Keywords

1 Introduction

Prior research has shown that young adults are a growing opportunity for political outreach via the Internet but that their user experience on sites used for political endeavors may not match their needs or expectations [1]. Reports from the US Department of Education use the age range between eighteen and twenty-four when analyzing education and young adults [2]. Pew uses the age range of eighteen to twenty-nine when discussing technology trends in social media and politics [3]. For this paper young adults are considered to be between eighteen and twenty-five.

The disciplines which developed e-commerce over the past several decades have provided technology, research, and ideology that government agencies and politicians are adopting at an ever-increasing pace. The overarching label given to such endeavors is E-government [4]. Related to e-government is e-democracy or digital democracy, the use of the World Wide Web and Internet by politicians, political campaigns, political organizations, and special interest groups [5, 6]. Opinions abound regarding the effectiveness of the Internet in the advancement of democracy, yet Bimber and Davis [7] point out that there exists little scientifically collected data regarding this phenomenon.

The purpose of this study is to explore the maturation of the Internet and World Wide Web in political campaigns within the United States from the privacy perspective in order to understand how these factors affect young adult trust in political campaign websites. A framework is presented that models the campaign website privacy policy preferences of young adults. This framework is built by deploying an online survey that collects quantitative data and a series of interviews that obtain qualitative data. The information gleaned from data collection and analysis contributes a privacy model for campaign websites that positively influences young adults’ trust in the website and candidate.

2 Literature Review

2.1 Political Communication and Campaign Websites

Blumler and Kavanagh [8] described what they labeled the third age of political communication, which is characterized by an increase in “cyber politics.” Over a decade later this predicted age has arrived and the use of the World Wide Web (WWW) and the Internet within the political scene has matured consistently with each election [9]. The Internet as a new media used extensively for campaigns is due in part to its lower costs and flexibility for campaign control of message [10]. There is a body of literature that focuses on elections around the world and the impact of the Internet and the World Wide Web upon these campaigns and races [11]. A survey conducted after the 2005 UK elections found that when comparing the use of the Internet and websites for campaign purposes by young adults and older adults, young adults had a higher level of use frequency or had engaged in a broader array of digital activities [12].

Website use by political campaigns debuted for presidential elections in the 1992 Bush-Clinton race, and by the mid-1990s national parties and most congressional candidates had implemented a web presence, with national campaigns relying more heavily on web sites by the turn of the millennium [10]. In their study of how and when campaigns use web sites, Druckman et al. [13] looked at the issues of interactivity, data, and personalization as features and opportunities for campaign sites. Yet, their review of 2002 and 2004 congressional race websites is lacking a human factors component. A study of U.S. Presidential campaign websites performed by Benoit and Benoit [14] looked at both the content of said sites as well as some usability factors. In the study, an analysis of Bush and Gore websites in 2000 was performed just before the election in which the factors of identification, navigation, readability, irritability, information accessibility, interest level, information breadth and depth, adaptation to audience, and interactivity were examined and detailed to provide a framework whereby political sites of the future could be evaluated [14]. Modern websites provide an ever-increasingly rich user experience on political websites due to increased interactivity, social media and other Web 2.0 features [15].

2.2 Human Factors

Trust.

Walczuch and Lundgren [16] conducted a survey to identify the major psychological factors that affect trust in e-commerce conducted via the web. Their analysis revealed that of the factors studied (perception-based, personality-based, knowledge-based, attitudinal, and experience-based factors), trust had a strong link to perception-based and cognitive factors [16]. In another study that examined trust in e-commerce, Cases, Fournier [17] look at trust and acceptance in e-mail campaigns as they relate to purchase intention. This study revealed that a direct link between attitudes regarding e-mail campaigns and websites does not exist but that in their own right trust and acceptance are factors that can be applied to both [17]. Additionally, Lindgaard et al. [18] performed several experiments that show a budding correlation between visual appeal and trustworthiness of homepages. These studies show that this ethic of trust carries into the design space of interfaces and web design. A political candidate and campaign seek to garner trust from their constituents.

Privacy.

Privacy is an ethical concern that can have an impact on user satisfaction and trust and can evoke strong emotions from users. This study researches privacy, data collection, and data sharing within the context of political campaign websites. It examines users’ preferences regarding their online privacy and privacy policies. The idea of privacy is a construct rooted in the social and cultural fabric of a people and, therefore, the rights, challenges, and definitions are subject to change over time [19]. This means that different sociological cohorts will have differing views on privacy. This dissertation focuses on the current cohort aged eighteen to twenty-five. A recent definition of privacy is “the quality of or state of being apart from company or observation” and a “freedom from unauthorized intrusion” [20].

Privacy Policies.

Privacy policies are used by many organization but have an incredibly long way to go before they are user centric [21]. Such documents should be accessible, readable, clear, and not just placed into existence to provide legal cover [21]. The survey and interview guide used in this research project ask questions that examine user expectations and wishes regarding the accessibility, readability, and purpose of privacy policies on political campaign websites. Hong and Thong [22] performed a broad review of the literature regarding what they label “Internet Privacy Concerns” in which they detail nearly two decades of privacy research that frequently examine data collection and awareness as significant variables. More recently, the privacy paradox, the difference between what users say about their privacy preferences and their actual actions or behaviors, has become interesting to researchers [23]. This paper extends the prior research by examining these variables coupled with privacy policy readability and visibility as human factors that impact trust on campaign websites.

2.3 Research Questions

Research Question One (RQ1).

What is the level of knowledge among the users of political websites regarding the records of donations given to a political candidate via a website being public information by law?

The Federal Election Commission was founded in 1975 to provide oversight and governance for the election process of federal candidates – particularly as it relates to issues of finance. Many state governments also house agencies or bureaus responsible for monitoring not only state elections but also the financing of campaigns. The laws that govern campaign finance require that most donations made by individuals or organizations be made public – a publication which in many cases is made available via a searchable web-based database. Hunter [24] indicates that the process of e-campaigning has ushered into existence a new privacy issue with regards to democracy. Gardner [25] attempted to construct a new framework for considering policies which evaluate anonymity and public record and their influence on the democratic process. This study considers the awareness of website users regarding existing regulation regarding donations they may make. While website visitors may be concerned with issues of privacy and anonymity, federal and state laws require some data to be shared and made public. Regardless of user preference, these laws exist and the campaign websites must comply with data collection and sharing requirements. How privacy measures are implemented on a campaign website is affected by these laws and visitors’ knowledge regarding their legal rights and responsibilities may be unknown to them.

Research Question Two (RQ2).

Do political website visitors believe that they can trust the candidate more if the website has a clearly written privacy policy when compared to campaign websites which do not?

RQ2 centers on the issue of privacy perceptions on political websites as a human factors design concern. Privacy has recently seen an increase of attention both in the media and research and both theoretical and domain-specific research can be found in the HCC literature. Parayitam et al. [26] discovered that of all company policies on all e-commerce websites examined in their study, more attention was given to firm’s privacy policy than any other topic. Proctor et al. [27] examine the usability of web privacy policies and determine that amid all of the attention given to these policies readability and general usability is still lacking. It is this research question’s objective to ascertain the extent to which the perception of privacy during the use of a political website matters with regard to the readability of the campaign website’s privacy policy.

Research Question Three (RQ3).

Do political website visitors believe that they can trust the candidate more if the website has a clearly visible privacy policy when compared to campaign websites which do not?

The work of Flavián et al. [28] reveals that visitor trust of a website has a relationship to a site’s “perceived usability.” Green and Pearson [29] performed a study built on Lee and Kozar’s work which formulates an B2C e-commerce acceptance model as it relates to the HCI and e-commerce websites that finds that user perceptions play an important role in the user’s acceptance of the site. Furthermore, Walczuch and Lundgren [16] write that perception, experience, and knowledge have an influence on credibility in e-commerce. These studies exhibit that the notion of trust is a key component of a successful e-commerce website and that perceptions play a key role in credibility. RQ3 examines the visibility of a privacy policy as it affects the young adult users’ perception of the candidate.

3 Methodology

3.1 Survey

Surveys are a common data collection instrument in political science, political communication, and particularly human factors and user satisfaction research [30]. Adams and Cox [31] believe that surveys are an excellent way of discovering what is important or trivial to the participant or user. Holbert and Bucy [32] point out that “survey methods remain a dominant methodological tool for political communication analysis” [32].

The survey used in this study is a cross-section survey that is offered within a single point in time and its goal is not to measure changes over time but rather to present a “snapshot” in time [33]. Hoffman and Young [34] list voting intention and behavior, political knowledge, and media use as the three key constructs that political communication surveys measure. Attention is given to measuring political knowledge in this survey. Multiple questions on the survey and interview guide obtain this data.

Following the initial disclosure and consent form, the participant is provided with instructions for completing the survey. The first part of the survey consists of the questions related to the research questions. Questions regarding demographics and political activity comprise the latter part of the survey.

The survey is composed of closed-ended questions. Question types include multiple choice, multiple selection, rankings, and short answer. Several questions are in multiple parts or are designed to elicit responses regarding several concepts. This may take the form of questions that ask participants to “select all that apply” or to rate or rank the given options. In addition to the multiple choice, ranking, and open-ended questions the use of questions that ask the participate to select their response using a five-point Likert scale. The five-point scale provides an opportunity for the participant to express their response based on the extent to which they agree with a given question [30].

The researcher recruited participants by sending an email to computer science, information systems, and information technology professors at the two participating institutions. The professors contacted by the researcher decided if they would pass the invitation to participate to their students and they determined if any extra credit would be given to their students who participate in the study. All participants were currently enrolled college students and while their majors vary, all participants were enrolled in an undergraduate, computer-related general education or major course at one of the participating institutions. Some of these courses may discuss privacy within the curriculum. In effort to mitigate the opportunity for bias by limiting the time professors had to discuss topics such as ethics and privacy, data was collected during the first few weeks of the term at both institutions.

Study participants completed the survey using the web-based tool offered by Survey Monkey (www.surveymonkey.com). Access to the survey was provided via a link provided in an email forwarded by the participants’ professor or posted in a Learning Management System (e.g. BlackBoard). Survey data collection took place during the first weeks of the spring 2016 semester. Some professors that were asked to invite their students to participate declined, others marketed the study and did not offer any benefit to students, and others provided extra credit to students that participated in the survey. The survey response window was four weeks in length and offered early in the term. Approximately one third of the responses needed to be removed from the dataset prior to analysis because the participants were self-selected and fell outside of the required age range or neglected to state their age in the survey. This was due to user error or due to the fact that college students who desired extra credit in their course took the survey even though they did not qualify as study participants.

3.2 Interviews

The second data collection method for this research study used semi-structured interviews. The purpose of using semi-structured interviews in this study was to obtain a set of information from the participants while allowing for the flexibility of open-ended questions and follow-up questions if necessary [35]. Each of the research questions has one or more interview questions associated with it. The interview allows the participant to express his or her feelings and thoughts regarding privacy personalization on political campaign websites. A major benefit of the interview is the open-ended nature of the questions. The open-ended questions present the participant not only the opportunity to state their preference in response to the question, but also to solicit the rational for their response by following up with a second question, “why?”

The first questions in the interview collect data to answer the research questions and the questions at the end of the survey collect demographic data. The interview consisted of primarily open-ended questions that ask the participant to describe their thoughts, feelings, wishes, or observations regarding campaigns, privacy, websites, and data.

Participants for the interviews were recruited at the same two post-secondary institutions as survey participants. Flyers were posted in multiple buildings at both campuses near study areas, high-traffic areas, and classrooms. Participants were offered compensation of twenty dollars for a thirty minute interview consisting of twenty-four questions. Participants were given the option of interviewing at either of the two campuses that recruitment took place. Subjects notified the researcher of their interest in participation by email and a mutually agreed upon time to conduct the interview was selected. The interviews took place in a private office space with no third party present and the participants’ responses were separate from their identities to ensure anonymity in the results. Participants were provided an informed consent form, the interview process was explained, and participants were asked to sign the form if they were interested in proceeding. After the interview was conducted each participant was compensated in the amount of twenty dollars in cash. Most interviews required ten to twenty minutes including the greeting, informed consent review, interview, and compensation. The in-person interview recruitment process yielded ten participants. Saturation was achieved and triangulation between the data collection methods yielded plentiful results.

4 Results and Discussion

This section presents the analysis of the data collected using the online survey (N = 209) and the in-person interviews (N = 10). The survey collected quantitative data that was analyzed and the interviews collected qualitative data and which was coded. The results of both analyses were then compared and contrasted to triangulate findings. A discussion section follows the analysis for each research questions.

4.1 Demographics

The survey analysis contains data from 209 survey takers between the ages of eighteen and twenty-five with a mean age of 20.94. Sixty-four percent of survey takers report being male, thirty-five percent report being female, and one percent provided no response. The three most frequent race and ethnicity identifications were Caucasian Non-Hispanic, African American, and Asian. Eighty-five of these participants were enrolled in a public two-year institution and the remaining fifteen percent were enrolled in a public four-year institution. Of the 209 participants, thirty-seven percent report having a computer-related major; forty-eight percent report having a non-computer major; and fifteen percent reported that they are general studies, indicated they are undecided, or provided no response. Participants were of a class levels from freshman to senior.

Ten interviews were conducted with four participants being female and six participants being male. The participants were all students at a four-year institution and the sample did not include freshmen. The mean age of the participants is 21.6 and the mode is 22 and the participants reported being Caucasian Non-Hispanic, Asian, or reported multiple races or ethnicities.

4.2 Research Question One Results and Discussion

Results.

Amajority (79%) of survey participants rated privacy as an important, somewhat important, or very important factor when discussing campaign websites and campaign contributions. Interestingly, while survey participants considered privacy a concern when making donations on a campaign website, a large segment of survey takers were unaware of the record keeping practices and public government databases related to campaign donation data. Fewer than fifty percent of survey participants indicated that prior to their involvement in this study they knew that government entities record campaign donor information. That number is twenty percent when subjects responded to questions asking if they knew that donor data collected by government agencies are public via a database and/or website.

Interview participants revealed a moderate knowledge regarding what donation data is made public by the government and how it is released. Half of interviews indicated some level of knowledge while the other half of responses categorically denied having any awareness. Most of the interviewees provided one to three word responses like “absolutely” (P3), “I was not” (P4), or “yeah” (P1). A few respondents exemplify the lack of awareness of young adults regarding this topic. Furthermore, these same responses reveal that the conscious thought process of young adults is not focused on the government’s collection and sharing of donation data.

Discussion.

Hunter [24] reminds that digital technology use in campaigning has brought relatively new privacy concerns to the table when considering democracy. Participants from both data collection methods report a general unawareness regarding the data collected by political websites when donations are made. Thus, the answer to this research question is that political campaign website users have very little knowledge regarding the data collected from donors and made publicly available. Just as Gardner [25] assembled a framework for considering policies which anonymity and public record and their influence on the democratic process, so campaign designers must decide how they wish to use their site and what perception they want their site to receive from young adults.

4.3 Research Question Two Results and Discussion

Results.

In response to the fifty-four percent of participants indicated that they agree or strongly agree that they could trust a candidate when the candidate discloses his or her policies. Fifty-two percent of survey takers said that a privacy policy should contain an even balance of “legal language when compared to common language” and fifty percent were of the opinion that a campaign website privacy policy should “protect the candidate and visitor equally.

While sixty-nine percent of survey takers agree or strongly agree that reading and understanding a website’s privacy policy is essential only fourteen percent of the survey participants indicated that they read website policies on every or most sites that they visit. In contrast, thirty-seven percent of participants shared that they read website policies occasionally and forty-nine percent reported that they read these policies very rarely or have never read one.

The interview asked participants to explain their perceptions regarding the relationship between the readability of privacy policies and their trust of the candidate. Four participants responded by clearly and concisely sharing their “trust” of the candidate if the privacy policy was hard to read – each remarked their trust would be reduced or low. Others described their feelings or thoughts about hard-to-read privacy policies as “red flags” or convoluted (P1), “as bad as not having it as all” (P3), and as “not being considerate of us” (P10). Participant six clearly articulated advice for campaigns and developers by sharing that he/she things that “that it should be easier to read, that way a wider range of voters would be able to understand it.” Participants saw a complicated, legalese-oriented privacy policy as a negative factor in the relationship between the candidate the voter.

Discussion.

While studying trust in e-commerce, Walczuch and Lundgren [16] found that the psychological factor of perception plays a significant role in trust. The young adults studied in this study stated that a hard to read privacy policy has an impact on their trust of the candidate. This finding also contributes to the HCI literature because it confirms Furnell and Phippen’s [21] general recommendations for privacy policies by applying them within the political website and new media space. Such documents should be accessible, readable, clear, and not just placed into existence to provide legal cover [21]. While young adults believe that a privacy policy should protect both the organization and the site visitor, designers must clearly define what they mean by the terms privacy and anonymity in language that the site users can understand. Campaigns can increase trust among young adult website visitors by paying careful attention to the crafting of their privacy policy.

4.4 Research Question Three Results and Discussion

Results.

Of the survey takers, seventy-eight percent believe that campaign websites should have a privacy policy and eighteen percent are unsure if they believe that campaign websites should possess a privacy policy. Interview questions asked participants to describe their feelings about and trust in a candidate based on the visibility of their campaign website’s privacy policy and their ability to locate the aforementioned privacy policy. Responses to the question “How does your ability to locate a candidate’s privacy policy affect your trust of a candidate? Why?” consistently reveal that privacy policies should be visible or the trust of the participant will erode. Six of the ten interviewees shared that they felt that not having the policy clearly visible suggests that the candidate may be hiding something. Participant three went as far as to express the opinion that “generally, there’s this image of politicians are all liars and thieves and I think that’s actually generally true.” In trying to explain how important full disclosure, including privacy policies, is to the public, another participant referenced President Obama’s lack transparency regarding documentation during his campaign as an example of how not being up front and clearly presenting documentation can affect public trust in the campaign (P7). Participant eight noted that they would not feel comfortable donating on the candidate website if the privacy policy is not visible but that that would not affect their vote and participant nine stated that “[the site] would be a pretty weak…political website.”

Discussion.

The results of this study answer the research question in the affirmative. By making navigation to the privacy policy very obvious, designers can increase confidence in the candidate. These findings align with the work of Furnell and Phippen [21] whose research found that organizational privacy policies should be accessible. To make a campaign website privacy policy very easy to find for young adults and to increase trust designers can place links to the policy near data collection points on the website. The designers can also make navigation to the privacy policy a part of the entry to the home page and place a link in the more traditional locations (i.e. near the bottom of the web pages).

5 Conclusion

5.1 Implications

Young adults experience the privacy paradox when visiting political campaign websites. Young adults have concerns about the privacy implications of data collection on political campaign websites. As such, campaigns should make privacy policies highly visible and clearly written. They also should use an opt-in data collection model when designing for young adults. These steps can mitigate mistrust among young adults and present a more positive digital presence for political candidates.

5.2 Limitations

This research has three primary limitations – a limited demographic, sample representativeness, and external political or media influences.

Because this study examined young adults between the ages of eighteen and twenty five, the privacy and personalization preferences of adults outside this range were not captured. Adults older than twenty-six may have different preferences from their younger counterparts. Furthermore, because the participants resided in or near Baltimore, MD, the preferences of those from other metro areas or regions were not accounted for in the results.

Because all participants were enrolled in a post-secondary institution a limitation exists. Young adults not enrolled in college were not represented in the sample. According to the US Department of Education 40% of young adults between the ages of eighteen and twenty-four were enrolled in college in the US in 2014 [2]. A large percent of young adults in the US were not accounted for in this research project. The excluded individuals may provide data that would provide different preference results.

Because data was collected during a national Presidential primary campaign participants may have had more exposure to political advertising or political news thus creating bias that could have affected their responses to the survey. This limitation means that the participants reported preferences may be different than their responses might have been during a non-election or local election year.

5.3 Future Research

Several additional demographics could be studied in the future as the privacy and personalization preferences of different individuals could provide additional insights for campaigns and designers. Recommended demographic variables include age cohort, metro area, US region, socio-economic statues, and education level.

Since there are variations in communication dynamics when comparing individual conversations to group conversations the research in this dissertation could be extended by utilizing focus groups for data collection. Compiling responses to select questions from the survey and interview guides then comparing them to responses between private may prove interesting. This additional research may provide understanding as to how removing a semblance of privacy and, certainly, anonymity may bias conversations regarding privacy, politics, and web-behavior.