致谢
二十载求学路将尽，行文至此，落笔为终。求学之路始于家乡，而今将终于
徐州，一路行之如人饮水，冷暖自知。硕士学习生涯始于2019年秋，终于2022
年夏。三年青春打马而过，三千往事浮现眼前，感慨万千。
落其实者思其树，学其成时念吾师。二十岁桃李年华之际，有幸拜入江红艳
老师门下，得恩师传道授业解惑，更习得温良恭俭让等品行，无以为报，揖礼还
授。而今又受恩师江红艳点拨指导之恩，方能作此文。经师易遇，人师难遇，微
微寸心难报之。谢恩师多年来对吾眷顾。师之斥，忆之皆是对吾之策、劝勉。
借此片纸,聊表谢忱。感谢江老师对论文付出的巨大努力以及帮助，感谢江
老师能够同意让我再去完成最后的实验。感激涕零，无以言表。还要感谢同门师
姐、师兄和师妹们对我撰写毕业论文的帮助，在组会上对论文提出的宝贵意见以
及建议以及冯乐师兄在数据收集以及数据处理方面给我提供的巨大帮助和感谢
王璐师姐对修改实验刺激材料的帮助。同时，还要特别感谢杨浙帅老师，对论文
各个方面提出的宝贵意见，对我完成这篇论文有很大的帮助。此外，还要感谢预
答辩的老师，对论文提出的宝贵意见以及建议。是大家的努力和帮助，让我能够
顺利完成我的毕业论文。
树高千尺，根深沃土。感谢父母对我的支持，二十余载的疼爱与教育，源源
不断的信任与鼓励，是他们无尽的关怀与无条件的支持才能使得我能够顺利的完
成我的学业。
最后，祝愿大家都能远离疫情，平平安安，快快乐乐。也祝愿国家能够早日
摆脱疫情造成的影响和困扰，福泽披华夏，安平绕寰宇。
山水相逢，终有一别，大学路一号——中国矿业大学，后会有期。
摘要
近年来随着移动设备的普及和发展，消费者能够越来越方便地以在线评论的
形式为其他消费者分享其关于产品、服务和品牌的经验、观点以及反馈。在线评
论作为一种新型的口碑形式，在全球消费者中变得越来越流行。然而在线评论数
量的快速增长带来了信息过载、内容参差不齐等问题，识别有用的在线评论更能
使消费者和企业受益。以往的研究较少关注在线评论中比较信息与非比较信息之
间的差异。鉴于此，本研究将企业生成内容（FGC）中的比较广告的概念引入到
用户生成内容（UGC）下的在线评论中，较新地提出比较型评论的概念，并探究
评论者先前经验（有 vs.无）如何调节比较型评论（是 vs.非）对在线评论有用性
的影响及其潜在作用机制。
本文综合利用实验法和二手数据分析方法，通过三个研究对上述研究问题加
以验证。研究一结果发现：当评论提及评论者先前经验时，相对于非比较型评论，
消费者认为比较型评论更有用；当评论没有提及评论者先前经验时，消费者认为
比较型评论与非比较型评论的有用性之间无显著差异。研究二在研究一的基础上
进一步验证感知可信度在评论者先前经验和比较型评论的交互作用对在线评论
有用性的影响中发挥中介作用，并通过扩展实验刺激材料的产品品类与评论效价
以提高研究的外部效度。研究二结果发现：当评论提及评论者先前经验时，相对
于非比较型评论，消费者认为比较型评论更可信，进而认为评论更有用；当评论
没有提及评论者先前经验时，消费者认为比较型评论与非比较型评论之间的可信
度和有用性无显著差异。进一步，研究三通过对 Amazon上的真实评论数据
进行分析，再次验证前面研究的结论，增强了本论文研究结论的普适性和可靠性。
本文基于精细加工可能性模型和语言期望理论的视角辨析了比较型评论（是
vs.非）对在线评论有用性发挥影响的边界作用条件及其内在作用机制，整合了不
同文献证据之间的分歧，具有重要的理论贡献。同时，本文研究结论对在线零售
商和在线购物平台如何筛选有用的在线评论，以及社交媒体上个人品牌的塑造等
方面具有一定的营销实践启示。
该论文有图 6幅，表 21个，参考文献 224篇。
关键词：比较型评论；在线评论有用性；评论者先前经验；感知可信度
I
Abstract
In recent years, with the popularity and development of mobile devices, consumers
can more and more conveniently share their experience, views and feedback on
products, services and brands for other consumers in the form of online comments. As
a new form of word-of-mouth, online reviews are becoming more and more popular
among consumers all over the world. However, the rapid growth of the number of online
reviews has brought problems such as information overload and uneven content.
Identifying helpful online reviews can bring benefits to consumers and enterprises.
Previous studies have paidless attention to the difference betweencomparative
information and non comparative information in online reviews. In view of this, by
introducing the construct of comparative advertising in enterprise generated content
(FGC) into online reviews under the context of user generated content (UGC), this
research puts forward the concept of comparative review,and explores how the
reviewer’s previous experience (yes vs.no) moderate the impact comparative review
(yes vs.no) on online review helpfulness and its potential mechanism.
This thesis comprehensively adopts the experimental method and secondary data
analysis method to verify the above research questiones through three studies. The
results of study 1 showed that when the review mentioned the reviewer’s previous
experience, consumers perceived comparative review were more helpful than non-
comparative review; Whenthe review did notmention the reviewer’sprevious
experience,there wasnosignificantdifferenceon perceivedhelpfulbetween
comparative review and non-comparative review. On the basis of study 1, study 2
further verified that perceived credibility played a mediating role in the interaction
between reviewer’s previous experience and comparative review on onlie review
helpfulness, and expanded the product categories and review valence of experimental
stimulus materials to improve the external validity. The results of study 2 showed that
when the review mentioned the reviewer’s previous experience, compared with the non-
comparative review, consumers perceived the comparative review was more credible
and thus the review was more helpful; When the review did not mention reviewer’s
previous experience, there was no significant difference in reliability and helpfulness
between comparative review and non-comparative review. Further, study 3 verified the
conclusions of the previous research by analyzing the real review data on Amazon,
which enhanced the universality and reliability of the conclusions of this thesis.
From the perspective of the Elaboration Likelihood Model and the Language
II