中文摘要：

在营销实践中,多位代言人同时出现在一则广告中的组合代言广告日益增多,却很少有研究关注。本文探究了组合代言中代言人关系类型（合作型组合代言vs.冲突型组合代言）对广告效果的影响。研究发现,相对于合作型组合代言,冲突型组合代言会产生更好的广告效果;这种作用是通过消费者感知到的广告生动性所中介的;消费者的情感强度可以调节这种作用,对于情感强度较高的消费者,冲突型组合代言能产生更好的广告效果,而对于情感强度较低的消费者没有显著影响;此外,代言人之间的差异程度也可以起到调节作用,即当消费者对组合代言中代言人们的评价相差较小时,冲突型组合代言能提升广告效果,当对代言人们的评价相差较大时,则没有显著的促进作用。文章还对这些发现的理论贡献、营销启示和未来研究方向进行了讨论。

英文摘要：

Marketing practice has recently seen an increasing number of ads featuring more than one celebrity endorser in one ad. Although many studies have been conducted to investigate the effect of endorsement, little research attention has been given to group endorsement and hardly any studies have focused on the relationships between endorsers （cooperative vs. conflicting） in one ad. Building on the studies about advertising effectiveness and conflict, the current research proposes that the conflicting group endorsement brought stronger advertising effectiveness than the cooperative one. The mediating effect of consumers＇ perceived vividness of the ad and the moderating effects of consumers＇ affect intensity and the degree of discrepancy between endorsers were also examined. Two pretests and three main experiments were conducted to test the proposed hypotheses. Following the methods in previous studies, three products matching up with the two endorsers used in this study were screened in Pretest 1. The validity of the manipulation of endorsers＇ relationship types were verified in Pretest 2. In Experiment 1, a one factor （endorsers＇ relationship type： cooperative vs. conflicting） between subjects design with 99 participants was adopted to examine the effect of endorsers＇ relationship types on advertising effectiveness and the mediating effect of consumers＇ perceived vividness. The descriptions of product, endorsers and their relationship types verified in the pretests were used and advertising effectiveness and perceived vividness were measured by established scales. Following a similar procedure, a 2 （endorsers＇ relationship type： cooperative vs. conflicting） ~ 2（affect intensity： high and low） between subjects design with 91 participants was adopted to examine the moderating effect of consumers＇ affect intensity in Experiment 2. The participants＇ affect intensity were measured by established scales and divided into two groups by the median. In Experiment 3, a 2 （endorsers＇