Content-Type: text/html The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election Communication Technology& Policy Division The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election Tai-Li Wang Assistant Professor 011-886-2-27211109, or 011-886-2-27724713 [log in to unmask] or [log in to unmask] Department of Radio, TV & Film Shih-Hsin University #1, Lane 17, Sec.1, Mu-Cha Rd., Taipei 116, Taiwan, R.O.C. Submitted to the Communication Technology & Policy Division of the Association for Education in Journalism and Mass Communication at its annual convention in Washington, D.C., August, 2001. The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election Abstract Taiwan's 2000 presidential election is a milestone for the Internet to be a campaign medium in a nation-wide election. By conducting field experiments during the campaign period, this study intended to find out the effects of web interactivity on campaign communication. It is assumed that, by adding human-to-online media or human-to-human interaction, the more interactive campaign websites will generate more campaigning effects. Results showed that the interactive level of a candidate website might not linearly contribute to positive web campaigning effects. In some cases, moderate interactivity may produce better effects than higher interactivity. In other cases, it may be voters' political interest in election campaigns that predominantly determine online campaigning effects. The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election A Milestone for the Internet as a Campaign Medium in Taiwan From "Citytalks" in Amsterdam (Schalken & Tops, 1995) to the "Democracy Networks" in the United States (Dutton, 1996), there has emerged a global trend to conduct field research to determine whether the uses of the Internet will improve political dialogues between politicians and citizens. The influence of web campaigning, particularly, is the focus of the attention. Such a trend is rapidly spreading among several democratic societies in Asia, which enjoys the fastest growth of the Internet around the world. By the end of the millennium, the Internet population in Taiwan mounted to six million, which composed over one quarter of the whole population (23 million). In terms of both the Internet population and the density of Internet usage, Taiwan was listed in the Global Top Fifteen (The Ministry of Economics, 1999). According to a recent survey (NetValue), Demark has the highest percentage of Internet households in its population, trailing by the United States, Singapore, and Taiwan. The Republic of China on Taiwan is, therefore, considered to be an adequate society in Asia to explore the potential of the Internet in politics. The 2000 presidential election in Taiwan provided a timely opportunity for testing the development of the web as a campaign medium. This election was a nerve-racking campaign since the very beginning, for the three leading candidates were running so close that whoever won would have to squeak home by only the narrowest of margins. There were thousands of voters who could not deal with the uncertainty and fear. Psychiatric wards reported a sharp rise in depression and anxiety as being diagnosed related to the election. Life Lines (a phone-counseling service) was inundated with calls from voters who were paralyzed by indecision. Perhaps no Taiwanese election has ever been so bitterly contested. The outcome of the election dramatically ended the longtime rule of the Koumingtang in Taiwan. The candidate of the opposition party (the Democratic Progressive Party or the DPP), Chen Shui-Bian, won this tight battle. At a massive celebration, Chen claimed that a new stage in Taiwan's democracy had begun. Undoubtedly, the 2000 presidential campaign is unique in Taiwanese history. However, it is equally significant in another respect: It is a milestone for the Internet to be a campaign medium in a nation-wide election. This is particularly of the interest for the current study. Specifically, this study concerns the effects of web interactivity on the campaign process. Since 1996, several experimental studies on this topic have been attempted (Sundar, 1998, 1999; Ahern & Stromer-Galley, 2000), but very few were conducted in the context of a real election. The current study, which conducted online experiments in a natural setting, intended to examine whether web interactivity would get voters more involved with the campaign process as the way advocates of "electronic democracy" predict it to be. Literature Review Politics on the Net Some scholars hold that the introduction of computer-mediated political communication (CMPC) is one of the most noteworthy phenomena to inspire practical and scholarly speculation in the field of political communication since the broadcast of the Nixon-Kennedy debates (Tedesco et al., 1999). Indeed, communication has been an essential part of politics, particularly in a democratic society. Political communication, traditionally, involves four main actors: the public, the print and broadcast media, the government/politicians, and interest groups. Since most citizens do not have direct contact with politicians or public affairs, they rely heavily on media or social networks as news sources (Perloff, 1998). In such processes of political communication, the flow of political information is from the media to the public, and this is usually a one-way process. However, the rise of computer-mediated communication (CMC) may have the potential to change such one-way communication, particularly the rapid growth of the Internet. The open and interactive features of the Internet make users of online information not only the "receivers," but also the "senders" in the realm of political communication. The Internet may also remove some gatekeeping functions of the mainstream media because online users now have direct access to much information or to public records that were not previously available to the public. In the context of political campaigns, the Internet may provide the electorate with means of comparing candidate positions or ensuring political accountability (Jacques & Ratzan, 1997). It is believed that when people are sufficiently informed about public issues during the campaign periods, they are more able to evaluate the candidates, and then to make higher-quality voting decisions. Taking the position of a deliberative theory of de mocracy, the Internet is therefore often touted as a promising medium to enable voters to become better informed about campaign issues. The effectiveness of democratic processes may be greatly enhanced through such developments, and this has been called "electronic democracy" or "teledemocracy"(Dutton, 1996). However, critics disagree with such technological enthusiasm, because the latter seems to frame a seductive description of progressive social changes by neglecting several potential problems (Kling, 1991). First of all, given the potentially two-way communication which the Internet may offer, political candidates may still utilize it in an essentially one-way communication process, if that seems a better way to win more votes. Evidence shows that, during the 1996 campaigns in the United States, only 2 % of the political websites provided opportunity for communicating with the campaign staff or the candidates (Davis, 1999). Most of the candidates' websites only allowed visitors to "interact" with the campaign messages that favored the candidates. The content of online political discussion during campaign periods is another problem. Foucault (1991) argued that anonymity enables an individual to escape the "socio-political and socio-cultural bonds that fix his/her identity and consequently his/her subjective potential" (p.237). The anonymity of the computer medium may indeed contribute to flaming behavior in online political forums. The lack of face-to-face communication and nonverbal cues may lead to uninhibited and insulting messages in the CMC system. Therefore, anti-utopian theorists contend that, although CMC may offer more political discussion opportunities, most of the online discussion is banal and uncivil, and hardly benefits the deliberation process in a democratic society. Web Campaigning The influence of the Internet on politics may be observed in various political arenas, such as voters' political participation, voters' cynicism or political campaigning. This study concerns particularly the effects of online political campaigning, and therefore focus on the candidate website as a new channel and new form of political communication. The candidate web pages represent a candidate-controlled source of political communication other than paid political advertisements; they are much less expensive and allow more room for presentation of complete policy information. In the United States, the broader application of the web for political parties to set up freely accessible websites for campaign purposes did not appear until the 1996 presidential election. The emergence of the graphical web browser in 1994 inspired the extension of multimedia to the web, and thus made the Internet an enhanced tool in the 1996 political campaigns. The 1996 presidential debates on the web were found to improve political accountability by enabling voters to keep track of candidates' public actions and speeches. In the era of print and broadcast media, public figures could easily avoid statements on controversial matters because most statements would effectively disappear over time. It was difficult for citizens to keep track of the threads of meaning and logic. In contrast, a live debate on the Internet allowed voters to compare a candidate's current statements to his/her past records through the hypertext capabilities provided in web pages (Jacques & Ratzan, 1997). Experiments with the 1996 web campaigning suggested that candidates' web pages were rapid response vehicles for responding to opponents' charges, and the web technologies had direct implications for a national discussion of ideas. Voters went to the Internet, looking for affiliation, support and affirmation via the online discussions in order to reach a better voting decision (Willock, 1997). The Internet was used even more extensively during the 1998 senatorial and gubernatorial elections. One example of a successful website was that of Jesse Ventura, the candidate of the Reform Party in the Minnesota gubernatorial race. Raney (1998b) argued that Ventura's campaign relied heavily on e-mails to organize and mobilize voters throughout the state. Reney cited Ventura's website director, who said that "The Internet was not the difference; it was the mobilization" (p.1). Ventura himself described his website as "tailored-made for this campaign" and "reaching a huge amount of people at a very low price." Voter turnout topped 62 % in this election, which was much higher than the 38 % nationally. Exit polls indicated that young, first-time voters turned out in record numbers and provided Ventura the margin of victory (Kamark, 1998). In the aforementioned election campaigns, the candidates' websites were used mostly for fund raising, mobilizing volunteers, getting out the vote and enhancing media relations (Martin, 1998; Rash 1998; Resnick, 1998). In New Zealand, where elections are contested by political parties instead of by individual candidates, the web was adopted as an alternative political marketing channel (Roper, 1998). With so many new political parties (22) attempting to gain Parliamentary seats, no party could afford the risk that another party might gain an advantage through the Internet. However, upon content- analyzing these party websites, Roper concluded that a lack of knowledge of the Internet resulted in the content and format of these sites resembling those of traditional media. None of these websites allowed for "open and direct" interaction among users. The key issues covered in these sites were found to reflect the parties' campaign agendas rather than those of the voters. In the case of Taiwan, the development of web campaigning was slow in pace prior to the 2000 presidential election. In the 1996 presidential election, the first time in Taiwanese history that the Taiwanese electorate directly elected the leader of the country, the Internet was just in its primitive development. Its major users were confined within the academic circle and high-technology industries. Studies showed that merely 8 percent of the voters had ever used new media (including underground radio stations, TV call in programs, BBS or the Internet all together) for campaign information (Sun, 1996). Given the small percentage of new media users, they were found to be significantly different from the non-new media users. The new media adopters were identified as pro-opposition voters, who tended to discredit the information disseminated by mainstream media, and relied mainly on alternative media for learning about the campaign. In 1997 and 1998, the Internet enjoyed unprecended growth in Taiwan. In the 1997 county magistrate election, most of the candidates devised political marketing strategies on the web (Chou & Pen, 1999). A content analysis of the KMT's and DPP's candidates' websites revealed that the majority of the candidates did not consider the "capacity issue" of the Internet when designing their sites. They tended to use big pictures, big frames and complicated Java languages on their websites, which resulted in a quite slow downloading pace and might impede users' navigational behaviors. The growth of the Internet population was also reflected at the end of 1998 when congressional and mayoral elections were held. In a study on profiles of the Internet users, the younger and better-educated voters were found to be major users of online electoral information (Hu, 1999). They attended to Internet campaign information mainly out of a "ritual behavior," rather than a motivated communication behavior. In other words, "going online" symbolized a subculture of Taiwanese adolescences. The younger generation habitually goes online for all sorts of information, probably including campaign information during election periods. It may not indicate any specific motivation, nor indicate particular concern about the campaign itself. From the experiences of web campaigning in the United States, New Zealand and Taiwan, the primary role of the web during campaigns seems to serve the candidates instead of the voters. The main goal of viable candidates is to win. Their websites basically have been used to affect the electoral outcomes, not to educate or communicate with voters. As a result, most of the previous studies on web campaigning point to a common conclusion: The candidate websites were used primarily to demonstrate a candidate's awareness of current trends and his/her commitment to the latest technologies. Little contact with candidates was actually offered to voters. Most of the campaigning sites, either in the United States or in other democratic societies, have not utilized the interactive features of the Internet to generate direct contact between candidates and voters, or to produce more in-depth reports on candidates' policies or stances. Diamond and Geller (1995) pointed out that there are at least two worlds presented on campaign websites. One world consists of public empowerment, including bulletin boards and chat rooms, where individual citizens can post any information at will, without regard to objectivity or facts. The other world includes real information and hard political data without the influence of emotion or possible bias. Candidate websites seemed to fall somewhere between these two worlds, depending on what effects they were intended to achieve. Most candidates tend to produce and display web messages for their own best benefits. The "public empowerment" opportunity offered to the voters seldom seems to be the main concern. Website Interactivity, Affinity and Involvement New media researchers persistently claim that interactivity is a key variable for studying the uses and effects of new media technologies. For instance, Williams et al. (1988) identified three dimensions of new media research, and "interactivity" is the first one. Most discussions of interactivity draw from its roots in the cybernetic theory mapped out by Wiener (1948). Compared to Shannon and Weaver's communication model, which was available at the same time, yet was given much more attention, Wiener's model incorporated the concept of feedback into the communication process. This ability for message receivers to respond to message senders developed into a principal component of the conceptualization of interactivity. Based on Wiener's model, recent research on new media develops more fully conceptual deliberation on interactivity. Williams et al. (1988) defined interactivity in new media as "the degree to which participants in a communication process have control over, and can exchange roles in their mutual discourse." "Mutual discourse" means the degree to which a given communication act is based on prior series of communication acts. Analyzing group computer-mediated communication, Rafaeli (1988) emphasizes the concept of "third-order dependency," stating that interactivity is "an expression of the extent that in a given series of communication exchanges, any third (or later) transmission (or message) is related to the degree to which previous exchanges referred to even earlier transmissions" (p.111). In other words, to distinguish between different levels of interactivity, one must ask to what extent later messages recount earlier messages. In his continuum of interactivity, one-way communi cation only allows the source to set the agenda without feedback from receivers, while two-way communication is present as messages flow bilaterally between two parties, and reactive communication requires, in addition, that later messages refer to earlier messages. Other scholars have focused more on technological structure. Steuter (1992) reported "interactivity" as "the extent to which users can participate in modifying the form and content of a mediated environment in real-time" (p.84). His concept is governed by the speed, range and mapping capability of a medium, and therefore is of great benefit for constructing operational definitions for interactivity. In line with such an argument, Ghose & Dou (1998) proposed an operational measurement of interactivity. They called for constructing an "interactivity index" for each site to denote the maximum number of interactive functions within that site. The interactivity index then can be used to quantify the highest possible level of interactivity a visitor may experience in a site. Based on these varied definitions of interactivity, Severin and Tankard (2000) assert that interactivity in new media at least can be defined in two senses: users interact with the computer (as in interactive games) or some users interact with other users (as two users exchange e-mail messages). The former can be characterized as a kind of human vs. machine interaction, which focuses on the technological features that a user can experience in an online medium. The latter can be characterized as a kind of human vs. human interaction, which encompasses the concept of the interconnected relationship between exchanged messages. Similar theorization was provided by another study specifically on online political communication(Stromer-Galley, 2000), which stated that interactive media provide an opportunity for the message receiver to engage in two different types of interaction: computer-mediated human interaction (i.e., communication with people via a digital network) and media-interac tion (i.e., communicating with the network itself). The current study develops its operational definition of "interactivity" based on these two kinds of interactions-human-to-online media and human-to-human via online media. Elaboration Likelihood Model Broadly speaking, the current study is concerned with voters' attitudinal changes resulting from exposure to a particular type of campaign communication. The Elaboration Likelihood Model (ELM) outlines a relevant theory of attitude change by providing a framework for understanding the basic processes underlying the effectiveness of persuasive communications. To account for the differences in the persistence of communication- induced attitude change, the ELM postulates two distinct routes to persuasion: the central route and the peripheral route. The central route occurs when motivation and ability to scrutinize issue-relevant arguments are relatively high. The peripheral route occurs when motivation and/or ability to process a persuasive message are relatively low, and thus, attitudes are determined mostly by positive or negative cues in the persuasion context. The current study considers interactivity to be a vital feature of online communication because it may serve as a "peripheral cue" encouraging people to process political information. According to the Elaboration Likelihood Model (Petty & Cacioppo, 1986), when people lack the motivation to carefully process a communication message, persuasion can occur through a peripheral cue that induces attitudinal changes without people scrutinizing the central merits of the issue-relevant information. In other words, when people's elaboration likelihood for processing a certain message is low, they will not make much use of cognitive efforts, and may accept or reject a persuasion message based on positive or negative cues associated with this message. Therefore, as motivation to process information is decreased, peripheral cues become more important in determining persuasion. In the past decade, the ELM model has been successfully applied to research fields such as psychotherapy, counseling, mass media advertising and sales. With its focus on exposure to persuasive communication, the ELM has been recently applied to the field of online political communication. For instance, Sundar et al. (1998) conducted two experimental studies on interactivity in the context of political communication. First, they examined the relationship between website interactivity and psychological affinity toward a fictitious candidate on a campaign website. They found out that interactivity has the potential for improving voter affinity toward a candidate. In a later study, Sundar and his colleagues refined the previous experimental designs and made two distinct conceptualizations of interactivity-the Dialogue View and the Contingency View (Sundar et al., 1999). While the Dialogue View considers interactivity as conducting a dialogue or exchange between the user and the interf ace, the Contingency View argues that interactivity is a process in which subsequent messages are contingent or dependent upon previous messages. The later study applies the Contingency View of interactivity, and finds that interactivity can contribute significantly to individuals' psychological responses to online media content. The level of website interactivity influences subjects' formation of their impression of the candidate as well as their level of agreement with his/her policy positions. The current study intends to investigate the effects of website interactivity in the context of real elections and real candidates. Will the effects associated with fictitious candidates occur with a real web campaign? If not, in what ways will the effects differ from those in the existing literature? Research Hypotheses Based on literature reviews on several dimensions of online political communication, the current study formulate four hypotheses regarding the effects of online campaigning websites: Hypothesis 1. The more interactive a candidate's campaign site is, the greater the voters will like the site[1]1. Hypothesis 2. The more interactive a candidate's campaign site is, the greater affinity the voters will feel towards the candidate[2]2. Hypothesis 3. The more interactive a candidate's campaign site is, the more likely voters will feel the campaign issues covered in the site to be relevant to their lives[3]3. Hypothesis 4. The more interactive a candidate's campaign site is, the greater involvement the voters will feel with the campaigning process[4]4. Evidence suggests that both the human-to-machine interaction and the human-to-human interaction lead to more favorable perception of communication partners; this, in turn, produces higher-quality decisions (Burgoon et al., 1999). The more interactive the medium, the greater the increase in learning after exposure to mediated information (Schaffer & Hannafin, 1986). Thus, the current study proposes that the more interactive a campaign site is, the more the visitors will like the site (Hypothesis 1). Moreover, voters develop their perceptions of a candidate by tallying their likes and dislikes for this candidate, and then vote for the candidate with the highest positive number (Kelly & Miere). Candidate image is identified to be the primary factor affecting voter affinity. In fact, recent election studies both in Taiwan and in the United States point out that the rise in personality politics contributes to voters' primary attention to candidates' images or personal characteristics instead of to issue positions (Weaver, 1980; Hart, 1994; King, 1994; Huang, 1997; Chen, 1999). Consequently, recent election campaigns in Taiwan have centered on promoting candidate images. While previous studies traced the rise in personality politics to factors associated with television, this study proposes that, as the Internet is becoming an influential medium, some of its major characteristics (such as interactivity) may lead to the greater affinity voters feel towards candidates. The concept of affinity relates to how close a voter feels towards a candidate, based on the candidate's image in a voter's mind. Image here refers to the attitudes and feelings that a voter has towards a candidate running in a given race (Boiney & Paletz, 1991). Image is considered to be the primary factor affecting voter affinity. With technological advantages such as e-mail, web pages, discussion groups, online data bases and updated campaign news, the Internet is explored as one way to provide a candidate with better feedback on issues of concern to his or her voters. In doing so, voters may feel greater affinity towards the candidate, because he/she creates a two-way interaction channel for voters to voice their requests and concerns to the candidates (Hypothesis 2). Moreover, candidates' issue positions are found to be much less emphasized, compared to their image counterparts, in print or TV media coverage. In contrast, the Internet allows candidates to present their issue positions directly to the voters without interference by gatekeepers (Raney, 1998). Recent research indicates that an increase in the interactivity of a political website contributes to more positive impressions of the candidate as well as of his/her policy impressions (Sundar, et al., 1998, 1999). The current study proposes that, when voters' motivation or involvement to process campaign information is low, the interactivity of campaign websites may exert a more important role in determining persuasion. When the campaign issues are presented in a format that allows voters to control the reading pace, the presenting order, and the preferred modality (text, audio or video), they may be more motivated to process the information. An online discussion, following the processing of relevant campaign information, may further bring voters into the campaigning process. By interacting both with the online media (the web) and with other voters, voters may feel these discussed issues are more relevant to their lives (Hypothesis 3). Another aspect of interactive media is that it is easier than non-interactive media for message receivers to confirm and corroborate information. When electorates exhibit a lack of interest in following campaign information because of candidates' attack speeches, deception or demagoguery, the interactive campaign websites may serve as an alternative channel encouraging voters to gather unfiltered information for voting decisions. Research on political involvement suggests that voters usually weight mentally the rewards and punishments of getting involved in some kind of political participation. Voters are reluctant to get involved in the absence of tangible benefits (Kendrick & Fleming, 1983). However, the Internet provides voters a new channel to get involved in campaign processes at convenient times and places. They can check out information about any candidates, and easily compare their policy stances on different campaign issues in the online environment. If they prefer talking to someone about the campaign, there will always be someone available in chatrooms or forums on some kind of political websites. More significantly, they can do so without disclosing their identities. They can more freely express their opinion, which may be less socially acceptable in face-to-face communication occasions. Speaking of volunteering in campaign activities or giving donations, it takes only one or two clicks of mous e on the Internet. In short, it seems that the higher interactive a political website is, the more likely that it will get voters involved in some kind of campaign activities (Hypothesis 4). Methodology This study established an experimental campaign website during Taiwan's 2000 presidential election with technical help from the IS.net (An Internet service and content provider in Taiwan). This site served as a campaign website featuring the leading three candidates in the 2000 presidential election in Taiwan; it was designed as a campaign information center, which contained instructions for participating in the study, candidates' information, candidates' policies on three major campaign issues, e-mail contact with candidates and bulletin boards (for the high-interactive version only). The theme of the website, "The New President in 2000" appeared on the right-hand side of the front page, accompanied by a flashing theme logo of the site and cartoon pictures of the three major candidates. On the left-hand side of the front page were listed the frames included in the site: Instructions for the Experiment, About the Candidates, Candidates' Platforms, Post-Questionnaire, and Bulletin Board (for the high-interactive condition only). All the presented information about the candidates or their policies were drawn from the genuine candidates' websites with the candidates' permission[5]5. The subjects were recruited by classified advertisements both in print newspapers and online news sites, in order to maximize the possibility of including voters who might not be heavy Internet users. The ads described the study as one about online campaigning, and participants were promised payment of $20 in return for a fifty-minutes online experiment. In a period of 7 days, about 400 people responded to the ads either by telephone or e-mails. When they gave oral and e-mail consent to participate in the study, they were asked to provide some personal information, such as name, address, telephone numbers, age, and educational background. The researcher selected 250 respondents out of those who called, in order to approximate the voter profile of the entire population. After thorough telephone interviews for confirming their personal data, these 250 selected subjects received the URL of the pre-questionnaire on February 14 and were asked to fill out the online questionnaire by February 16. There were 220 subjects completed the pre-questionnaires and successfully submitted them to the researcher. The pre-questionnaire measured subjects' political knowledge (4 questions), and their interest in politics (10 questions). According to subjects' scores on political interest in the pre-questionnaire, they were categorized as politically lower-interested (the lower 50 %) or politically higher-interested (the upper 50%) voters, and then were randomly assigned to the three interactive conditions. One third was assigned to the low-interactive condition; one third to the medium-interactive condition; and another one-third to the high-interactive condition. These six groups operationalized the 2x3 factorial design (two levels of political interest x three levels of interactivity). Considering the potential generalizations of the findings, this study implemented a field experiment, instead of lab experiments. It is argued that the use of online experiments will maximize the generalizability of the current study, and may discover more accurate effects of web interactivity on political persuasiveness. Acknowledging that the use of online web experiments may hurt the internal validity of the study in many ways, several approaches were undertaken to minimize such consequences. First, the study adopted a password technique advised by previous web experiment studies in psychology (Reips, 1997b, 1999). Individual user IDs and passwords were created and set in the web server, and were e-mailed to the 220 subjects before the online experiment was conducted. The subjects were then required to key in the assigned ID and password before entering the experimental site. Upon entering the site, they were reminded several times that this site was an experimental campaign site, not the genuine campaign websites of any presidential candidates. In addition, the Cookie set in the web server in advance also helped to check the data quality. Each subject was told in advance that his/her participation in the study was assisted by Cookie techniques, and the information they provided in the experimen t would be kept confidential and only be used for research purposes. If they felt uneasy in the midst of the experimental course, they could leave the experimental site anytime. The experimental website was then able to keep track of the time each subject spent participating in the study. Those who spent an unreasonably short time on the site or answering post-questionnaires were eliminated from final statistical analyses. In addition, the post-questionnaires contained several questions to test the subjects' memory of the website content in order to check whether subjects received the experimental treatment properly. The 220 subjects got a reminder e-mail message the day prior to the day the online experiment was to be held. They were provided detailed self-instructions for participating in the study, including the URL of the website, according to their assigned condition. The researcher's telephone numbers (three lines) were also provided on the experimental site to help answering subjects' inquiries during the experimental course. The websites in the three conditions were identical in content and differed only in the level of interactivity. The interactivity condition was operationalized on the basis of Severin and Tankard's conceptualization of interactivity in cyber communication (Severin & Tankard, 2000). Interactivity was operationalized as the degree to which the page could be manipulated by the users (the human-to-online media communication), but also as the degree to which the site allowed interaction between candidates and voters, and among voters themselves (the human-to-human communication). The high-interactive condition provided the most hyperlinks for voters to manipulate the content, provided candidates' e-mail links for voters to send e-mails to candidates, and provided a bulletin board for voters to post their opinions on the campaign issues covered in the experimental site. The medium-interactive condition contained fewer hyperlinks than the high-interactive condition, but did not provide e ither the candidates' e-mail links or the bulletin board. The low-interactive condition contained the least number of hyperlinks; besides the entry page of this experimental site, it did not contain any hyperlinks. Subjects in this condition needed to scroll down the screen to learn about the campaign material. The MONOVA tests (Multivariate Analysis of Variance) were used to assess differences among the three groups varying in the interactive levels of treatment. These analyses dealt with web campaigning effects produced by the factor of interactivity. But before testing the proposed hypotheses, control measures were assessed first to test whether subjects' memory differences for website content or their political interest mediated the main effects of the online experiments. The 2x3 factorial analysis of variance, with both subjects' memory for the website content and their political interest as the dependent variables, yielded no statistically significant effects. The level of interactivity did not seem to influence memory for website content, F (2, 177)=.639, p>.05, nor did the subjects' level of political interest, F=(1, 177)=.814, p>.05. Major Findings The four research hypothesis propose that the effects of a campaign website will increase as a function of increasing interactivity levels of this site. However, the MONOVA tests showed somewhat surprising and mixed results. Hypothesis 1 proposes that voters' liking for a campaign website will increase as a function of increasing interactivity levels of this site. Unlike the predicted direction, Table 1 shows that the average for the medium interactive condition produced the highest ratings of the site (mean=7.41), with the high interactive condition yielding a lower rating (mean=7.15) and the low interactive condition producing the lowest rating (mean=6.7). The MONOVA test with "liking for the site" as the dependent variable showed a significant difference among groups, F=15. 453, p<.001. However, the main effect was produced by the factor of political interest (F=71.984, p<.000), rather than by the interactive level of the website (F=0.57, P>.05). The interaction effect between the two independent variables-political interest and site interactivity-was statistically insignificant, F(2, 176)=1.422, p=.24. In other words, subjects' liking for the political site was affected by their perceived interest or enthusiasm towards politics, not by the interactive level of the campaign website. Hypothesis 1 is not supported. Hypothesis 2 concerns the relationship between the interactive level of a candidate's campaign website and voters' perceived affinity towards this candidate. The mean ratings of voters' affinity and the F test are run for each candidate (James Soong, Chan Lien and Shui-Bian Chen) individually[6]6. First, the mean ratings of subjects' perceived affinity towards James Soong (Table 2) showed that the medium interactive candidate site produced the highest ratings (mean=7.64), followed by the high interactive site (mean=7.21), with the low interactive site generating the least affinity (mean=6.66). The MANOVA test further showed a significant main effect for both interactivity and political interest, F=3.105, p<.01. The main effect was produced both by the factor of interactivity (F=4.29, p<.05), and by political interest (F=6.59, p<.05). There was no significant two-way interaction between interactivity and political interest. The post hoc analysis showed that the differences among experimental groups existed mainly between the low and medium interactive groups, but not between the medium and high interactive groups. Therefore, although an "inverted-V "pattern (Sundar, 1999) for the effects of interactivity seems to exist among experimental conditions, the post hoc analysis only lent partial support to such pattern. In Chan Lien's case, the mean ratings of voters' perceived affinity towards Lien (Table 3) revealed that the medium interactive group felt stronger affinity towards Lien (mean=5.87), followed by the low interactive group (mean=5.64) and the high interactive group (mean=5.40). However, the MANOVA test indicated that the differences among experimental groups did not reach statistical significance. None of the independent variables, including interactivity, political interest or interaction between the two, produced significant differences among groups. Although the inverted-V pattern for the effects of interactivity seemed to emerge among experimental groups, the MANOVA test did not lend support to such a pattern. Table 4 illustrates Chui-Bian Chen's case. The mean ratings of voters' perceived affinity towards Chen (Table 8-5) revealed that the medium interactive group felt stronger affinity towards Chen (mean=7.88), followed by the low interactive group (mean=7.65) and the high interactive group (mean=7.14). The MANOVA test showed a significant main effect for interactivity and political interest, F=3.80, p<.01. The main effect was produced both by the factor of interactivity (F=2.22, p<.05), but not by political interest (F=.70, p>.05). The two-way interaction between interactivity and political interest was significant (F=2.66, P<.05). The post hoc analysis further showed that, for the subjects with high political interest, the differences among experimental groups existed between the low and medium interactive groups, between the medium and high interactive groups, and also between the low and high interactive groups. That is, the medium interactive site produced the highest ratings of perceived affinity, followed by the low interactive site and the high interactive site. Therefore, an inverted V-pattern for the effects of interactivity seemed to exist for the subjects with high political interest, but such a pattern did not occur among subjects whose political interest is low. For people who are not very interested in politics, the low interactive version of Chen's page seemed to work better than the medium and the high interactive versions. Taken altogether, the factors of interactivity seemed to have different effects for voters' perceived affinity towards different candidates. While an inverted V-pattern seemed to emerge for the effects of interactivity in Soong's and Lien's cases, only Soong's website produced a statistically significant result. In Chen's case, only the subjects with higher political interest followed the inverted V-pattern. In other word, the interactive levels of a candidate website did not affect voters' perceived affinity towards a candidate's image in a linear way as hypothesis 2 postulated. Hypothesis 3 measured the proposed linear relationship between interactive levels of a candidate website and voters' perceived relevance of the candidate's policies covered in this site. The measures of voters' perceived relevance of candidates' policies were composed of three items: the perceived relevance of the candidate's policies on the future of Taiwan, for reforms of political systems in Taiwan, and for social problems in Taiwan. The mean ratings of these three items will be computed first for each candidate. A MANOVA test will further determine which factors contribute to any differences existing among groups. Generally speaking, the findings for hypothesis 3 repeated what was found for hypothesis 2, that is, the "inverted-V" pattern for the online campaigning effects. While the mean ratings of voters' perceived relevance of three candidates' policies all showed that the medium interactive versions consistently produced the highest ratings, the MONOVA tests reported that the differences among the three experimental conditions reached statistical significance only in James Soong's and Shui-Bian's pages. The MONOVA test for Chan Lien's page did not lend support for the "inverted-V" pattern (Table 5,6,7). Therefore, hypothesis 3 is not supported, either. Hypothesis 4 assumes that the interactive level of a candidate's campaign website will affect voters' perceived involvement with the campaign process. The results (Table 8) showed that the medium interactive websites generated the best results (with the highest mean ratings of 6.9), followed by the high interactive version (mean=6.72) and the low interactive version (mean=6.35). The MANOVA test further showed a significant difference in subjects' perceived involvement with the 2000 presidential campaign, F=8.56, p<.001. However, it is primarily the factor of political interest that produced the main effect, F=35.846, p<.001, not the website's level of interactivity, F=2.47, p=.086. Voters' perceived involvement with the campaign process was, therefore, primarily affected by subjects' political interest. Hypothesis 4 is not supported. Conclusions & Implications for Further Research The current study examined the effects of interactivity on a 2000 presidential campaign website, in terms of site users' liking for the site, site users' perceived affinity towards the presidential candidates, site users' perceived relevance of the candidates' proposed policies covered in the site, and site users' perceived involvement with the 2000 campaign process. It was originally proposed that, the more interactive a campaign website is, the greater campaigning effects will be generated. However, the MONOVA tests showed a somewhat surprising result. The interactive levels of a campaign website did influence voters' perceived affinity towards the candidates and voters' perceived relevance of the candidates' policies covered in their sites (Hypothesis 2& 3), but not in a linear way as predicted in the first hypothesis. There generally emerged an inverted-V pattern across most of the experimental conditions, with one exception occurring with Chen Shui-Bian's page. The interactivity did not linearly contribute to positive web campaigning effects. Instead, the experiment demonstrated that there existed such a thing as "too much interactivity" on a political website. Subjects in the high interactive conditions of our experiment consistently gave lower ratings than their counterparts in the other two conditions as testing these two hypotheses. Moderate interactivity, instead, seemed to increase voters' perceived affinity towards the candidates and contribute more to vote rs' perceived relevance of the candidates' proposed policies covered in the site. Such findings corroborated previous research on fictitious elections and candidates (Sundar, 1999). Hypothesis 1 & 4 examined the relationship between the interactive levels of a campaign site and site users' liking for this site, as well as site users' perceived involvement with the campaign. The factor of interactivity was not found to significantly affect voters' liking for the campaign site, not to affect their perceived involvement with the 2000 campaign process. It was subjects' political interest that predominantly determined these effects. That is, regardless of the interactive levels of a campaign site, subjects with higher political interest tended to like this site more and to report more involvement with the campaign process. In sum, the current study did confirm that the uses of the interactive campaign websites could produce some desirable communication effects, but the uses of the interactivity needed to be cautiously controlled. While some amount of interactivity may be effective in promoting politicians or in engaging the voters with the campaign process, too much interactivity has the potential to take away such advantages. It is worthy of noting that the MANOVA test also found that voters' predisposition (for whom they would cast their votes) may be a covariate of voters' perceived affinity towards the candidates, as well as their perceived relevance of the candidates' policies. For instance, for those who decided to vote for Chen, they would perceive more affinity towards Chen's than those who decided not to vote for him. Chen's supporters also perceived his proposed policies to be more relevant to their lives than those who decided not to vote for him. The interaction between interactivity and vote was significant, indicating that among Chen's supporters, the perceived affinity of his image or the perceived relevance of his policies tended to be a linear function of the level of interactivity of his website. The high interactive version produced the highest ratings, followed by the medium interactive version and the low interactive version. In other words, the original hypotheses that assumed a li near relationship between interactivity and the effects of candidates' campaign websites were supported by a candidate's supporters, but not by other segments of the voters. Another discovery in the online experiment pertained to the absence of a significant interaction between interactivity and political interest in many measures of the dependent variables. In other words, the magnitude of the effects of the interactivity did not differ substantially between voters with high and low political interest. The current study categorized the subjects into two groups according to their levels of political interest, with an assumption those who had little interest in politics might be more attracted to a highly interactive political website than would those who had higher interest in politics. In line with such arguments, it was proposed the interactive levels of a political website might be considered a peripheral cue by the site users as they processed the political messages. However, the results did not show that voters with lower political interest would be more likely than voters with higher political interest to exhibit more communication effects due to the increased interactivity. This indicates that the interactivity of a political website may not serve as a "peripheral cue" to encourage people to process political information when they lack the motivation to do so. Of course, it may be possible that a one-shot experiment was not able to cause less interested voters to be more interested in processing political information, no matter how it was presented. The process of political communication is seldom a magic or a one-shot thing. It may take much longer than expected to observe the effects of interactive web campaigning on those who are not particularly interested in political information. On the other hand, the results from the experiments also point to several possible methodological problems with the study. First of all, the choice of online experiments over lab experiments may generate an unanticipated problem-the subjects' differences in Internet transmission speed. The original design of this study did not take this factor into consideration. However, it is speculated that the differences in the subjects' modem speed, the quality of their ISP (Internet Service Providers), or even the different ways of calculating the ISP service charge may have affected the findings. For those subjects whose modem speed is relatively slow (such as slower than 56K), or whose ISP provides a less satisfactory service, the more interactive websites mean more downloading time, which may result in a frustrating experience when using these websites. The way the ISP charges users (a fixed monthly charge or a charge based on the minutes used) may also have affected the site-users' reactions towards the messages presented in the experimental site, since a more interactive site may have cost them more money. Even the time the subjects chose to go online could have affected the outcomes (during the peak time or during the off-peak time). All of these reasons may have caused more negative experiences in navigating the more interactive websites, thus canceling out the positive communication effects brought by the highly interactive campaign messages. Perhaps as a result, the moderate interactive websites, which required shorter processing time, produced more desirable web c ampaign effects than the highly interactive ones did. Of course, the results may also lead us to reconsider the operationalization of interactivity in this study. The current research design defined "interactivity" in two ways-the human-to-machine interaction and the human-to-human interaction. The former type of interactivity was characterized by extra hyperlinks presented on the website. The latter type of interactivity was presented by the provision of an interactive bulletin board, which allowed the subjects to exchange opinions concerning the campaign messages covered in the site, as well as by the provision of e-mail links to the presidential candidates in the 2000 campaign. These two types of operationalization might characterize only a particular defining quality of the interactive nature of the Internet at a certain time. In fact, many qualities of the Internet (such as hypertextuality, multimedia, synchronicity, nonlinearity, or the message receiver as the message-sender) are bundled together in the web interface. In this light, it is important to parse out interactivity from potential confounds, and examine it as a separate, distinct technological variable. The current study defined the interactivity as the numbers of hyperlinks (involving hypertextuality, multimedia, nonlinearity), or the availability of an interactive forum (involving synchronicity, nonlinearity, and the message receiver as the message sender), but did not examine each quality as a separate, distinct variable. The entanglement of these structural features of the Internet may have generated confounding results that were not easy to interpret. The current designs for a field experimental study also may limit the possibilities to explore more dimensions of the concept of interactivity, such as vividness or playfulness. Some scholars (Steuer, 1994; Gopal, 1996) argued that vividness is the key to experiencing interactivity over the web. By vividness, they meant the number of senses that were engaged in processing web content. Therefore, text was rated low in vividness and moving images or animations with voice were rated high in vividness. However, website designs with higher vividness would inevitably take much longer time for downloading and processing. For the subjects with lower-speed modems, they were very likely to avoid processing such designs, no matter how vivid they might be. An online experimental design was, therefore, not very likely to include this dimension of interactivity. Similarly, playfulness (games or quizzes that the site users can take part in) could be another key dimension of web interactivity (Ha& James, 1998). But the designs for interactive games or quizzes may create too much complexity for an online experimental design to work. In this case, a laboratory experiment may be more adequate to explore some particular dimensions of interactivity that require complicated work. Again, the current study operationalized two kinds of interactivity-human-to-human interaction and human-to-online media interaction. This particular definition could also have some impacts on the inverted-V pattern found in this study. Future studies that take various definitions of interactivity may find different patterns for web communication effects other than the inverted-V pattern. [1] 1. Liking for the site was measured by asking respondents "On a scale of 1 to 10, with 1 being not like at all and 10 being like very much, what would you say about your liking for this website?". [2] 2. Voters' perceived affinity towards the candidates' images were measured by the ten items modified from previous studies (Sundar, 1998, 1999). The ten items can be grouped into four categories: responsiveness, trusfulness, charisma and caring. [3] 3. Voters' perceived relevance of the candidates' policies was measured by three items, in which voters were asked to what extent do they identify a candidates' visions for the future of Taiwan, his/her positions on political reforms and his/her positions on social problems. [4] 4. Finally, the level of voters' involvement with the campaign process was determined by five items: the importance level voters perceived the 2000 campaign to them (importance), the attentive level voters were to the changes of campaign polls (concern), the influential level voters perceive their votes to the outcomes of the election (impact), the interested level voters feel towards the 2000 presidential campaign as a whole (interest in campaign) and finally, the interested level voters feel towards the web campaigning in the 2000 election (interest in web campaigning). [5] 5. Because this site presented information about the three candidates at one time, some procedures were taken to enhance the internal validity of the experiment. Counterbalancing was accomplished by assigning subjects either to the condition of presenting Chen's page first, to Lien's page first, or to Soong's page first. However, later statistical analysis did not find out any significant difference due to the presentation order of the candidate's page. [6] 6. James Soong was a former KMT stalwart and Taiwan governor. He runs for presidency independently. Chan Lien is KMT's vice president and Taiwan's vice president and runs for KMT. Shui-Bian was mayor of Taipei from 1994 to 1998, and runs for DPP. The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election Tables Table 1. Mean Ratings of Voters' Liking for the Site as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 8.10 8.50 8.33 8.31 Low 5.41 6.31 5.97 5.89 Combined 6.76 7.41 7.15 7.10 F=15.45, p<.001. (Note. Scores are shifted to an 11-point scale ranging from -5 to +5 for ease of interpretation. Higher scores indicate higher ratings of perceived liking for the site) Table 2. Mean Ratings of Voters' Perceived Affinity Towards James Soong as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 7.08 7.95 7.71 7.55 Low 6.19 7.33 6.78 6.75 Combined 6.66 7.64 7.21 7.15 F=3.11, p<.01. Table 3. Mean Ratings of Voters' Perceived Affinity Towards Lien as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 5.63 5.91 5.25 5.59 Low 5.64 5.82 5.54 5.66 Combined 5.64 5.87 5.40 5.63 F=.74, p>.05. Table 4. Mean Ratings of Voters' Perceived Affinity Towards Chen as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 7.55 8.52 6.97 7.66 Low 7.76 7.21 7.28 7.42 Combined 7.65 7.88 7.14 7.54 F=3.80, p<.01. Table 5. Mean Ratings of Voters' Perceived Relevance of Soong's Policies as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 6.94 7.65 7.40 7.33 Low 5.56 7.62 6.87 6.66 Combined 6.28 7.64 7.14 7.03 F=5.60, p<.001. Table 6. Mean Ratings of Voters' Perceived Relevance of Lien's Policies as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 6.67 6.83 6.28 6.60 Low 6.18 6.92 6.26 6.43 Combined 6.42 6.88 6.27 6.51 F=.547, p>.05. Table 7. Mean Ratings of Voters' Perceived Relevance of Chen's Policies as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 8.42 8.65 7.64 8.24 Low 8.06 8.49 7.39 7.98 Combined 8.24 8.57 7.52 8.11 F=3.86, p<.01. Table 8. Mean Ratings of Voters' Perceived Involvement with Campaign Process as Functions of Interactivity and Political Interest Interactivity Political Interest Low Medium High Combined High 7.60 7.92 7.43 7.66 Low 5.83 5.88 5.39 5.67 Combined 6.72 6.90 6.35 6.66 F=8.56, p<.001. The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election Reference Ahern, R.K., Stromer-Galley, J. & Neuman, W.R. (2000). When voters can interact and compare candidates online: experimentally investigating political web effects, paper presented to ICA, Acapulco, June 1-5, 2000. Boiney, J. & Paletz, D. L. (1991). In search of the model: political science vs. political advertising perspectives on voter decision making. In Biocca, F., eds., Television and Political Advertising, V1: Psychological Process. NJ: Lawrence Erlbaum Associates. Burgoon, J. K. & Bonito, J. A, et al. (1999). Testing the interactivity model: communication processes, partner assessments and the quality of collaborative work, Journal of Management Information Systems, 16 (3): 33-56. Chen, L. (1996). Presidential voting in the 1996 Taiwan's presidential election: an analysis, Election Studies, 5 (2): 161-184. China Times (2000). Election Anxiety Hit Undecided Voters, March 1, 2000. Chou& Pen (1999). Political Marketing Strategies on the WWW, unpublished master's thesis, Graduate Institute of Communications in Tamkang University. Davis, R. (1999), The Web of Politics: the Internet's Impact on the American Political System, NY: Oxford University Press. Diamond, E. & Geller, G. (1995). Will press coverage be virtual in the online political world? The National Journal, 27 (37): 2302. Dutton, W. H. & Taylor, J. (1996). The information polity: electronic democracy, privacy, and surveillance", in Information and Communication Technologies-Visions and Realities, pp284-299, Oxford University Press. Foucault, M. (1991). Remarks on Marx: Conversations with Duccio Trombadori, New York: Semiotexte. Ghose, S. & Dou, W. (1998). Interactive functions and their impacts on the appeal of Internet presence sites, Journal of Advertising, 38(2): 29-43. Gopal, Y. (1996). Selling in cyberspace: An investigation of modality effects on cognitive processing of persuasive communication on the internet. Unpublished doctoral dissertation, University of Georgia, Athens. Graphic, Visualization & Usability Center's (GUV) 9th WWW User Survey. (1998). [Online] Available: http://guv.gatech.edu/user_surveys/survey-1998-04/. Ha, L., & E.L. James (1998). Interactivity reexamined: a baseline analysis of early business websites. Journal of Broadcasting & Electronic Media, 42 (4): 457-474. Hart, R. P. (1994). Seducing America: How Television Charms the Modern Voter, Oxford: Oxford University. Hu, Y. & Lee, C (1999). Profiles of the Internet users' electoral campaign information, Theories & Policies, 51: 65-86. Huang, S. (1997). Voter involvement and participation in Taiwan's 1994 magistrate elections, Election Studies, 2 (1): 51-76. Jacques W.W. & Ratzan S.C. (1997). The Internet's World Wide Web and political accountability, American Behavioral Scientist, 40 (8): 1226-1237. Kamark, E. (1998). Campaigning on the Internet in Off Years Elections of 1998: A Snapshot in Time. Kelley, S. & Mirer, T.W. (1974). The simple act of voting, American Political Science Review, 68: 572-591. Kendrick, F. & Fleming, T. (1983). Strategies for Political Participation, New York: University Press of America. King, P. (1994). The press, candidates' images and voter perceptions. In McCombs, M. &Shaw, D. & Weaver, D. (1997), eds., Communication and Democracy: Exploring the Intellectual Frontiers in Agenda-Setting Theory, pp.29-40. New York: Mahwah. Kling, R. (1991). Hopes and horrors: technological utopianism and anti-utopianism in narratives of computerization', in Computerization & Controversy, Boston: Academic Press. Martin, J.P. (1998). Nationwide, candidates spin the Internet, [Online] Available: http://www.washingtonpost.com/wrary/mdelect/candidatestites.htm, pp. 1-4. Ministry of Economics Report (1999). The Annual Report for the National Internet Infrastructure Development. NetValue Report (2001). Cited from Reuter's News (San Francisco) on March.26, 2001. Perloff, R.M. (1998). Political Communication: Politics, Press and Public in America, New Jersey: Lawrence Erlbaum Associates, Inc. Petty, R.E. & Cacioppo, J.T. (1986). Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York: Springer-Verlag. Rafaeli, S. (1988). Interactivity: from new media to communication. In Hawkins, R., Weimann, J. & Pingree, S. (Eds.) Advancing communication science: Merging mass and interpersonal process, 124-181. Newbury Park, CA: Sage. Rash Jr., W. (1997). The politics on the nets: wiring the political process. New York: W. H. Freeman and Company. Reney, R.F. (1998b), Former wrestler's campaign got boost from the Internet, Technology Cybertime. Available at http://www.nytimes.com/library/cyber/articles/06campaign.htm. King, P. (1994). The press, candidates' images and voter perceptions. In McCombs, M. &Shaw, D. & Weaver, D. (1997), eds., Communication and Democracy: Exploring the Intellectual Frontiers in Agenda-Setting Theory, pp.29-40. New York: Mahwah. Kling, R. (1991). Hopes and horrors: technological utopianism and anti-utopianism in narratives of computerization', in Computerization & Controversy, Boston: Academic Press. Martin, J.P. (1998). Nationwide, candidates spin the Internet, [Online] Available: http://www.washingtonpost.com/wrary/mdelect/candidatestites.htm, pp. 1-4. Ministry of Economics Report (1999). The Annual Report for the National Internet Infrastructure Development. NetValue Report (2001). Cited from Reuter's News (San Francisco) on March.26, 2001. Perloff, R.M. (1998). Political Communication: Politics, Press and Public in America, New Jersey: Lawrence Erlbaum Associates, Inc. Petty, R.E. & Cacioppo, J.T. (1986). Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York: Springer-Verlag. Rafaeli, S. (1988). Interactivity: from new media to communication. In Hawkins, R., Weimann, J. & Pingree, S. (Eds.) Advancing communication science: Merging mass and interpersonal process, 124-181. Newbury Park, CA: Sage. Rash Jr., W. (1997). The politics on the nets: wiring the political process. New York: W. H. Freeman and Company. Reney, R.F. (1998b), Former wrestler's campaign got boost from the Internet, Technology Cybertime. Available at http://www.nytimes.com/library/cyber/articles/06campaign.htm. Resnick, D. (1998). Politics on the Internet: the normalization of cyberspace, in Loader, B.D., eds., The Governance of Cyberspace: Politics, Technology and Global Restructuring, pp. 155-174, New York: Routledge. Roper, J. (1998). New Zealand political parties online: the WWW as a tool for democratization or for political marketing? In Chris, T. eds., The Politics of Cyberspace, New York and London: Routledge. Schaffer, L.C. & Hannafin, M.J. (1986). The effects of progressive interactivity in learning from interactive video, Educational Communication and Technology, 34 (summer): 89-96. SeedNet (1998). The 1998 Internet Users Poll. [Online Available: http://www.seednet.com.tw/polls/1998.htm] Steuer, J. (1992). Defining virtual reality: Dimensions determining telepresence. Journal of Communication, 42, 73-93. Steuer, J. (1994). Vividness and source of evaluation as determinants of social responses toward mediated representations of agency. Unpublished doctoral dissertation, Stanford University, CA. Stromer-Galley, J. (2000). Online interaction and why candidates avoid it. Journal of Communication. Sun, S. (1996). Analysis of the effects of voters' mass media use on political behaviors in the 1996 Taiwan's presidential election, Election Studies, 3 (2): 71-106. Sundar, S.S., Hesser, K.M., Brown, J. & Kalyanaraman, S. (1998). The effects of website interactivity of political persuasion, paper presented to IAMCR, Glasgow, UK, July 26-30, 1998. Sundar, S.S., Brown, J. & Kalyanaraman, S. (1999). Reactivity vs. interactivity: impression formation effects of message contingency in political websites, paper presented in ICA, San Francisco, May 27-31, 1999. Tedesco & Miller & Spiker (1998). Political campaigning on the information superhighway: an exploration of content and form," in Kaid & Bystrom, eds., The Electronic Election, pp. 51-63, NJ: Lawrence Erlbaum. Rafaeli, S. (1988). Interactivity: from new media to communication. In Hawkins, R., Weimann, J. & Pingree, S. (Eds.) Advancing communication science: Merging mass and interpersonal process, pp. 124-181. Newbury Park, CA: Sage. Reips, U.-D. (1997b). Psychological experimenting on the Internet, in B. Batinic, eds., Internet for Psychologists, pp.245-265. Reips, U.-D. (1999a). Theory and techniques of web experimenting, In B. Batinic, eds., Online Research. Gottingen: Hogrefe. Roper, J. (1998). New Zealand political parties online: the WWW as a tool for democratization or for political marketing? In Toulouse, C. & Luke, T. W., eds., The Politics of Cyberspace, New York: Routledge. Severin, W. J. & Tankard, J.W. (2000). Communication Theories: Origins, Method sand Uses in the Mass Media, NY: Longman. Schalken, C. & Tops, P. (1995). Democracy and virtual communities: an empirical exploration of the Amsterdam digital city, in Van de Donk et al., pp. 143-54. Westen, T. (1998). Can technology save democracy? National Civic View, 87(1): 47-56. Whillock, R.K. (1997), Cyber-Politics, American Behavioral Scientist, 40(8): 1208-1225. Wiener, N. (1948). Cybernetics, or Control and Communication in the Animal and the Machine. Cambridge, Mass.: Technology Press. Williams, F., Rice, R.E. & Rogers, E.M. (1988). Research Methods and the New Media, New York: Free Press. The Effects of Interactivity on Web Campaigning in Taiwan's 2000 Presidential Election (Note) 1. Liking for the site was measured by asking respondents "On a scale of 1 to 10, with 1 being not like at all and 10 being like very much, what would you say about your liking for this website?". 2. Voters' perceived affinity towards the candidates' images were measured by the ten items modified from previous studies (Sundar, 1998, 1999). The ten items can be grouped into four categories: responsiveness, trusfulness, charisma and caring. 3. Voters' perceived relevance of the candidates' policies was measured by three items, in which voters were asked to what extent do they identify a candidates' visions for the future of Taiwan, his/her positions on political reforms and his/her positions on social problems. 4. Finally, the level of voters' involvement with the campaign process was determined by five items: the importance level voters perceived the 2000 campaign to them (importance), the attentive level voters were to the changes of campaign polls (concern), the influential level voters perceive their votes to the outcomes of the election (impact), the interested level voters feel towards the 2000 presidential campaign as a whole (interest in campaign) and finally, the interested level voters feel towards the web campaigning in the 2000 election (interest in web campaigning). 5. Because this site presented information about the three candidates at one time, some procedures were taken to enhance the internal validity of the experiment. Counterbalancing was accomplished by assigning subjects either to the condition of presenting Chen's page first, to Lien's page first, or to Soong's page first. However, later statistical analysis did not find out any significant difference due to the presentation order of the candidate's page. 6. James Soong was a former KMT stalwart and Taiwan governor. He runs for presidency independently. Chan Lien is KMT's vice president and Taiwan's vice president and runs for KMT. Shui-Bian was mayor of Taipei from 1994 to 1998, and runs for DPP.