Money Talks … to Online Opinion Leaders
What Motivates Opinion Leaders To Make Social-Network Referrals?
MENGZE SHI Rotman School of Management, University of Toronto mshi@rotman.utoronto.ca ANDREA C. WOJNICKI Independent marketing consultant Andrea.Wojnicki@ gmail.com

This study investigated the effectiveness of intrinsic versus extrinsic motivations for consumers’ online social-network referrals, specifically across “opinion leaders” and “non–opinion leaders.” The authors utilized a unique dataset that matched a survey with an online field experiment. The empirical results indicated “money talks”—that is, online referral rates were higher when extrinsic rewards were conferred. Notably, the effect of an extrinsic reward was significantly stronger among opinion leaders. In this paper, the authors highlight the significance of reputational concerns and referral motivations in this context. Opinion leaders may have developed a reputation of intrinsically motivated referrals across their social networks, shielding them from potential loss of social capital associated with extrinsic rewards.
INTRODUCTION Given the commercially and academically documented impact of social networks in the marketplace, many marketers seek to harness the power of consumer referrals or word of mouth (WOM), whether it be online through social media (Kaiser, Kröckel, and Bodendorf, 2013) or offline. In the quest to inspire consumers to talk, some marketers offer rewards (i.e., discounts or gift certificates) for social-network referrals—the implicit rationale being that consumers need some tangible incentive
Bloomberg Business Week, “I Sold It through the Grapevine,” May 29, 2006. The Wall Street Journal, “If You Get Paid to Tweet, Disclose It, U.K. Says,” Paul Sonne, January 10, 2011.
1

to generate WOM. Meanwhile, the media and consumer watchdogs tend to discourage the contamination of this otherwise pure inter-consumer communication, 1 and research outside of the consumer-referral context indicates that these incentive strategies may backfire (Kamenica, 2012; Frey and Oberholzer-Gee, 1997).

• This research substantiates and reinforces the significant role of online “opinion leaders” in the marketplace. Referral campaigns not only should be targeted but offer valued, tangible rewards to opinion leaders.

• Empirical results indicate that consumers’ online referral rates are higher when extrinsic (i.e., financial) rewards are conferred, particularly among opinion leaders.

• Opinion leaders may have developed a reputation of intrinsically motivated referrals, shielding them from potential loss of social capital associated with extrinsic rewards.

• Despite the rhetoric that incentive strategies may backfire, tangibly compensating consumers for making product referrals to other consumers in their social network may be effective, particularly as opinion leaders may be most likely to respond to such tactics.

DOI: 10.2501/JAR-54-1-081-091

March 2014

JOURNAL OF ADVERTISING RESEARCH

81

MONEY TALKS … TO ONLINE OPINION LEADERS

This study sought to establish the online audience that most likely would leverage an extrinsic reward as compensation for online referrals—a group the authors refer to as “opinion leaders”—and compare their responses to those disinclined to accept such offers—“non–opinion leaders” in the authors’ vernacular. By doing so, the study sought to develop a better understanding of opinion leaders’ motivations. The authors focused on specific intrinsic (as opposed to extrinsic) rewards (Godes et al., 2005). Consumers who generated referrals—based on their interest and involvement in a product category and their desire to help others—would be classified as intrinsically motivated. The rationales for engaging in intrinsically motivated behaviors are inherent in the behaviors themselves. Extrinsically motivated behaviors, by contrast, mediate an external reward; participation is a “means to an end” (Kruglanski, 1975). Marketplace examples of extrinsically motivated online referrals by opinion leaders include pharmaceutical companies paying senior experts to influence attitudes regarding new drugs (Moynihan, 2008) and cell phone manufacturers providing free phones to influential bloggers and encouraging them to mention the phones in their blogs (Kozinets, de Valck, Wojnicki, and Wilner, 2010). The current study’s authors asked a key question: Do influential opinion leaders, who are assumed to be intrinsically motivated to generate referral across their social networks, make incremental referrals when they are extrinsically motivated? The rest of the paper is organized as follows: • Theoretical Grounding and Hypothesis The authors consider the literature on opinion leaders (King and Summers, 1976; Childers, 1986; Iyengar, Van den Bulte, and Valente, 2011); intrinsic and

extrinsic motivations (see above); and “social capital,” defined here as “the social resources embedded in one’s social network” (Lin and Erickson, 2008). Although intrinsically motivated consumer referrals may enhance social capital, referrals motivated by extrinsic rewards can do the opposite: When consumers receive referrals from “friends” who are compensated for spreading the word, the receiver may make a judgment regarding the motivation behind the referral. • Addressing the Key Question The authors propose and demonstrate that opinion leaders may have earned an abundance of social capital based on their reputation for providing helpful recommendations to their social networks. As a result, more incremental referrals can be induced by extrinsic rewards from opinion leaders because their expected social capital loss is much smaller than it would be for non–opinion leaders. • Discussion about the Data The data in the current study matched a sizeable survey with an online field experiment, and the survey measures Internet opinion-leadership scores and general demographics. The subsequent experiment randomly assigned participants to intrinsic and extrinsic motivation conditions and monitored online referrals for a Web site. • Analysis and Results Very few participants in the intrinsicmotivation condition made referrals. When extrinsic rewards were offered, however, the referral rate was substantially higher. Moreover, the effect of extrinsic rewards significantly increased with opinion-leadership scores. (A simple economic model of referral that

supports and formalizes this theory and pattern of results is presented in the Appendix.) • Conclusion The authors summarize key results, managerial implications, and future research directions. CONCEPTUAL DEVELOPMENT Intrinsic Motivation versus Extrinsic Rewards for Referrals Consumers readily seek to establish and maintain their identities (Thompson, 2013), knowing that others may infer motivations from their behaviors. Motivations for referral behaviors can be classified into two main groups: • Intrinsic Motivations: The reasons for engaging in intrinsically-motivated behaviors are inherent in the behaviors themselves—such as entertainment value, the opportunity to improve skills (such as consumer expertise), and social obligation (Kruglanski, 1975). These reasons readily translate to the WOM domain: Opinion leaders may generate referrals based on their involvement and knowledge in the category (Jacoby and Hoyer, 1981; King and Summers, 1976) or for social reasons. When they generate WOM, consumers may earn (or lose) “social capital” (defined as the “social resources embedded in one’s social network”; Lin and Erickson, 2008). Like financial capital, social capital can be accumulated by helping others and, inversely, can be retrieved by profiting from others (Kadushin, 2004). Past research has documented that social capital—in the form of social status, power, or self-enhancement—is a common, inherent motivation associated with WOM (Gatignon and Robertson, 1986; Hennig-Thurau, Gwinner,

82

JOURNAL OF ADVERTISING RESEARCH

March 2014

MONEY TALKS … TO ONLINE OPINION LEADERS

Walsh, and Gremler, 2004; Wojnicki and Godes, 2011). • Extrinsic Motivations: The rationale for participating in an extrinsically motivated behavior is mediated by the reward, with the focus on extrinsic compensation, such as cash, discounts or free goods (Kruglanski, 1975). These rewards are typical of those offered by marketers to consumers who make referrals. Offering an extrinsic reward for something that otherwise was intrinsically motivated can lead to an “overjustification” (Lepper, Greene, and Nisbett, 1973) or “crowd-out” (Kreps, 1997) effect, whereby consumers become less likely to participate in the incented activity. These effects have been demonstrated in a number of non-consumer-referral contexts, such as employees’ incentives and productivity (Murdock, 2002; Benabou and Tirole, 2003), and residents’ price incentives and citizenship behaviors (Frey and Oberholzer-Gee, 1997). Not all referrals enhance social capital; the direction of change in social capital depends on the value of information and the perceived motivation behind the referral. Specifically: • When a consumer’s referral is perceived to be motivated intrinsically, the consumer gains social capital; and • when a referral from a consumer is perceived to be induced by extrinsic rewards for the consumer’s own benefit (e.g., a cash reward), the consumer may suffer a loss of social capital. Although a receiver does not directly confer the external reward to the referrer, it is their shared social relationship that facilitates the transaction (i.e., the referral) that confers the social capital or intrinsic

reward. Conceptually, the evolution of social capital follows a reciprocal process with consumer referrals, where referrers deposit social capital through intrinsically motivated referrals and withdraw social capital by profiting from referral rewards. When a consumer does not receive a reward for making a referral, the receiver is likely to infer that the referral was intrinsically motivated. As a result, the sender gains social capital. When a consumer receives a reward for making a referral, however, it is uncertain whether the referral was intrinsically motivated or induced by the reward. This may partially explain research results indicating that extrinsic rewards are particularly effective in increasing referrals among consumers with weak ties (Ryu and Feick, 2007). Reputation Matters The assumption of intrinsic motivation usually invokes the recall of their past referral interactions. If a consumer has made more intrinsically motivated referrals in the past, it is more likely that the consumer’s current referral will be perceived as intrinsically motivated. In other words, a consumer’s reputation matters. When receiving a reward for making a referral, if a consumer has historically made intrinsically motivated referrals, the current referral would be more likely perceived as intrinsically motivated. For those consumers with strong reputations of making intrinsically motivated referrals, a referral—despite being associated with an extrinsic reward—still may generate a positive gain in social capital. If a consumer has seldom made referrals before, however, no information regarding past motivations is available, and the current referral very likely would be perceived as induced by the reward. With the absence of reputation for making intrinsically motivated referrals, one expects negative social capital from receiving referral rewards.

Opinion leaders are those consumers who generate a higher proportion of WOM in a given category (King and Summers, 1976; Childers, 1986), even (or perhaps particularly) in the absence of extrinsic incentives. As opinion leaders often are experts in their fields of interest (Jacoby and Hoyer, 1981; Iyengar, Van den Bulte, and Valente, 2011), it takes less time for them to understand product categories and identify ideal products for referral. Following earlier discussions, when an opinion leader receives a referral reward, the consumer’s referral motive more likely is to be perceived as intrinsic relative to a non–opinion leader. Non–opinion leaders’ referrals, however, are likely to be perceived as induced by the reward. When a marketer provides an extrinsic reward for referrals—and no other considerations have influenced the exchange— the reward should boost the consumer’s incentive to make a referral. However, the consumer may anticipate the potential loss of social capital due to the reward. Compared with the case without extrinsic rewards, the consumer likely will make a referral if—and only if—the consumer’s value of extrinsic reward exceeds the expected loss of social capital. The net effect of extrinsic reward on referral propensity clearly depends on opinion leadership: The incremental loss of social capital due to receiving an extrinsic reward is smaller for opinion leaders than for non–opinion leaders. A long history of intrinsically motivated referrals reduces the likelihood that the current reward-associated referral will be perceived as reward-inducing. By contrast, non–opinion leaders do not have a reputation for making intrinsically motivated referrals. As a result, when receiving rewards from current referrals, these referrals likely will be perceived as selfish and extrinsically motivated. As the opinion leaders are expected to suffer a smaller

March 2014

JOURNAL OF ADVERTISING RESEARCH

83

MONEY TALKS … TO ONLINE OPINION LEADERS

amount of social capital loss, they likely will respond to external rewards. Reflecting this discussion, the authors hypothesize:

TABLE 1 Summary Statistics of Survey Data
Observations Mean Gender (=1 for Female, =0 for Male) 3,219 3,219 3,219 2,726 3,219 3,219 0.406 Std. Dev. Min Max 0.491 0 18 0 1 1 6 1 89 1 11 7 30

H1:

Opinion leaders are more responsive to extrinsic rewards.

Age (years) Married (=1 if married, =0 otherwise) Household income (scale)

48.197 12.768 0.667 4.960 4.693 19.156 0.471 1.898 1.275 4.989

FIELD EXPERIMENT AND DATA These data were collected by a Fortune 500 manufacturer of durable, male-skewed products.2 The general objective of the research was to study consumer responses to a new Web site created to provide general product-category advice. The authors selected an entry in an expensive consumer-goods category where the purchase process tends to be highly involved. As most consumers consider several product options and extensively seek product information before making their purchase decisions, the authors believe that this product category was an appropriate category in which to study referral behavior. The data came from two sources—a large online survey and an online field experiment. The online survey was managed by an independent market-research firm that invited 50,000 online panel members to participate. Fully 14,216 panel members logged in, and 13,031 completed the survey, resulting in a response rate of 26 percent. All panel members were U.S. residents ages 18 and older. At the end of the survey, participants were asked whether they were interested in reviewing a beta-stage Web site. “Yes” was answered by 10,171, or 78 percent of respondents (See Table 1).
2

Education (scale) Opinion leadership

The Survey • Sample: In the matched sample of 3,219 participants, 40.6 percent were female, and 66.7 percent were married (See Table 1). Ages ranged from 18 to 89 years, with a mean of 48.2. Household income was divided into 11 levels, with a median income of $50,000 to 74,999. Education level was classified into seven levels, with the mean falling between “some college” and “college graduate.” • Procedure and Stimuli: Survey participants answered questions regarding WOM behavior and propensities and demographics. A six-question opinion leadership scale (King and Summers, 1976) was administered in reference to the “Internet” using five-point scales. • Survey Data Descriptive Statistics: The scale sum for opinion leadership ranged from 6 to 30, with sample distribution approximately Normal (See Table 1). The mean sum for the opinion leadership scale was 19.156 (SD = 4.989). The Field Experiment Invited participants who completed the survey were e-mailed an invitation to
3,219 invitees covered by these two conditions. Importantly, after comparing the matched sample of 3,219 subjects with the initial survey sample (N = 13,031), there are no significant differences in demographics or opinion leadership scores.

preview a new Web site. This Web site was noncommercial in nature in that it asked consumers to enter as much (or as little) information as they wished regarding their category preferences before they generated a list of recommended products. The list included products across all manufacturers in the category based on an algorithm that was validated by highly regarded industry sources. On the screen where consumers were presented with their list of recommended products, they also were offered an opportunity to “click here to e-mail a friend” (See Appendix 1). If a participant chose to click on the “refera-friend” link, a new window popped up, allowing the participant to enter up to five e-mail addresses. Participants could click on the link as many times as they wished.4 Generalization beyond two specific manipulation designs may introduce validity issues because the results likely vary by the level of the inducement and the functionality within manipulation classification. For example, the extrinsic manipulation included a $500 sweepstakes. Other possible versions of a tangible manipulation that could result in very different referral rates include a sweepstakes of different
4

3

The identity of the firm and category cannot be disclosed to maintain competitive confidentiality. 3 6,775 of the 10,171 were randomly selected and later e-mailed an invitation to the Web site; 400 (6 percent) of these e-mails bounced back, leaving 6,375 valid invitees. These 6,775 invitees were randomly assigned to four conditions in the initial research commissioned by the firm. As only two of the four conditions are relevant to this study, the authors focus exclusively on the relevant sample of

Here the analysis is limited to the decision on whether to make a referral. Given the small number of respondents who had made the referrals, the analysis on the number of referrals did not yield any new insights.

84

JOURNAL OF ADVERTISING RESEARCH

March 2014

MONEY TALKS … TO ONLINE OPINION LEADERS

prizes (e.g., a $250 or a $10,000 sweepstakes), or a discount on future purchase, which also could vary by level (e.g., 10 percent off or 50 percent off). As such, comparison among referral rates is limited to the versions and levels specified here. Future research should include sensitivity analyses to determine consumer thresholds and optimal amount of rewards. ANALYSIS AND RESULTS With the merged survey and experiment data, participants’ referral behaviors were matched with their personal profiles. Among the 3,219 valid observations, 67 of the participants made a referral, resulting in a mean referral rate of 2.08 percent— which was reasonable compared to an industry average of around 1.5 percent. The Power of a Cash Incentive: Money Talks Referral rates were substantially different under the two manipulation conditions: • In the “intrinsic-motivation” condition, there were virtually no referrals. Specifically, there were four referrals of 1,591—a 0.25 percent response rate. Thus, these 1,591 participants were not A possible reason for low intrinsic motivation is that the information contained in this Web site was not perceived as valuable, particularly because it was in beta-form. As a result, the expected gain from referral was judged to be too small to deserve referral efforts. • The “extrinsic reward” condition represented substantially more referrals. Specifically, 63 of 1,628 (3.87 percent) made referrals. Thus, the extrinsic rewards significantly increased referral propensity
5

(3.87 percent versus 0.25 percent)—a difference that is statistically significant at p < 0.001. Opinion Leadership As expected, the opinion-leadership scores were balanced between the two manipulation conditions. Thus, different referral propensities across the two manipulation conditions were not associated with different levels of opinion leadership. Specifically: • Mean opinion leadership scores were 19.127 for the intrinsic-motivation conditions; and • mean opinion leadership scores were 19.108 for the extrinsic-motivation conditions. To examine whether the referral rate varies with opinion leadership, the authors plotted the relationship for each of two manipulation conditions (See Figure 1). Sufficient

variance in opinion leadership scores existed under each manipulation condition. Specifically: • Under the “intrinsic-motivation” condition, the standard deviation was 5.213 for opinion leadership. Clearly there were a fair number of participants with high opinion-leadership scores. Nevertheless, the referral rate was very low under this condition among both opinion leaders and non–opinion leaders. Under this condition, there was no significant difference in referral rate between consumers with different opinion leadership scores. • By contrast, under the “extrinsic-reward” condition, the authors found a positive relationship between the referral rate and the opinion leadership score. As the opinion-leadership scores increased from six to 30, the referral rate rose from 2 percent to approximately 6 percent.

5

Intrinsic Motivation 0.06

Extrinsic Reward

Referral Rate

intrinsically motivated to make referrals.

0.04

0.02

0 0 10 20 30 0 10 20 30

Opinion Leadership Score

DMA (Direct Marketing Association) “Statistical Fact Book 2013,” p. 115.

Figure 1 Referral Rate vs. Opinion Leadership Score under Different Motivation Conditions (Intrinsic vs. Extrinsic)
March 2014

JOURNAL OF ADVERTISING RESEARCH

85

MONEY TALKS … TO ONLINE OPINION LEADERS

To better understand this relationship, the authors explored the data in two ways: • Among the 1,628 participants in the “extrinsic reward” condition were 63 referrers. The average opinionleadership score was 20.9 for referrers and 19.0 for non-referrers, indicating a clear difference (t-value = 3.04, p < 0.01). In other words, opinion leaders responded to extrinsic incentives more than non–opinion leaders. • The authors conducted a median-split analysis with median opinion leadership score at 19. The current study excludes median data points and computed referral propensity for opinion leaders (above median) and non–opinion leaders (below median).6 Results indicated that, under the extrinsic reward condition, the referral rate for opinion leaders (opinion leadership scores > 19) was 4.85 percent, whereas the referral rate for non–opinion leaders (opinion leadership scores < 19) was just 2.69 percent. The difference again was significant statistically (t = 2.31, p < 0.05). Thus, opinion leaders were more responsive to an extrinsic reward than non–opinion leaders. Interactions with Intrinsic and Extrinsic Manipulation To investigate the potential interaction between the opinion leadership scores and the intrinsic-versus-extrinsic manipulation conditions more formally (See Figure 1), the authors ran a linear regression analysis with respect to referral rates and summarized the results (See Table 2).
6

TABLE 2 Regression Analysis Results
Coefficient Education Age Married (=1 if married, =0 otherwise) Gender (=1 if female, =0 if male) Extrinsic Reward (=1 for extrinsic reward condition, =0 for intrinsic motivation condition) Opinion Leadership Score Extrinsic Reward × Gender* Extrinsic Reward × Age Extrinsic Reward × Married Extrinsic Reward × Education Extrinsic Reward × Opinion Leadership Score* Constant
Note: Sample size = 3219; * indicates statistical significance < 5 percent.

Std. Dev. 0.0028 0.0003 0.0078 0.0076 0.0349 0.0007 0.0106 0.0004 0.0110 0.0040 0.0010 0.0247

t-value 0.66 –0.09 0.52 –0.06 0.60 0.45 –2.39 –0.71 –0.24 –0.22 2.46 –0.68

0.0019 –0.00002 0.0041 –0.0005 0.0210 0.0003 –0.0254 –0.0003 –0.0026 –0.0009 0.0024 –0.0168

The current study found significant interaction between opinion leadership and the positive effect of extrinsic rewards (p-value < 5 percent) (See Table 2). Recall that the opinion leadership scores range from 6 to 30, with a median value of 19. This result indicates that—when the opinion leadership score increases by 10 units (e.g., from medium to the maximum)—the positive effect of extrinsic rewards on referral rate will increase by 10 × 0.0024 = 2.4 percent. Though the results seem to indicate a main effect of extrinsic reward (See Figure 1), the difference was not significant in the secondary data examination (See Table 2). This discrepancy may be attributable to scaling effect of the opinionleadership score (Irwin and McClelland, 2001). More specifically, since the opinionleadership scale has a minimum value of 6, the main (positive) effect of extrinsic reward on referral rate is captured by especially when the referral probability under one condition is close to zero (Berry, DeMeritt, and Esarey, 2010).

the interaction term. The authors have run separate regressions with mean-adjusted values of opinion leadership and found significant main effects (See Appendix 2). The current study also found that positive effect of extrinsic rewards on referral rates was much higher for male participants (See Figure 2). On average, the increase in referral rate induced by the extrinsic reward was higher for male participants by 2.52 percent. To further understand this difference, the authors examined the relationship between mean referral rates and opinion leadership scores for male and female participants separately. As the referral rate was virtually zero under the intrinsicmotivation condition, the analysis used only the participants under the extrinsic reward condition. The results clearly showed that male opinion leaders were very responsive to the extrinsic rewards. Though the referral rate went up to 10 percent for those male participants with the highest

7

A significant proportion of the opinion leader scores were at the median values (N = 113 or 8 percent). Results remain significant when including all of these observations as high or low. 7 The authors believe a linear regression model is a better way to test the interaction effect than a binary Logit model,

86

JOURNAL OF ADVERTISING RESEARCH

March 2014

MONEY TALKS … TO ONLINE OPINION LEADERS

referrals from opinion leaders versus non–
Male 0.08 Female

opinion leaders. Participants in the current study were invited to “click here to refer a friend” via e-mail. Future research should be conducted to determine whether these insights

Referral Rate

0.06

might be extrapolated across other online platforms such as Twitter and Facebook. Though the main finding that extrinsic rewards are equally effective across other

0.04

platforms, it is viable that different types of opinion leaders—“knowledge opinion leaders,” for instance, or “discussion opinion leaders”—who have different motivations (Kaiser, Kröckel, and Boden6 30 6 Opinion Leadership Score 30

0.02

dorf, 2013) may differentially self-select across the various platforms, thus affecting dissemination patterns. Future research also could • include a sensitivity analysis to determine the incremental referrals a firm can expect from both intrinsic and extrinsic rewards of various types and sizes and how other consumer variables such as altruism and involvement (controlling for opinion leadership) affect the pattern of intrinsically-versus-extrinsically motivated referrals; and • assess whether consumers who receive referrals assume that opinion leaders are intrinsically motivated to help, even when they were extrinsically rewarded. How do consumers perceive referrals from opinion leaders who receive small discounts? What about opinion leaders who receive free phones? Importantly, the use of extrinsic rewards also may affect the proportion of referrals that are ultimately converted into purchases and profits (Tuk, Verlegh, Smidts, and Wigboldus, 2009). The empirical results and conceptual model apply to one-time use of extrinsic rewards. The conclusions drawn also

Figure 2 Referral Rate vs. Opinion Leadership Scores for Participants of Different Gender (Male vs. Female) under the Extrinsic Reward Condition opinion-leadership scores, the relationship was much flatter for the female participants. This potentially was due to the fact that females were more sensitive to socialcapital loss. Another possible reason could be that that product category in the study was male-skewed and females simply may have had fewer social ties who would be interested in the product’s Web site. DISCUSSION AND CONCLUSION This paper investigated the relative effects of intrinsic versus extrinsic motivations for consumers to make Web-site referrals. Results indicated that the participants generally were not inspired to refer the Web site to their social networks based on intrinsic motivations. When offered more tangible, extrinsic rewards, however, referral rates were substantially higher. Importantly, the incremental referral rate stimulated by extrinsic rewards was significantly higher for opinion leaders; thus, “money talks.” The authors provided a theoretical interpretation (and a simple economic model; see Appendix 2) for these empirical findings. Not surprisingly, previous research had highlighted that consumers constantly seek to establish and maintain their identities, whether online through social media or elsewhere (Thompson, 2013). Building on this basic premise, it is conjectured that opinion leaders establish positive reputations based on their history of intrinsically motivated referrals. Positive reputations—or “social capital”—can shield opinion leaders from potential reputation loss when their social network referrals occasionally are rewarded with tangible, extrinsic incentives. By contrast, when non–opinion leaders’ referrals are associated with extrinsic rewards, these consumers may be perceived as purely selfish and therefore suffer a loss of social capital. Consequently, as demonstrated by the theoretical analysis, extrinsic rewards may stimulate more

March 2014

JOURNAL OF ADVERTISING RESEARCH

87

MONEY TALKS … TO ONLINE OPINION LEADERS

are intended to be descriptive in nature, not prescriptive. If marketers repeatedly employed extrinsic rewards to stimulate social network referrals, there may be negative long-term and strategic consequences: • The reputation effect shielding the opinion leaders may diminish. • Imagine a multiple-period setup where in each time period a consumer faces referral opportunities. If the vast majority of these referral opportunities are not associated with tangible rewards, the reputation effect for opinion leaders will hold. If a significant proportion of referral opportunities extrinsically are rewarded, however, the reputation effect for opinion leaders cannot be sustained. Once opinion leaders have received a sufficient number of extrinsic rewards, they no longer will enjoy the reputation of being intrinsically motivated category experts. • Competition among competing brands for opinion leaders’ attention may rise. Since, as this study demonstrates, extrinsic rewards can encourage the referral intensity of opinion, marketers may consider offering rewards to opinion leaders. This could lead to an incentive war and eventually a tragedy of the commons. As firms compete for the opinion leaders’ referrals with more valuable rewards, the cost of referrals will increase. • Greater availability of extrinsic rewards eventually could damage the reputation of opinion leaders and, ultimately, reduce referral rates. Of course, these issues (and the models) assume full transparency of extrinsic rewards. As a result of these reputational concerns, marketers and opinion leaders, alike,

may seek to not disclose extrinsic referral rewards, which could work to the detriment of both parties. • Falsely incentivizing—or bribing— opinion leaders for making referrals may raise other ethical concerns. The conceptual model (and the economics model in Appendix 2) show that extrinsic rewards may entice opinion leaders to refer products that they would otherwise not recommend. Opinion leaders may take advantage of their established reputation and recommend products with incentives instead of those products that best fit with others’ needs and preferences. This research offers other specific insights for marketers who are designing consumer referral campaigns: • The results substantiate and reinforce the significant role of opinion leaders in the marketplace. • Results show that the “over-justification” (Lepper et al., 1973) or “crowd-out” effects (Benabou and Tirole, 2003; Frey and Oberholzer-Gee, 1997; Kreps, 1997; Kamenica, 2012) may vary with the presence of motivation uncertainty and reputation effects. Much of the literature on the subject examined people who had high social capital but had not established a reputation for making intrinsically motivated referrals. For these people, extrinsic rewards could reduce referral propensity. A message to marketing managers: The adage that “explicitly rewarding consumers for referrals may backfire” is not necessarily true—supported by reasons that include consumers’ uncertainty regarding opinion leaders’ motivations and the effects of opinion leaders’ reputations.

• The conclusions in this paper also may explain an earlier finding that rewards are effective in increasing referral to weak ties but less so to strong ties (Ryu and Feick, 2007). Consumers may not be willing to sacrifice social capital with their strong ties, and, as such, managers could encourage consumers to make referrals and recommendations to their acquaintances (as opposed to their close friends). • Perhaps most importantly, the results of the current study suggest that campaigns targeted should but also not only be offer tangi-

ble, extrinsic referral rewards to opinion leaders.

menGze shi is an associate professor of marketing at Rotman School of Management, University of Toronto. His research is focused on influencing through incentives and relationship programs, with special investigation of loyalty rewards, sweepstakes, group-buying discounts, and direct mails. Dr. Shi’s work can be found in Marketing Science, Management Science, and Journal of Marketing Research, among other journals.

andrea woJniCKi is an independent marketing consultant in Toronto. Her research focuses on consumers’ motivations to generate word of mouth in both offline and online environments. Dr. Wojnicki’s academic and professional career includes a recent assistant professorship at the Rotman School of Management, University of Toronto, and various marketing positions at Kraft Foods, Toys R Us, and Ogilvy & Mather. Her work has been published in the Journal of Marketing.

ACKNOWLEDGMENT
The authors thank anonymous reviewers for offering many valuable ideas discussed in this last section.

88

JOURNAL OF ADVERTISING RESEARCH

March 2014

MONEY TALKS … TO ONLINE OPINION LEADERS

REFERENCES
B enABou , r., and J. t iroLe . “Intrinsic and Extrinsic Motivation.” Review of Economic Studies 70, 3 (2003): 489–520.

i yengAr , r., c. V An

Den

B uLte , and t. W.

L ePPer , m. r., D. g reene , and r. e. n isBett . “Undermining Children’s Intrinsic Interest With Extrinsic Rewards: A Test of the “Overjustification” Hypothesis.” Journal of Personality and Social Psychology 28 (1973): 129–137.

VALente. “Opinion Leadership and Social Contagion in New Product Diffusion.” Marketing Science 30, 2 (2011): 195–212. J AcoBy , J., and W. D. h oyer . “What if B erry , W., J. D e m eritt , and J. e sArey . “Testing for Interaction in Binary Logit and Probit Models: Is a Product Term Essential?” American Journal of Political Science 54, 1 (2010): 248–266. c hiLDers , t. L. “Assessment of Psychometric Properties of an Opinion Leadership Scale.” Journal of Marketing Research 23, 2 (1986): 184–188. Frey, B. s., and F. oBerhoLzer-gee. “The Cost of Price Incentives: An Empirical Analysis of Motivation Crowding-Out.” American Economic Review 87, 4 (1997): 746–755. g Atignon , h., and t. s. r oBertson . “An Exchange Theory Model of Interpersonal Communication.” Advances in Consumer Research 13 (1986): 534–538. g oDes , D., D. m AyzLin , y. c hen , s. D As , c. D eLLArocAs , et al .

Opinion Leaders Didn’t Know More? A Question of Nomological Validity.” In Advances in Consumer Research, Kent B. Monroe, ed., vol. 8. Ann Arbor, MI: Association for Consumer Research, 1981. k ADushin , c. “Too Much Investment in Social Capital?” Social Networks (2004): 1–16. kAiser, c., J. krÖckeL, and F. BoDenDorF. “Simulating the Spread of Opinions in Online Social Networks when Targeting Opinion Leaders.” Information System E-Business Management 11 (2013): 597–621. kAmenicA, e. “Behavioral Economics and Psychology of Incentives.” Annual Review of Economics 4 (2012): 427–452. k ing , c. W., and J. o. s ummers . “Overlap of Opinion Leadership Across Product Categories.” Journal of Marketing Research 7 (1976): 43–50.

L in , n., and B. e rickson . “Theory, Measurement, and the Research Enterprise on Social Capital.” In Social Capital: An International Research Program, N. Lin and B. Erickson, eds. Oxford, UK: University Press, 2008. Lu, y., k. JerAth, and P. V. singh. “The Emergence of Opinion Leaders in Online Review Communities.” Management Science 59 (8): (2013): 1783–1799. m oynihAn , r. “Key Opinion Leaders: Independent Experts or Drug Representatives in Disguise?” BMJ 336, June 21 (2008): 1402–1403. murDock, k. “Intrinsic Motivation and Optimal Incentive Contracts.” Rand Journal of Economics 33, 1 (2002): 650–671. r yu , g., and L. F eick . “A Penny for Your Thoughts: Referral Reward Programs and Referral Likelihood.” Journal of Marketing 71, January (2007): 84–94. thomPson, c. J. “The Politics of Consumer Iden-

“The Firm’s Management

of Social Interactions.” Marketing Letters 6, 3/4 (2005): 415–428. h ennig -t hurAu , t., k. g Winner , g. W ALsh , and D. g remLer . “Electronic Word-of-Mouth via Consumer-Opinion Platforms: What Motivates Consumers to Articulate Themselves on the Internet?” Journal of Interactive Marketing 18, 1 (2004): 38–52. irWin, J. r., and g. h. mccLeLLAnD. “Misleading Heuristics and Moderated Multiple Regression Models.” Journal of Marketing Research 38, 1 (2001): 100–109.

kozinets, r., k.

De

VALck, A. WoJnicki, and s.

tity Work.” Journal of Consumer Research (2013): DOI:10.1086/673381. t uk , m.A., P. V erLegh , A. s miDts , and D. h. J. WigBoLDus. “Sales and Sincerity: The Role of Relational Framing in Word-of-Mouth Marketing.” Journal of Consumer Psychology 19, 1 (2009): 38–47. WoJnicki, A. c., and D. B. goDes. “Signalling

W iLner . “Networked Narratives: Understanding Word-of-Mouth Marketing in Online Communities.” Journal of Marketing 74, March (2010): 71–89. k rePs , D. “Intrinsic Motivation and Extrinsic Incentives.” American Economic Review 87, 2 (1997): 359–364. k r u g L A n s k i , A. W. “The Endogenous-

Success: Strategically Positive Word-of-Mouth.” Working Paper. Robert H. Smith School of Business, University of Maryland, 2011.

Exogenous Partition in Attribution Theory.” Psychological Review 83 (1975): 387–406.

March 2014

JOURNAL OF ADVERTISING RESEARCH

89

MONEY TALKS … TO ONLINE OPINION LEADERS

APPENDIX 1 Field Experiment Options
TWO REFER-A-FRIEND MANIPULATIONS Intrinsic: “E-mail your friends a link to this Web site to help them make their next ‘product’ purchase decision.” Extrinsic: “E-mail your friend a link to this Web site and your name will be entered into a $500 cash sweepstakes.”

APPENDIX 2 An Economic Model of Referral with Intrinsic and Extrinsic Motivations
Consider a consumer, denoted by i, who allocates the total endowed time T between making referrals and engaging in other utility production activities. Denote the consumer’s referral decision by rij where rij = 1 if consumer i makes a referral of product j and rij = 0 otherwise. If the consumer makes a referral, it costs the consumer tij units of time, leaving zi = T – tij amount of time for other activities. The consumer’s referral utility is separable from other utilities; that is, consumer i’s total utility is uij(rij) + ψizi, where ψi is the (constant) marginal utility from other activities. The consumer’s referral decision can be summarized as follows: max uij(rij) + ψizi rij , zi

(1)

s.t. tijrij + zi = T, rij ∈{0,1} and zi ≥ 0. Substitute zi = T – tij rij into the objective function and rewrite problem (1) as rij ∈{0 ,1}

max uij(rij) + ψi (T – tij × rij)

(2)

Assume uij (0) = 0. Then (2) indicates consumer i should refer product j if and only if uij(rij = 1) ≥ ψitij Now assume the following utility function to represent the referral motivations: uij(rij = 1) = hi(Rj) + gi(SCij) (4) (3)

where Rj is the extrinsic reward offered by project j and SCij refers to social capital gains from making referrals of product j. The term SCij captures all the non-monetary and psychological value that consumer i derives from enhanced social relations and reputation associated with making referrals of product j. Let hi(Rij) > 0 and gi(SCij) > 0. Combining equations (3) and (4): consumer i makes a referral of product j if and only if the following condition is satisfied: hi(Rj) + gi(SCij)> ψi tij (5)

A consumer’s social capital change depends on the perceived motivation behind the referral. Specifically, when a referral of product j from consumer i is perceived to be intrinsically motivated, then consumer i gains social capital, i.e., SCij > 0. The relative amount of gain in social capital depends on the value of information. On the other hand, when a referral from consumer i is perceived to be induced by extrinsic rewards, then consumer i will suffer a loss of social capital, i.e., SCij < 0. To summarize the above discussion, r η j if consumer i is intrinsically motivated to refer product j  SCij =   −λ if consumer i is motivated by an external reward to refer product j (6)

90

JOURNAL OF ADVERTISING RESEARCH

March 2014

MONEY TALKS … TO ONLINE OPINION LEADERS

When consumer i does not receive a reward for making a referral, the receiver infers that the referral was intrinsically motivated, and consumer i gains social capital by ηj. However, when consumer i receives a reward for making a referral, the referral motivation becomes uncertain to the receiver. When the frequency of intrinsically motivated referrals that have been made by consumer i in the past is high, consumer i’s current referral is more likely to be perceived to also be intrinsically motivated. Specifically, let nr,i and nI,i denote the number of reward-induced and intrinsically-motivated referrals consumer i has made in the past. Then, when consumer i receives a reward from making a referral, the referral will be perceived as reward-induced with probability of (1 + nr,i)/(nI,i + nr,i + 1) and as intrinsicallymotivated with probability of nI,i/(nr,i + nI,i + 1). Incorporating above discussions into (6) leads to equation (7). if R j = 0 η j ,  SCij =  nI , i nr , i  n + n + 1 η j − n + n + 1 λ , if R j > 0 r ,i I ,i r ,i  I ,i

(7)

Let opinion leadership (denoted by OLi) represent the consumer’s past intensity of making intrinsically-motivated referrals. Following standard opinion leadership scales, required time tij should decrease with OLi, and social capital should increase with OLi. Equation (7) then implies that (∂SCij(Rj > 0)/∂OLi) > 0, i.e., referrals from opinion leaders are more likely to be perceived as intrinsic. When consumer i faces an extrinsic reward Rj, the consumer’s incentive to make a referral is boosted by hi(Rj). Simultaneously, the consumer will anticipate the potential loss of social capital. Based on equations (5) and (7), the net effect of the extrinsic reward is hi(Ri) + gi(SCij|Rj > 0) – gi(SCij|Rj = 0)   nI , i nr , i ηj − λ  – gi(ηj). = hi(Ri) + gi   nI , i + nr , i + 1 nI , i + nr , i + 1    (8)

Thus, extrinsic rewards will increase a consumer’s referral rate if and only if the consumer’s value of extrinsic reward, hi(Ri), exceeds the expected loss of social capital, gi(SCij|Rj>0) – gi(ηij). The net effect clearly depends on opinion leadership because the incremental loss of social capital due to receiving an extrinsic reward is smaller for opinion leaders (with a larger nI,i) than for non–opinion leaders. Since opinion leaders are expected to suffer a smaller amount of social capital loss, they are more likely to respond to external rewards.

March 2014

JOURNAL OF ADVERTISING RESEARCH

91

Copyright of Journal of Advertising Research is the property of Warc LTD and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.