Abstract: The anonymity of the Internet, the difficulty of identifying shills, and the large number of Internet users make it worthwhile to explore the influence of the forum objectivity on WOM recipients’ attitudes, the boundary conditions of the influence, and the underlying mechanism. Based on the research on objectivity of individuals as information sources (Williams et al., 2004), objectivity of forums is defined as consumers’ overall opinion of the objectivity of all threads in the forum.
Four studies were conducted. With 140 Internet surfers participanting in study 1, we employed a 2 (objectivity of forums: objective versus not) × 2 (brand strength: strong versus weak) between-subjects design. In the “objective” condition, participants were told Haizhu wanted to buy an MP4, and he or she saw a thread on the exact MP4 in a large forum which he or she usually surf. Participants were also told that Haizhu usually saw some followers pointed out the initiator was a shill on the forum in the “not objective” condition. We describe the strong brand as one of the famous brands with high reputaion and the relatively weak brand as a not famous brand. All the participants saw the same thread recommending the MP4. The following three studies are similar to study 1. With 78 Internet surfers participating in study 2, we employed a 2 (objectivity of forums) × 2 (prior brand satisfaction: very satisfied versus not) between-subjects design. We told the two groups of participants Haizhu had been very satisfied or a little dissatisfied with the old model of the brand respectively. With 94 undergraduate students participating in study 3, we employed a 2 (objectivity of forums) × 2 (familiarity between the followers and initiator: familiar versus not) between-subjects design. In the “familiar” condition, 3 followers pointed out they are familiar with the initiator in their replies. In the “not familiar” condition, no follower did so. With 80 Internet surfers participating in study 4, we employed a 2(objectivity of forums)×2(consensus among followers and initiator: unanimous versus not) between-subjects design. In the “unanimous” condition, all followers agreed with the initiator. In the “not unanimous” condition, 3 followers did not agree with the initiator. We measured WOM recipients’ attitudes (Cronbach’s αs>0.82) and persuasion knowledge (Cronbach’s α=0.92). Prior work has used the scales (Kirmani & Zhu, 2007).
All manipulations were successful (ts > 2.11, ps < 0.04). ANOVA models were used. The main and interaction effects are significant in study 1 (Fs (1, 136)≥3.86, ps≤0.05) and study 2 (Fs (1, 74)≥3.02, ps≤0.09). The main effect of objectivity of forums and interaction effect are significant in study 3 (Fs (1, 90)>4.26, ps<0.05). The interaction effect is significant in study 4 (F (1, 76)=3.50, p<0.07). Recipients’ attitude for the product recommended by the thread is better when forum is objective than when it is not. The effect is mediated by persuasion knowledge. When the brand is relatively weak, when the reader was a little dissatisfied with the brand, when followers are not familiar with the initiator, or when the attitudes of followers and the initiator are not unanimous, the effect is prominent.
These findings enhance our understanding of the antecedents of online WOM influence and persuasion knowledge. Companies could apply our findings to improve online WOM communication.

Key words: online WOM, forum objectivity, persuasion knowledge, brand strength