the distinct attributes of opinion leaders and seek out and utilize those sources they perceive as opinion leaders (Engel, Blackwell, and Miniard, 1990; Solomon, 1992). This notion has not been empirically tested. Alternatively, information seekers may not specifically recognize the differential characteristics of these sources, at least not in the same terms by which consumer/marketing researchers measure and identify opinion leaders. Reingen and Kernan (1986), in a study of a piano tuner's referral network, provide evidence that "nonleaders" are frequently used as personal sources. Operationalizing opinion leadership as frequency of information provision within the network, they found that only seven of 157 sources were accessed more than once, and five of these were music professionals. By Reingen and Kernan's identification of opinion leaders, all other personal sources would be considered nonleaders. Arndt (1968b) also supplies empirical evidence to support this possibility. Utilizing the sociometric method for opinion leader identification, he tested the "two-step flow" hypothesis of Lazarsfeld, Berelson, and Gaudet (1948) and found that there were almost as many WOM episodes reported in which flow was from informant-identified "nonleaders" to "leaders" as there was from "leaders" to "nonleaders." The majority of WOM flows, he found, were congruent on informant-defined leadership, nonleader to nonleader and
whom they perceive to be opinion leaders. The Arndt (1968b) study and the theory of homophily suggest a competing hypothesis: H1 b: Information seekers will select interpersonal sources
whom they perceive to be congruent to themselves in terms of opinion leadership (i.e., opinion leaders will select sources whom they perceive to be opinion leaders, and nonleaders will select sources whom they perceive to be nonleaders). In addition to the opinion leadership perceptions of the receiver regarding him/herself and the source, a third perception is the perception of the WOM source regarding his/her own opinion leadership. Again, opinion leadership theory implies that opinion leaders will perceive themselves to be opinion leaders. This assumption is the basis for the self-designation method of opinion leadership identification (cf. Childers, 1986; King and Summers, 1970). Therefore, an individual who has been selected for WOM transmission by an information seeker should rate him/herself as an opinion leader. The following hypothesis will be tested: H2: Interpersonal sources selected by information seekers
will perceive themselves to be opinion leaders. Theory and the variety of opinion leadership identification methods used (receiver identification of opinion leaders and opinion leader self-designation) are based on the assumption that the source and receiver agree on the source's opinion leadership (Rogers, 1983; Jacoby, 1972). Either member of the dyad should reliably identify opinion leaders and nonleaders as such. A fourth hypothesis is offered: I-I3: Information seekers and their interpersonal sources will
be congruent in their perceptions of the source's opinion leadership status.
Knowledge Perceptions Product knowledge is commonly proposed as an important determinant of personal influence (cf. Solomon, 1992). Opinion leadership identification studies have frequendy found that opinion leaders are more knowledgeable in the product area than
228
J Busn Res 1995:32:225-237
nonleaders (cf. Baumgarten, 1975; Leonard-Barton, 1985; Price and Feick, 1984). Product knowledge has been measured by consumer researchers by three methods (Brucks, 1985): (1) objective test (cf.Jacoby and Hoyer, 1981), (2) subjective evaluation (i.e., perception of own knowledge) (cf. Park and Lessig, 1981), and (3) inference from product experience (e.g., amount of time product owned, times purchased, etc.) (cf. Punj and Staelin, 1983; Srinivasan and Ratchford, 1991). In the case of active selection of an interpersonal source for the gathering of product information, subjective knowledge might be expected to be the more important measure of product knowledge. The information seeker likely selects a source based in part on his/her perception of the source's knowledge. Further, the seeker's perception is likely to be at least partially based upon the source's previously articulated subjective perception of his/her own product expertise. Individuals selected as personal sources are likely to perceive themselves as high in product knowledge for two reasons. First, consistent with personal source theory and findings, personal sources are indeed likely to possess greater objective product knowledge. For example, Bloch, Sherrell, and Ridgway (1986) found that heavy ongoing searchers (assumedly in possession of significant objective information on the product) also reported high levels of word-of-mouth activity. They concluded, "This result is not surprising, since the product expertise gained through ongoing search makes the searcher sought after by other consumers" (p. 125). In addition, attribution theory (Heider, 1958) suggests that sources may perceive themselves to be high in product knowledge expressly because they have been selected by another to transfer product information. They may attribute their selection to possession of greater product knowledge (i.e., their selection is "evidence" of their high knowledge status). Thus we propose that individuals will select an interpersonal source for product information whom they perceive to have significant knowledge regarding the focal product and that the source will perceive him/herself as having significant product knowledge. Further, it is expected that seekers will choose sources they perceive to be more knowledgeable than themselves. However, perceptions of the source's knowledge will likely differ between the information seeker and the selected personal source. The source is simply in a better position to form a more accurate and complete perception of his/her own knowledge than the information seeker. The seeker chooses to access the specific source due in part to the seeker's perception of the source's knowledge. Therefore, the seeker's perception of the source's knowledge may be higher than the perception of the source regarding his/her own knowledge. The following hypotheses regarding product knowledge will be investigated:
H4: Both information seekers and personal sources will perceive the source as high in product knowledge.
H5: Information seekers will rate their own product knowl-
L.J. Yale and M. C. Gilly
edge as lower than their ratings of their selected source's knowledge. H6: Information seekers will perceive their selected source's product knowledge as higher than the source will rate his/her own knowledge. Product experience is sometimes used as a surrogate or additional measure of product knowledge (cf. Punj and Staefin, 1983; Srinivasan and Ratchford, 1991). Theoretically, greater product experience in terms of usage and purchase frequency are causally linked to greater product knowledge. (Punj and Staelin (1983) found empirical support for this relationship among car purchasers.) Personal sources should thus also have owned the product for a longer period of time and have purchased the product more frequently than their information seeker counterparts. The next hypothesis will be tested:
H7: Personal sources selected by information seekers will have greater experience-based product expertise than their respective information seekers.
Influence Perceptions Influence is often suggested as another component of opinion leadership and other types of personal sources (Midgley, 1976; Robertson, Zielinski, and Ward, 1984; Rogers, 1983). A variety of studies have indicated that WOM sources are the most important/influential in consumer product decision-making (cf. Katona and MeuUer, 1954; Kiel and Layton, 1981; Murray, 1991 ; Price and Feick, 1984). Personal sources who are perceived as experts are likely to be perceived as more influential than sources who are perceived to be less knowledgeable. The source's perception of influence is also expected to be linked to his/her perception of product knowledge possessed; that is, if sources perceive their own product knowledge to be high, they will also perceive their influence to be high. The following hypothesis summarizes these ideas:
H8: Both information seekers' and personal sources' ratings of source influence will be related to their perceptions of sources' product knowledge. The personal source and his/her respective seeker may or may not be congruent regarding the source's influence in the product decision. For example, the source may not have an accurate idea of the extent of his/her influence on the information seeker. Some sources may overestimate their influence (possibly due to attribution), whereas others may underestimate the influence, not knowing the seeker's preference for/reliance upon WOM information or the seeker's high opinion of the source's product knowledge. Thus, the following exploratory hypothesis will be tested to determine the relationship between influence perceptions.
H9: Personal sources and information seekers will be congruent on their perceptions regarding the influence of the personal source.
Dyadic Perceptions in Information Search
Method
J Busn Res 1995:32:225-237
Table 1. Descriptive Statistics of Information Seekers and Their Selected Personal Sources
Sample The data used to test the hypotheses are composed of 94 information seeker/information source dyads. Information from both the information seeker and his/her personal source was collected in a unique two phase study. In the first wave, 438 non-faculty employees of two large California university campuses were mailed a questionnaire regarding likely behaviors when faced with a hypothetical product purchase. As part of the study, respondents were asked to select a personal source (other than spouse) whom they would likely contact for information prior to purchase of a video cassette recorder (VCR). The VCR was chosen as the focal product based on Beatty and Smith (1987). They found that, of five consumer electronics products, purchase of a VCR gave rise to the highest level of personal source information search. It therefore was expected that the VCR is a product for which consumers would indicate high likelihood of personal source search. Indeed, information seeker respondents within the 94 dyadic pairs indicated a high likelihood of use of WOM information, second (of seven possible information source types) only to retail comparison shopping (5.84 and 5.97 respectively, measured on a seven-point likelihood scale). The response rate for the first wave was 70%. Data from wave 1 were used in a previous personal search study (Yale, 1991). The dyadic data for the present investigation were gathered in wave 2. Information seeker respondents from wave one were also asked to participate in an optional second-wave study. They were provided with a letter of introduction and a postage paid envelope to use in forwarding an endosed source survey to be completed by their selected personal source. Ninety-four second-wave information source surveys were returned and paired with their respective information seekers (a second-wave response rate of 31%). No determination of the number of source surveys forwarded by first-wave information seeker respondents was made. It was feared that any attempt to gather report of first-wave forwarding behavior may have led to a significant reduction in first-wave respondents. Descriptive statistics of the 94 paired seeker/source dyads are given in Table 1. Although the respondents are not a random sample and therefore are not representative of the general consumer population, it can be seen that they compose a relatively heterogeneous group. To determine whether paired first-wave respondents differed from nonpaired first-wave respondents, comparisons on approximately 40 variables were made between first-wave respondents not matched to a source (and therefore not included in this data set) and first-wave respondents who are matched to their responding personal sources. The groups did not differ significantly (p < .05) on any demographic variables. They do, however, differ on two variables. The "with source match-up" seeker respondents used in this study were somewhat more likely to rate their selected source as an opinion leader than
229
Information Seekers
VCR owner? Yes No Total Sex Male Female Missing Total Education High school graduate Some college College graduate Graduate degree Total Age 18-24 25-34 35-44 45-54 55-64 65+ Total
Personal Sources
n
(%)
n
(%)
75 19 94
(79) (20) (100)
85 9 94
(90) (10) (100)
45 49
(48) (52)
94
(100)
68 24 2 94
(72) (26) (02) (100)
4 34 32 24 94
(04) (36) (34) (26) (100)
8 30 36 20 94
(09) (32) (38) (21) (100)
4 21 42 14 11 2 94
(4) (22) (45) (15) (12) (02) (100)
8 36 31 12 7 0 94
(09) (38) (33) (13) (07) (100)
Note:Exceptfor VCRownership,all demographicvariablesare reportedby information seekers.
were the "no source match-up" wave I only respondents (none of whom is represented in the data for this study). This higher opinion leadership likelihood is because the "with source matchup" seekers were somewhat more likely than "no source matchup" respondents to indicate their source did influence other people concerning VCRs (the second question used in the Feick and Price opinion leadership measure). Interestingly, a smaller percentage of "with source matchup" seekers indicated their source owned a VCR, 93%, compared to 98% of "no source match-up" respondents. This difference in the two proportions is statistically significant (z ~ 2.06, p m .02) but obviously both percentages are large majorities.
Construct Measurement Exhibit i details the operationalizations of the major constructs used in this study. In sociometric terms, all first-wave respondents (information seekers) would be identified as hypothetical WOM receivers and all second-wave respondents (information sources) would be identified as "opinion leaders." In addition to this sociometric identification of interpersonal roles, perceptual measures of opinion leadership, product knowledge, and level of influence were taken for both first-wave respon-
230
J Busn Res 1995:32:225-237
L.J. Yale and M. C. Gilly
dents (information seekers) and the respondents' matched personal sources (second-wave respondents). Data regarding VCR ownership, year VCR first obtained, and number of VCRs purchased were gathered from both information seekers and their SOUrCeS.
Opinion leadership was measured via a two-item dichotomous scale based on that developed by Feick and Price (1987). (See Exhibit 1 for specific item wording.) They found that their short two-item measure correlated highly with the commonly used King and Summers' (1970) scale (r = 0.75). Affirmative answers to both items resulted in a respondent being selfdesignated as an opinion leader. A negative answer to either question (or both) placed a respondent in the nonleader category. Of the information seekers, only 13 (13.8%) perceived themselves to be opinion leaders. Of the personal sources, 44 (46.8%) rated themselves as opinion leaders. The first question within the opinion leadership measure Exhibit 1. Constructs, Measures, and Operationalizations Theoretical Construct
Measure
Opinion leadership
Subjective product knowledge Subjective influence
Product expertise
Ownership Usage experience Purchase experience Subjective product knowledge
Influence
Episodic influence
Operational Definition
Yes/No response re "very knowledgeable about VCRs" Yes/No response re "sometimes influence other people in their purchase of, or opinions about, VCRs" Yes answers to both questions resulted in labeling the respondent as an opinion leader (Feick and Price, 1987) Self-report (yes/no) Years since VCR first obtained (Kid and Layton, 1981) Times purchased VCR (Punj and Staelin, 1983) 1-7 scale; knowledge as compared to other people: 1 = "one of the least knowledgeable people"; 7 = "one of the most knowledgeable people" (Brucks, 1985) 1-7 scale; level of influence source would have on which VCR seeker would buy: 1 - "very little influence"; 7 = "complete influence" (Brown and Reigen, 1987)
Opinion leadership, product expertise, and influence self-perceptions were obtained from both source and seeker respondents. In addition, seekers provided subjective measurements of their respective sources' opinion leadership and level of influence. Note:
was also used as a dichotomous measure of perceived product knowledge. In addition, a continuous scale item was included. In comparison to "other people," information seekers were asked to rate themselves and their selected sources regarding knowledge of VCRs on a scale ranging from (1) "one of the least knowledgeable people" to (7) "one of the most knowledgeable people" (Alba, 1984; Brucks, 1985). Sources similarly reported their subjective perceptions of their own knowledge of VCRs. Table 2 presents the distribution of respondents on perceived knowledge. Perceived personal source influence on product choice was measured on the part of both the information seeker and the information source. Similar to measures of influence used by Brown and Reingen (1987) and Corfman (1991), seekers and sources were both asked to indicate how much influence they thought the source would have on which VCR the seeker would actually buy, using a seven-point scale anchored by "very little influence" (1) and "complete influence" (7). The single item influence measure used in this study has been found in a pretest and a related investigation to correlate well with three additional influence statements, resulting in Cronbach's ot of 0.82 and 0.79 respectively. Table 2 also provides the distribution of responses on source influence. Product experience was measured in three ways. Both information seekers and sources were asked to report whether they owned a VCR, the year they first obtained a VCR, and the number of times they had purchased a VCR (Newman and Staelin, 1972; Punj and Staelin, 1983; Srinivasan and Ratchford, 1991). See Table 3 for distributions. Table 2. Seeker and Source Perceptions of Product Knowledge and Influence Seeker Perception of Own Knowledge n (%)
Very knowledgeable 18 (19) Yes No 76 (81) Missing Total 94 (100) Knowledge 7-point scale Mean 3.75 Median 4.00 1.45 SD
Influence 7-point scale Mean Median SD
Seeker Perception of Source Knowledge n (%)
Source Perception of Own Knowledge n (%)
70 23 1 94
50 44
(53) (47)
94
(100)
(75) (24) (01) (100) 5.34 6.00 1.13
Seeker's Perception of Source Influence
Source's Perception of Own Influence
4.86 5.00 1.22
4.06 4.00 1.49
4.73 5.00 1.25
Dyadic Perceptions in Information Search
J Bush Res 1995:32:22S-237
Table 3. Distribution of Experience Measures Seeker
Source
(n)
(n)
1 0 4 1 2 3 3 6 10
1 0 2 2 3 4 8 7 21 8 11 7 7 4 9 94 1984.26 1984 2.74
VCR first obtained (year) 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 Do not own Total Mean Median SD
18 9 6 1 19 94 1984.49 1985 2.69
Times purchased a VCR (n) 0 1 2 3 4 5 >5 Total Mean Median SD
24 47 16 5 1 1 0 94 1.47 1 .81
11
13 43 26 4 5
94 1.84 1 1.41
Note: The calculationof the times-purchasedmean excludesnonpurchasers.
To summarize, perceptual self-measures of opinion leadership, product expertise, and episodic influence were obtained from both seekers and sources. In addition, seekers also provided measures of their perceptions regarding their respective source's opinion leadership, product knowledge and level of influence.
Results Table 4 provides a summary of the hypotheses and the results of the various statistical tests. Discussion of the results will be provided in the concluding section.
Opinion Leadership Hypotheses The data do not support the contention that information seekers select interpersonal sources whom they perceive to be opinion leaders (Hla). Only 34% (32) of the information seekers rated their source as an opinion leader using the Feick and Price (1987) measures (Z ~ -3.10, p < .01). Based on findings by Reingen and Kernan (1986), Arndt
231
(1968b) and the theory of homophily, a competing hypothesis was also offered (Hlb). Information seekers were proposed to seek sources whom they perceived as similar to themselves in opinion leadership status, that is, seekers who perceived themselves as opinion leaders were expected to disproportionately select a source whom they also perceived to be an opinion leader, whereas self-perceived nonleader seekers were proposed to disproportionatdy select a source whom they also perceived as a nonleader. Consult Table 5 for cross-tabulations regarding opinion leadership perceptions. Five seeker respondents perceived both themselves and their sources as opinion leaders, whereas 54 respondents rated both members of the dyad as nonleaders. Thirty-five seekers rated themselves and their sources as incongruent in opinion leadership status. Therefore, in 59 of 94 cases the dyadic members were perceived as similar in opinion leadership status by the information seeker. However, to test the alternative "homophily" hypothesis, relative frequencies of seeker opinion leadership perceptions must be considered. The proportion of selfperceived opinion leaders who selected a source that they also perceived as an opinion leader is 38%. Self-perceived nonleaders selected those whom they perceived as opinion leaders in 33% of the cases. Similarly, the proportions of self-perceived opinion leaders and nonleaders who selected nonleaders is 62% and 67%, respectively. Although the proportions are consistent with the hypothesis, a test for association indicates the association is not statistically significant (X2 < 0.01). The self-designation method of opinion leadership rests on the assumption that individuals who rate themselves as opinion leaders will, in actuality~ be sought for product information by information seekers. Therefore, it was proposed that individuals selected by information seeking consumers would perceive themselves as opinion leaders (H2). Forty-four sources in the study (47%) rated themselves as opinion leaders; 50 rated themselves as nonleaders. Therefore, H2 is not supported (Z = -0.58, p >. 10). Although using the sociometric method of identification all sources would be deemed episodic opinion leaders, less than half of them perceive themselves to be opinion leaders regarding VCRs. It was also proposed that information seekers and their interpersonal sources would be in agreement regarding the opinion leadership status of the source, that is, that opinion leadership identification would be congruent no matter which member of the dyad was questioned (H3). Fifty-four pairs (57%) agreed on the leadership status of the source. Forty pairs did not share perceptions of the source's opinion leadership. Although the proportion of agreement is greater than the proportion who disagree, the association is not statistically significant (X2 = 1.21, p > .10). Only 18 (41%) of the 44 self-perceived opinion leader sources were perceived as opinion leaders by their dyadic seeker counterparts (see Table 5). This finding casts further doubt on the validity of using sociometric and self-designation measures of opinion leadership as interchangeable means for opinion leadership identification.
232
L.J. Yale and M. C.
J Busn Res 1995:32:225-237
Gilly
Table 4, Summary of Results Hypothesis
Result
Opinion leadership Hla: Information seekers will select interpersonal sources whom they perceive to be opinion leaders. Hlb: Information seekers will select interpersonal sources whom they perceive to be congruent to themselves in terms of opinion leadership. H2: Interpersonal sources selected by information seekers will perceive themselves to be opinion leaders. H3: Information seekers and their interpersonal sources will be congruent in their perceptions of the source's opinion leadership stares. Product expertise H4: Both information seekers and personal sources will perceive the source as high in product knowledge. H5: Information seekers will rate their own product knowledge as lower than their ratings of their selected source's knowledge. H6: Information seekers will perceive their selected source's product knowledge as higher than the source will rate his/her own knowledge. H7: Personal sources selected by information seekers will have greater experience-based product expertise than their respective information seekers. Influence H8: Both information seeker's and personal sources' ratings of source influence will be related to their perceptions of sources' product knowledge. H9: Personal sources and information seekers will be congruent on their perceptions regarding the influence of the personal source.
Not supported a Not supported b Not supported a Not supported b
Supported a,e Supported a Supported a.a Not supported a
Supported e Supported a,e
az-test of one(two)proportion(s). bx2 test of association. CZ-testof populationmean. d(p - .06). ePearsoncorrelation.
Product Expertise Hypotheses Hypothesis four, that is, that both information seekers and personal sources will perceive the source as high in product knowledge, was tested by analyzing seeker other and source selfperceptions on two knowledge variables: (1) the dichotomous variable regarding the source as someone who is "very knowledgeable about VCRs," and (2) the 1-7 product knowledge scale variable. Seventy information seekers perceived their sources to be "very knowledgeable about VCRs." This large proportion of affirmative perceptions supports the hypothesis (Z - 4.65, p < .01). As an additional test, the mean knowledge score given to sources by their seeker counterparts (on the seven-point scale) is 5.34 (SD = 1.13). This rating is significantly greater than the midpoint (4) on the scale (Z = 11.47, p < .01). In response to the question, "Are you someone who is very knowledgeable about VCRs?," 50 source respondents (53%) answered in the affirmative, but this proportion is not significantly different from chance (Z = 0.58, p > .10). However, the mean rating given by sources concerning their product knowledge on the seven-point scale is 4.73 (SD = 1.25). This mean is significantly greater than the midpoint of the scale (Z = 5.68, p < .01). Although sources were somewhat hesitant at labeling themselves "very knowledgeable," they tended to rate themselves as more knowledgeable than "other people." The results described above thus lend support for H4.
Seekers were proposed to rate their own product knowledge as lower than that of their selected sources (H5). In only 10 cases did the information seekers rate their own knowledge as greater than their rating of their source's knowledge. In 14 cases, seekers rated their source's knowledge as equivalent to their own, and in 69 cases, they perceived their source's knowledge to be greater than their own, supporting the hypothesis (Z = 4.63, p < .01). Information seekers were expected to rate their source's knowledge at a level higher than the source would rate his/her own knowledge (H6). In 16 cases, seeker respondents rated their chosen source's knowledge at a level lower than that selfperceived by the selected source, whereas in 24 dyads the pair rated the source's knowledge at the same level. In the other 54 pairs, the seeker perceived the source to be more knowledgeable than the source perceived him/herself to be. Therefore, 58% of the seekers rated their source's knowledge at a level higher than the source rated his/her own knowledge. This proportion is consistent with the hypothesis and statistically significant at the .06 level (one-tailed test) (Z = 1.55). The personal sources selected by information seekers were predicted to have had greater experience with the focal product than the respective information seeker (H7). For analysis of this hypothesis, product experience was operationalized in two ways: (1) years difference in ownership, calculated by subtracting the year of first ownership of the seeker from that of the source (in five of the 94 dyads, neither the source nor the
Dyadic Perceptions in Information Search
J Busn Res
1995:32:225-237
233
Table 5. Opinion Leadership Perceptions Seeker Perception of Source's
Opinion Leader Seeker self-perception of opinion leadership
Opinion Leadership Nonleader
Total
Opinion leader
5 (38) [16]
8 (62) [13]
13 (14)
Nonleader
27 (33) [841 32 [34]
54 (67) [871 62 [66]
81 (86) 94
Source's Self-Perception of Opinion Leadership
Seeker perception of source's leadership
Opinion leader Nonleader
Opinion Leader
Nonleader
Total
18 (56) [411 26 (42) [591 44 [471
14 (44) [28] 36 (58) [721 50 [531
32 (34) 62 (66) 94
( ) Row percent;[ ] columnpercent.
seeker had yet obtained a VCR), and (2) difference in purchase experience, calculated by subtracting the times a seeker had purchased from the times a source reported purchasing a VCR. Of the 89 dyads in which one or both members owned a VCR, 49 sources (55%) reported first ownership in a year prior to the information seeker. In the remaining 40 cases, the source either had owned a VCR for a shorter time than the seeker or both source and seeker reported obtaining a VCR in the same year. Although the proportion with longer ownership is as predicted, it is not significantly different from chance (Z = 0.94, p > .10). In 13 of the dyads, the source had never purchased a VCR, in 24 of the cases the seeker had never purchased, and in two instances both lacked purchase experience. In 39 cases (42%), the source did report greater purchase experience than the seeker. In 53 cases, the source had either purchased the same number of times as the seeker or fewer times than the seeker. Therefore, H8 fails to be supported. In fact, the data suggest the counterintuitive opposite hypothesis (Z = -1.54, p = .06 lone-tailed test]).
Influence Hypotheses Theory suggests that an information seeker should yield influence in a product choice decision to a personal source in relation to his/her perception of the source's product knowledge (H8). The Pearson correlation for source's perceived influence and source's perceived product knowledge, both mea-
fluence and source's perceived product knowledge, both measured by the perceptions of the information seeker, was 0.61 (p < .01). Just as information seekers were expected to relate their assigned level of source influence to their perception of the source's knowledge, sources were also expected to relate their level of influence to their perception of their own product knowledge. The correlation between the source's own perception of influence and product knowledge was high (r = 0.66). The data support the hypothesis that both an information seeker's and a personal source's perception of product choice influence is related to his/her subjective evaluation of the source's product knowledge. Due to a variety of perceptual differences, information seekers and their personal sources may or may not be congruent with regard to their perceptions of the level of influence to be played by the source in the product selection decision (H9). The Pearson correlation for the two perceptions of influence was significant but quite modest (r = 0.22, p < .05). Using a strict test of the proposition, only 19 of the 94 pairs saw exactly "eye to eye" on the level of influence of the source. However, using a less stringent test, 57 pairs (61%) perceived the source's influence to be the same or within one interval of the other's perception. This proportion is significantly different from chance (Z - 2.13, p < .05), and therefore the exploratory hypothesis is supported by the data. The two members of the dyad were surprisingly consistent in their perceptions of the level of in-
234
J Busn Res 1995:32:225-237
fluence that the selected source would play in the VCR product choice decision.
Discussion Due to the dyadic nature of the data collection method utilized in this study, it was possible to compare the perceptions of interpersonal information seekers and their respective information sources. Hypotheses derived from basic assumptions underlying personal source theory and identification and previous empirical research were tested regarding dyadic perceptions of opinion leadership, personal source product knowledge and experience, and level of personal influence. Statistical tests indicated that the data supported some of the hypothesized relationships, but several hypotheses failed to find support and these results indicate a need for caution with regard to opinion leadership identification and use of experience as a surrogate for knowledge.
Opinion Leadership None of the four opinion leadership perception hypotheses was supported by the data. Neither information seekers nor their selected personal information sources tended to identify the sources as opinion leaders in the terms that have come to be commonly utilized within marketing and consumer behavior. This lack of consistency between the other and self-designation methods of opinion leadership identification indicates a weakness of the self-designation method at least from an individual influence episode perspective. Although from a sociometric standpoint, all source respondents in this study would be identified as "opinion leaders," only one-third of information seekers and less than half of the selected personal sources perceived the source to be an "opinion leader" in the product category, using opinion leadership scale items. This finding seems of particular concern due to the simulation method used in this study. Given the "role-playing" method, information seekers could have selected their "ideal source" as their hypothetical source (ability to actually access the selected source was not needed in this case). The ideal personal source is logically someone who possesses the greatest product knowledge and lifestyle similarity to the information seeker (Price and Feick, 1984). However, even with the ease of selection, only 34% of the seekers selected sources they themselves rated as opinion leaders and only 47% of the selected sources designated themselves as leaders. These findings suggest that, at least from a single episodic perspective, our traditional method of opinion leadership identification and our notion of opinion leadership are not robust. The fact that an individual is cited by another as a personal influencer does not mean that either individual thinks of the source as an "opinion leader." Although based on results of Arndt (1968b), we expected to see a tendency for seekers to select sources they perceived as congruent to themselves in opinion leadership status, the
L.J. Yale and M. C. Gilly
data analysis did not reveal a significant association between opinion leadership perceptions. The failure to find this homophily effect may in part be due to the small number of seekers who perceived themselves as opinion leaders. The lack of support for the opinion leadership propositions may indicate that self-identified opinion leaders are only leaders "in their own minds" and that the concept opinion "leadership" has been oversold, at least in terms of active information seeking. Only 18 of the 44 self-perceived "opinion leader" sources were similarly rated by their respective seekers. These results suggest that perceptions of seekers and their sources are likely to differ. Other- and self-identification of other influential types, such as innovators or market mavens, may yield differing seeker and source perceptions as did this investigation regarding opinion leadership. Although they did not attempt to compare seeker and source perceptions of any particular influential type, other studies provide suggestive support that consumers often select personal sources who would not think of themselves as being particularly influential in any general sense. Feldman and Spencer (1965) found that new residents turned to friends, neighbors, and co-workers when selecting a physician. Few of these sources exhibited any medical expertise but tended to be in similar life circumstances. For example, childless couples tended to ask advice from another childless couple, whereas those with children tended to ask others who had children. Bither and Wright (1977) conducted a laboratory experiment to study personal sources. Almost one-third of novice golfers indicated preference for a similarly novice source over a more expert source. Brown and Reingen (1987), in a referral network study of piano teacher selection, demonstrate from the example WOM pairs described in their article that most of the links between seekers and their sources were between physically or emotionally close friends or acquaintances who did not exhibit particularly high knowledge or previous influencing behavior in the category. Price and Feick (1984) and Yale (1991) provide theoretical (and limited empirical) support for these nonleader to nonleader selections. They suggest that seekers will frequently select others who are perceived to be cooriented with the seeker. A nonexpert but cooriented personal source would have no specialized product knowledge but would likely share similar values and life experiences with the information seeker, and therefore is likely to share common product needs and wants. These cooriented pairings may be highly likely in information search for services and products in which personal taste is important.
Product Expertise Information seekers were found to be likely to perceive their selected source as high in knowledge, and rate the source's knowledge as greater than their own. Although sources were somewhat hesitant to label themselves as "very knowledgeable," the personal sources selected by seekers also perceived them-
Dyadic Perceptions in Information Search
J Busn Res
1995:32:225-237
selves to be fairly high in product knowledge, even though, as was proposed, they rated their knowledge as somewhat lower than the rating given them by their respective seekers. Product knowledge appears to be particularly important in determining the potential of a WOM source. This study adds to suggestive evidence (cf. Bloch, Sherrefl, and Ridgeway, 1986) regarding the importance of knowledge by measuring it from the dual perspectives of information seeker and source. Information seekers were likely to rate their selected source as high in knowledge, whereas the sources were likely to rate their knowledge at a somewhat lower level. However, the data suggest caution in equating actual product experience with subjective product knowledge. The expectations that personal sources will have greater product experience (in terms of length of ownership) and greater product purchase experience than their respective information seeking partners are not supported by the data. Personal sources selected by information seekers were not found to have owned the product for significantly longer than the seeker, nor were they found to purchase the focal product more frequently. In post hoc analysis, simple linear regressions of seekers' and sources' subjective ratings of their individual knowledge and the year the VCR was first obtained did yield significant relationships. Times purchased was also found to be related to subjective knowledge. But the difference between seeker's experience and source's experience was not great enough to support the stated hypothesis, suggesting differences in "ownership" and purchase experiences do not underpin the difference in knowledge between seeker and source.
Influence Some scholars point to influence as the most relevant aspect of opinion leadership (cf. Engel, Blackwell, and Miniard, 1990; Rogers, 1983). In this study, both seekers' and sources' ratings of the source's product choice influence were significantly related to their perceptions of the source's knowledge level. Additionally, although members of a dyad might be expected to differ in their perceptions of an influence encounter, dyad members within this study were found to be consistent in the level of influence they assigned to the source. In conclusion, this dyadic study of a hypothetical personal source influence episode suggests that the dyad members are somewhat consistent in their perceptions of the selected source's product knowledge level and his/her level of influence on product selection. However, opinion leadership perceptions are not consistent with the theory: the concept of the opinion leader does not appear to be well-established in the minds of either the information seeker or the personal source. Caution must be taken when we discuss opinion leadership in the context of active information seeking, whereas similar findings in Arndt's (1968b) study of passive information receipt call for additional research in all word-of-mouth contexts. We may need to rethink opinion leadership research and give greater attention to other perceptions, such as homophily, between seeker and source.
235
Limitations of the Study The findings of this study are subject to limitations that-must be noted. In order to alleviate the sociometric method problem of contacting a large sample in order to generate a significant number of sources, we utilized a hypothetical purchase situation scenario method. Although the method was particularly suited to the collection of data regarding paired seekers and sources, the hypothetical nature of the investigation limits the validity and generalizability of the study (see Murray, 1991 for arguments supporting the use of scenarios in information search studies). Given our sampling frame, many sample members were not in the market for a VCR at the time of our survey. Indeed, the mean likelihood for VCR purchase "within the next year" was only 2.37 on a seven-point scale. We do not know how this low likelihood of near-future purchase may have affected the realism of the scenario method, nor how responses may have been different if respondents had been actually in the process of purchasing a video cassette recorder. For example, respondents may have simply reconstructed what they did in their previous VCR purchase. However, one would expect the results concerning opinion leadership to be even weaker if the setting were changed to a real episode, in which ideal access situations are less likely to occur. Replication of the study using recent purchasers or in-the-market likely purchasers must be conducted in order to validate our results and conclusions. The findings may also be somewhat dependent on the focal product. Scholars generally have found that opinion leadership is monomorphic, limited to a single product category (Feick and Price, 1987). Information search for products other than the VCR, especially services (Murray, 1991), may yield results different from those found here. Although respondents represent "real world consumers," the sample used in the investigation is somewhat nonrepresentative of the consuming public at large, especially with regard to education and ownership of the product. These factors may also have influenced the results, most probably with regard to the weak relationship between product experience and product knowledge.
Future Research Additional research should be conducted to replicate and extend the findings of this study. Other products, especially those which depend more on taste or opinion than on expertise (e.g., movies, restaurants), could be used as focal products in replications of this study. It would be especially interesting to conduct a multi-product study in which differences in personal source selection and perceptions for products that are high in search, experiential, or credence qualities (Darby and Karni, 1973; Nelson, 1970; Zeithaml, 1981) are examined. Knowledge may be of less importance, and experience of more importance, in the case of products high in experiential or credence qualities. Additionally, a study comparing personal search for a recendy introduced product, a product in the growth stage
236
J Busn Res 1995:32:225-237
and a mature product may yield interesting, differing results regarding interpersonal source selected and seeker and source perceptions. A replication of this study using actual personal search episodes would be helpful in determining the possible impact of the hypothetical nature of this study on the results found here. In addition, passive WOM episodes should be investigated to compare dyadic perceptions in the non-active search case. Given the findings concerning opinion leadership, additional studies utilizing other opinion leadership self-designation scales (e.g., Childers, 1986) are called for. Although the measure used here has been validated (Feick and Price, 1987), other measures may provide other results and distinctions. In addition, similar studies which incorporate measures of other influential types, such as market mavens and innovators, would also be interesting. Further, other mutual perceptions, such as involvement, could help explain some of the results, especially those regarding the weak link between subjective knowledge and product experience. Finally, the dyadic method should be used more often in the future when studying interpersonal search and influence. The retention of the dyadic nature of the communication could yield new insights into the word-of-mouth process. For example, research in which paired seekers and sources reported on their typical external search behavior could shed light on the dyadic linking of consumers high and/or low on opinion leadership and opinion seeking, i.e., the socially integrated, the socially independent, the socially dependent, and the socially isolated (Reynolds and Darden, 1971; Schiffman and Kanuk, 1991).
Conclusion In conclusion, the study reported here offers a unique dyadic method for studying perceptions of interpersonal information seekers and their sources. By maintaining the bond between them a number of theoretically implied hypotheses were able to be tested. Many of the assumed relationships between information seekers and their sources failed to be supported by the data, in particular those regarding opinion leadership and product experience. Results suggest that perceptions of seekers and sources require additional study in order to further explain and understand the process of interpersonal information source se~ lection and influence.
References Alba, joseph W., The Effects of Product Knowledge on the Comprehension, Retention, and Evaluation of Product Information. Advances in Consumer Research 10 (1983): 577-580. Arndt, Johan, Word-of-Mouth Advertising and Informal Communication, in Risk Taking and Information Handling in Consumer Behavior. Donald F. Cox, ed., Graduate School of Business Administration, Harvard University, Boston, 1967a, pp. 188-239. Arndt, Johan, Role of Product-Related Conversations in the Diffusion of a New Product, J. Marketing Res. 4 (August 1967b): 291-295. Arndt, Johan, Word-of-Mouth Advertising and Perceived Risk, in Per-
L.J. Yale and M. C. Gilly
spectives in Consumer Behavior. Harold Kassarjian and Thomas Robertson, eds., Scott, Foresman, & Company, Glenwood, IL. 1968a, pp. 330-336. Arndt, Johan, A Test of the Two-Step Flow in Diffusion of a New Product, Journalism Quarterly 45 (Autumn 1968b): 457-465. Baumgarten, Steven A, The Innovative Communicator in the Diffusion Process. J. Marketing Res. 12 (February 1975): 12-18. Bither, Stewart W., and Wright, Peter, Preferences Between Product Consultants: Choices vs. Preference Functions. J. Consumer Res. 4 (June 1977): 39-47. Bloch, Peter H., Sherrell, Daniel L., and Ridgway, Nancy M., Consumer Search: An Extended Framework. J. Consumer Res. 13 (June 1986): 119-126. Brooker, George, and Houston, MichaelJ., An Evaluation of Measures of Opinion Leadership, in Marketing: 1776-1976 and Beyond. Kenneth L. Bernhardt, ed., American Marketing Association, Chicago. 1976, pp. 564-567. Brown, Jacqueline Johnson, and Reingen, Peter H., Social Ties and Word-of-Mouth Referral Behavior. f. Consumer Res. 145 (December 1987): 350-362. Brucks, Mettle, The Effects of Product Class Knowledge on Information Search Behavior. J. Consumer Res. 12 (June 1985): 1-16. Childers, Terry L., Assessment of the Psychometric Properties of an Opinion Leadership Scale. J. Marketing Res. 23 (May): 184-188. Corfman, Kim, Perceptions of Relative Influence: Formation and Measurement. J. Marketing Res. 28 (May 1991): 125-136. Darby, Michael R, and Karni, Edi, Free Competition and the Optimal Amount of Fraud. J. Law and Econ. 16 (April 1973): 67-86. Engel, James F., Blackwell, Roger D., and Miniard, Paul W., Consumer Behavior (6th ed.), Dryden Press, New York. 1990. Feick, Lawrence F., and Price, Linda L., The Market Maven: A Diffuser of Marketplace Information. J. Marketing 51 (January 1987): 83-97. Feldman, Sidney P., and Spencer, Merlin C., The Effect of Personal Influence in the Selection of Consumer Services, Marketing and Economic Development. Peter Bennett, ed., American Marketing Association, Chicago. 1965, pp. 440-452. Heider, Fritz, The Psychology oflnterpersonal Relations, Wiley, New York. 1958. jacoby, Jacob, Opinion Leadership and Innovativeness: Overlap and Validity, in Proceedings, Third Annual Conference, The Assoc. for Consumer Research. 1972, pp. 632-649.
Jacoby, Jacob, and Hoyer, Wayne, What if Opinion Leaders Didn't Know More? A Question of Nomological Validity. Advances in Consumer Research 8 (1981): 299-303. Katona, George, and Mueller, Eva, A Study of Purchasing Decisions, in Consumer Behavior: The Dynamics of Consumer Reaction. L. H. Clark, ed., New York University Press, New York. 1954, pp. 30-87. Katz, Elihu, and Lazarsfeld, Paul F., Personal Influence: The Part Played by People in the Flow of Mass Communications, The Free Press, Glencoe, IL. 1955. Kiel, Geoffrey C., and I~yton, Roger A., Dimensions of Consumer Information Seeking Behavior. J. Marketing Res. 18 (May 1981): 233-239. King, Charles, Fashion Adoption: A Rebuttal to the "Trickle Down" Theory, in Toward Scientific Marketing. Stephen Greyser, ed, American Marketing Association, Chicago. 1964, pp. 108-125. King, Charles, and Summers, John, The New Product Adoption Research Project: A Survey of New Product Adoption Behavior Across a Wide Range of Consumer Products Among Marion, Indiana Homemakers, In-
Dyadic Perceptions in Information Search
stitute for Research in Behavioral, Economic and Management Science, Purdue University, Lafayette, Indiana. 1967. King, Charles, and Summers, John, Overlap of Opinion Leadership Across Consumer Product Categories. J. Marketing Res. 7 (February 1970): 43-50. Kohn, Carol A., and Jacoby, Jacob, Operationally Defining the Consumer Innovator, in Proceedings of the American Psychologlcal Association. American Psychological Association, Washington, DC. 1973. Lazarsfeld, Paul, and Merton, Robert, Friendship as a Social Process: A Substantive and Methodological Analysis, in Freedom and Control in Modem Society. Monroe Berger et al., eds., Van Nostrand, New York. 1954. Lazarsfeld, Paul, Berelson, Bernard, and Gaudet, Hazel, The People's Choice (2nd ed.), Columbia University Press, New York. 1948. Leonard-Barton, Dorothy, Experts as Negative Opinion Leaders in the Diffusion of a Technological Innovation. J. Consumer Res. 11 (March 1985): 914-926. Midgley, David F., A Simple Mathematical Theory of Innovative Behavior. J. Consumer Res. 3 (June 1976): 31-41. Midgley, David F., Innovation and New Product Marketing, Halsted Press, John Wiley & Sons, Inc., New York. 1977. Midgley, David F., Patterns of Interpersonal Information Seeking for the Purchase of a Symbolic Product.J. Marketing Res. 20 (February 1983): 74-83. Midgley, David F., and Dowling, Grahame R., Innovativeness: The Concept and Its Measurement. J. Consumer Res. 4 (March 1978): 229-242. Murray, Keith B., A Test of Services Marketing Theory: Consumer Information Acquisition Activities.J. Marketing 55 (January 1991): 10-25. Nelson, Philip, Advertising as Information. J. Political Econ. 81 (JulyAugust 1970): 729-754. Newman, Joseph W., and Staelin, Richard, Prepurchase Information Seeking for New Cars and Major Household Appliances.J. Marketing Res. 9 (August 1972): 249-257. Park, C. Whan, and Lessig, V. Parker, Familiarity and Its Impact on Consumer Decision Biases and Heuristics.J. Consumer Res. 8 (September 1981): 223-230. Price~ Linda L., and Feick, Lawrence F., The Role of Interpersonal Sources in External Search: An Informational Perspective. Advances in Consumer Research 10 (1984): 250-255. Price, Linda L., Feick, Lawrence F., and HiDe, Robin A., Information Sensitive Consumers and Market Information. J. Consumer Affairs 21 (1987): 328-340.
J Busn Res 1995:32:225-237
237
Punj, Girish, and Staelin, Richard, A Model of Consumer Information Search Behavior for New Automobiles. J. Consumer Res. 9 (March 1983): 366-380. Reynolds, Fred D., and Darden, William R., Mutually Adaptive Effects of Interpersonal Communication. J. Marketing Res. 8 (November 1971): 449-454. Reingen, Peter H., and Kernan, Jerome B., Analysis of Referral Networks in Marketing: Methods and Illustration.J. Marketing Res. 23 (November 1986): 370-378. Richins, Marsha L., and Root-Shaffer, Terri, The Role of Involvement and Opinion Leadership in Consumer Word-of-Mouth: An Implicit Model Made Explicit. Advances in Consumer Research 15 ( 1988): 32-36. Robertson, Thomas S., Zielinski,Joan, and Ward, Scott, Consumer Behavior, Scott, Foresman, & Company, Glenview, IL. 1984. Rogers, Everett M., Diffusion of Innovations, The Free Press, New York. 1983. Rogers, Everett M., and Bhowmik, Dilip K., Homophily-Heterophily: Relational Concepts for Communication Research. Public Opinion Q. 34 (1971): 523-538. Rogers, Everett M., and Shoemaker, F. Floyd, Communication of Innovations: A Cross-Cultural Approach (2nd ed.), The Free Press, New York. 1971. Ronchetto, John R., Jr., Hutt, Michael D., and Reingen, Peter H., Embedded Influence Patterns in Organizational Buying Systems. J. Marketing 53 (October 1989): 51-62. Schiffman, Leon, and Kanuk, Leslie Lazar~Consumer Behavior (4th ed.), Prentice Hall, Englewood Cliffs, NJ. 1991. Solomon, Michael R., Consumer Behavior, Allyn and Bacon, Boston, MA. 1992. Srinivasan, Narasimhan, and Ratchford, Brian T., An Empirical Test of a Model of External Search for Automobiles. J. Consumer Res. 18 (September 1991): 233-242. Yale, Laura, The Role of Expertise and Coorientation in Personal Source Selection, in Marketing Theory and Applications, Vol. 2 (1991 AMA Winter Educators Conference Proceedings). Terry L. Childers et al., eds., American Marketing Association, Chicago. 1991, pp. 395-402. Zeithaml, Valerie, How Consumer Evaluation Processes Differ Between Goods and Services, in Marketing of Services. James H. Donnelly and William R. George, eds., American Marketing Association, Chicago. 1981, pp. 186-190.