763 Advances in Consumer Research Volume 36, © 2009 What Do People Talk About in Word-of-Mouth Communications? Sema Barlas, McGill University, Canada Lei Huang, Dalhousie University, Canada 1 1 This article is based on the second author’s dissertation research and is funded by a grant provided by the Social Sciences and Humanities Research Council of Canada. Authors are grateful for the comments of Joshua Klayman, Jackie Gnepp, Ashesh Mukherjee, Ulf Bockenholt, and Renaud Legoux on earlier ver- sions of this article. EXTENDED ABSTRACT Consumers perceive Word-of-Mouth information (WOM) as more credible than marketer-originated information (Bickart and Schindler 2001). Consequently, extant research focused on the antecedents and outcomes of WOM, and the question of what information consumers exchange with others (depending on “why”, “about what”, and “to whom” they are talking) has not received much research attention. It has been commonly assumed that people engage in WOM to improve quality of decisions, and thus, exchange diagnostic information to reduce their uncertainty with respect to offerings (Herr, Kardes and Kim 1991). Recent research indicates that people also engage in WOM to establish social connections (Sassenberg 2002). This paper aims to demonstrate that the social bonding goal increases the value of engaging market- ing information in WOM. In our context, information is considered as socially engaging whenever it spurs shared emotions among communicators (e.g., funny or interesting information). More im- portantly, we also demonstrate that social-bonding efforts in WOM depend on “to whom” (e.g., friends vs. acquaintances) and “about what” people could talk (e.g., nature of marketing information). According to our theoretical framework, people make trade- offs between different communication goals, depending on the characteristics of available topics and relationships among the communicators. Availability of highly engaging topics stimulates the social- bonding goal, and availability of relatively significant incentives resulting from attending to informative topics, facilitates the uncertainty reduction goal. Similarly, establishing social bonds via talking about engaging marketing information might be more prevalent among people with weak social-ties than among people with strong-social ties, especially when the given information is not about product benefits, and thus, does not trigger conversations about intimate consumption experiences (e.g., a funny joke in an advertisement). That is, impersonal marketing information might be a good conversation topic among strangers who lack other common topics. Lastly, we assume that goals influence selection of particular topics for conversations; therefore, observed bias for engaging information in social settings is an indication of the social bonding goal. More specifically, we predict that: H1: Engaging information becomes significantly more valu- able when it is consumed socially than when it is con- sumed individually. H2: Acquaintances that are in need of social connections are more likely to promote the social bonding goal via engaging conversations about marketing information than other goals. H3: The choice of engaging option increases over time during the WOM episode and conversations at early stages of a WOM episode exert disproportional influence on later choices. We applied an experimental paradigm in which participants were asked to choose between furniture and personal ads for proofreading to earn between $5 and $9, depending on the number of misspelled words they could identify over ten trials. Furniture and personal ads had, respectively, seven and three misspelled words. The task required participants to make a tradeoff between engaging conversations with others and their individual earning/ savings as in consumer choice between discount ads and humorous ads. Furthermore, although participants were told in advance that furniture ads had more spelling mistakes and that personal ads were more interesting, they did not know the exact number of spelling errors; nor did they know how much more interesting personal ads were relative to furniture ads. Finally, there were two independent conditions in the study: subjects made their choices either privately or together with three other participants, in which case they could talk about the ads. All participants distributed 100 points between the spelling-error rate and the content to reflect the importance of these attributes in their choice of ads at the end of the study and indicated the number of people they were friends with in their group. All conversations were audio-taped for further content analysis. Data support our predictions (p<.05). In general, engaging information became significantly more valuable when it was con- sumed socially than when it was consumed individually (H1) and the advantage of engaging information in social setting increased in conversations with weak ties (H2). More specifically, the impor- tance of the content in making choices was significantly larger in the WOM condition than in the individual condition, suggesting that WOM participants intentionally shifted their goals in the direction of the social bonding goal. The intentional shift was larger for acquaintances than for friends. Also, WOM participants talked about the most relevant information for decisions, the spelling-error rate/money, only 23% of the time despite the fact that the impor- tance of the spelling-error rate in making their choices was 42%. Note that the importance of money in the individual condition was 68%. That is, WOM groups deliberately switched their interest or preferences in favor of content, but they conversed about money even less than what would be expected given the intentional shifts in goals. In contrast to the conversations about money, intentional shifts determined the frequency of conversations about personal ads to a great extent; participants talked about personal advertise- ments 60% of the time and the importance of the content was 58% (32% for individuals); however, conversations systematically di- verged from the intentional shifts in goals depending on the social ties. In particular, exclusively-acquaintance (mean=0.74) and ex- clusively-friend (mean=0.66) groups talked about the personal advertisements significantly more than the mixed groups (mean=0.56) did. Also, exclusively-friends groups talked about personal advertisements less than they intended to whereas exclu- sively-acquaintance groups talked about personal advertisements more than they intended to. Furthermore, WOM participants chose personal ads more than they intended to, compared with individu- als. While the individual goal determined the initial choices, later choices were governed by the goal of the group and the nature of the conversations during the early stages of communications (H3). Overall, our results suggest that WOM participants herd into engaging options disproportionately more with conversations over time in comparison to the individual decision-makers. Further-