網際網路是繼電子計算機發明後，帶給人類最大衝擊的一項革命，它突破了地理的籓籬，資訊的流動可以在極短暫的時間內跨越國界，這是其他媒體所望塵莫及的。而這種傳播特性最為人所詬病的問題之一，就是網路謠言的充斥。透過E-mail、BBS、NEWS、WWW等各種不同的管道散播出來的謠言，不僅可能造成個人的名譽受損、企業的形象破壞，許多的犯罪手法也因而產生。就企業主而言，網路謠言的殺傷力莫過於對公司形象與品牌聲譽的損害，而如何防範與澄清消息的正確性，已成為企業公關與危機管理的重要一環。 本研究則分析現有的謠言與消息來源理論，探討謠言的結構、傳播過程以及其社會心理成因等，並比較電腦中介傳播 (CMC) 與一般謠言口耳相傳 (FtF) 之下的謠言特性，以及探討傳統大眾媒體與網路媒介傳播謠言效果之差異，歸納出影響謠言傳播的可能因素包括謠言的可相信度、渾沌不安、訊息特性、涉入度、網路訊息惰性、人口統計變數與網路使用行為等。利用這些可能的變數，本研究設計兩組實驗來衡量謠言散播與闢謠效果，實驗一利用可相信度與渾沌不安兩個變數操弄謠言，並以小樣本的散播實驗來衡量謠言操弄的結果，探討謠言散播模式與影響變數，並以計數器追溯謠言散播的情形，來了解謠言散播的真實面貌；實驗二則以類似的實驗方法，選擇一則已發生的網路謠言案例來研究謠言散佈模型，並以發送闢謠信件的方式來探討其謠言防治效果。 實驗一在經網路謠言散播實驗之後，發現操弄變數的差異並不顯著，故實驗一改採Regression檢定，最後發現可相信度、訊息特性與渾沌不安是影響謠言散播的主要因素，網路訊息惰性則對謠言散播有間接影響效果，而可相信度有部分的中介效果；在電子郵件闢謠實驗方面，實驗二發現可相信度與網路訊息惰性對訊息的散播有顯著影響效果，而可相信度有部分的中介效果。由此可知，兩個實驗的散佈模式並不相同，其原因有可能是因為涉入度的差異。另外，兩個實驗皆發現經由電子郵件散播的訊息符合創新採用理論，其中實驗一的謠言與實驗二的闢謠訊息為Rogers所說的不完全散佈模式，而實驗二的謠言則為所謂的「S」形創新散佈曲線。 此外，本研究利用謠言的追蹤計數，得到了散播曲線與訊息散播方程式，並估計而求出了謠言的散播速度。由追溯的結果得知，網路謠言傳播的速度十分快速，尤其實驗二的謠言在第2.2日時就達擴散曲線的反曲點，一星期內散佈的人數高達原來樣本數的兩倍；相反的，不被眾人散播的謠言可以在短短幾天之內就趨於平息。而本研究也對網路訊息的「散播者」與「非散播者」作出比較，發現散播者的散佈可能性、對謠言的相信度、渾沌不安等都比非散播者高，但網路訊息惰性則較低。在粗淺的瞭解這群可能的網路「意見領袖」之後，企業界可以專注於這群人作有效的闢謠公關活動，讓網路謠言的管理更有效率。 最後，本研究彙整謠言的應對策略，並對企業因應謠言的危機管理的模式，提出許多的方案與建議，讓企業界在面對網路謠言時，能有參考的依循，並將損害減至最低。 / Not only has the internet increased the efficiency of communication, it has also brought out severe crises in the real world. Messages can be circulated all over the world in just seconds. As a result, geographical barriers have been virtually broken down. However, while rumors and hoaxes can easily be distributed through e-mail, BBS, news group, and so on, such false information may cause severe damage to individuals and corporations concerned, even trigger criminal offenses. How to prevent potential damage of commercial reputations which may cause by internet rumors and clarify the existing grapevines is one of the most important issues each corporation needs to address in the present information society. This thesis is inclined to find out the differences between Computer Mediated Communication (CMC) and Face-to-Face (FtF) Communication in terms of the variables which affect the diffusion of rumors, i.e. Credibility, Anxie-uncertainty, Message Property, Involvement, Internet Inertia, demographical variables, and personal internet behavior. Two experiments are designed in this research. Exp.1 operates Credibility versus Anxie-uncertainty into four rumors and illustrates the correlation between Diffusion Possibility and each of the aforesaid independent variables which can facilitate establishing the diffusion model for Internet Rumors accordingly. Exp.2 is an empirical study based on the diffusion model built up in Exp.1. A pre-selected existing rumor is used as the research sample. It remodels the rumor diffusion and measures the clarification effect through e-mail as well. Moreover, both experiments use Site Meter to trace the mechanism of rumor diffusion. Numbers of the rumor being viewed and its responses and clarifications can be easily counted. According to experiments, Credibility and Anxie-uncertainty are originally well operated in pretests. However, the manipulated variables are not fully distinguished in Exp.1. Hence, Regression is employed to test the hypotheses and as a result, Credibility, Anxie-uncertainty, and Message Property are the key variables that directly influence rumor diffusion in Exp.1 while Internet Inertia has indirect effect on the outspread of a rumor. In Exp.2, both Credibility and Internet Inertia play important roles in the diffusion behavior while Credibility acts as a partial mediating variable in this case. Additionally, this research also shows the possibility and degree that Involvement might be the factor which causes discrepancy between these 2 models. However, it needs to be confirmed by further research. Moreover, both models fit the Distribution Theory of Innovations developed by Rogers. Rumors in Exp.1 and clarify message in Exp.2 are both Unsuccessful Diffusion while rumor in Exp. 2 belongs to S-shaped Normal Diffusion. Incidentally, diffusion curves and the function of rumor diffusion are obtained to work out through simulation equations and so are the diffusion rates, which surprised us for their rapid distribution. The inverse point was reached in the 2.2 days after the rumor message was distributed in Exp.2, and almost two folds of the original group numbers were reached in just one week. On the other hand, rumors which were not circulated would be quelled in few days. Comparisons between the Distributor and the Non-distributor in the Internet society are also made in this study in facilitating the management of corporate grapevines. The Distributor believes rumors and are more aggressive in rumor distribution and shows higher Anxie-uncertainty, while their Internet Inertia are lower than Non-distributor. Knowing more about the so-called Internet “Opinion Leader” helps corporations focus on crisis management by initiating effective public activities not only because they are cores of grapevines but are also easier to approach than others. Finally, various corporate strategies of confining rumors are developed in the thesis. Suggestions and alternatives are provided in conclusion. Corporate administrators can therefore take them for reference when dealing with rumor-caused crises.