JCMC 4 (1) September 1998 

Message Board

---

Table of Contents

• Abstract
• Introduction
• Method
• Results
• Discussion
• Footnotes
• References
• About the Author

---

This study presents four models of funding for content in computer-mediated communication (CMC). These models emerge from analysis of 395 sites on the World Wide Web (Web). Key factors underlying the models are interactivity, ideologies related to intellectual property, and audience size. Relationships between these factors and funding of sites were tested by examining data collected through both content analysis of Web sites and surveys of individuals who manage the creation of content at those sites.

The data reported in this study indicate that multiple funding models can co-exist in the current CMC environment. These include content that supports organizational objectives in Sales and Promotion sites, pre-packed information and entertainment in Sponsored Content sites, a wealth of information in Public Information sites, and content provided by individuals and non-profit groups in Community Content sites.

However, the future infrastructure of CMC may play a major role in the funding of content. If systems of the future significantly increase the cost of developing and/or receiving content, the diversity of CMC could be lost. Public Information and Community Content sites may become too costly to maintain and commercial voices may gain dominance in Sales and Promotion and Sponsored Content sites. Thus, responsible public decisions should be made in the public sphere about how future CMC infrastructures are to be financed.

As computer-mediated communication has evolved from its roots in science and academe, a question that has been repeatedly raised by both academics and communication professionals is: Who will pay for content? (Lyle & McLeod, 1993; Oppenheimer, 1996)

Berkman conducted a study that found that "users aren't likely to pay for content, even news they consider valuable, if free content can be accessed elsewhere" ("Number of Online Newspapers Continues to Swell," 1996). Advertising is one method of making content "free." Additionally, some Web sites receive some or all of their funding from the organization that developed the site. Users can often access these sponsored sites at no cost. However, sometimes users do pay for content. Users of commercial services, such as America Online, pay a monthly subscription fee. Another alternative for making money from content disseminated on the Internet is the pay-per-use model.

Content can also exist in two types of "public markets." The first is online commerce (Hoffman, Novak, & Chatterjee, 1995). In this case, the content of the Web site is secondary to the primary purpose of selling. The second form of the public market is a marketplace of ideas. Governments may provide information for citizens; or individuals or groups may provide information for others who share similar interests (McMillan & Campbell, 1996). The unfunded or minimally funded content found in the marketplace of ideas includes e-mail messages and postings to newsgroups.

The previous discussion identifies primary sources of funding for content in computer-mediated environments. However, the goal of this study is to go beyond description to a model that is at least normative and perhaps predictive. To this end, three bodies of literature are reviewed: interactivity, ideologies related to intellectual property, and audience size.

Interactivity. Much of the popular and scholarly literature on computer-mediated communication describes it as an interactive medium without defining interactivity. However, discussions of interactivity in CMC seem to be organized around three primary perspectives: users (Chen, 1984; Kayany, Wotring, & Forrest, 1996; Walther, 1994), structure (Rice & Williams, 1984), and process. The process view is a dominant research theme and can be subdivided into research on roles and behaviors of the participants (Rice, 1984; Williams & Rice, 1983), content of messages (Newhagen, Cordes, & Levy, 1995), and sequences of actions (Bretz, 1983; Newhagen & Rafaeli, 1996; Rafaeli, 1988; Steuer, 1992). The literature also reflects six user-oriented dimensions of interactivity: threats (Markus, 1994), benefits (Ang & Cummings, 1994; Schaffer & Hannafin, 1986; Shaw, Arnason, & Belardo, 1993; Sproull & Keisler, 1991), sociability (Fulk, Flanagin, Kalman, Monge, & Ryan, 1996), isolation (Beniger, 1987; Dorsher, 1996; Mantovani, 1994; Spears & Lea, 1994), involvement (Borgman, 1986; Trevino & Webster, 1992), and inconvenience (Stolz, 1995; Thomas, 1995).

Heeter positions interactivity in the structure and processes of the medium. She identifies six characteristics of interactivity: complexity of choice available, effort users must exert, responsiveness to the user, monitoring information use, ease of adding information, and facilitation of interpersonal communication (Heeter, 1989). Her definition is one of the few analyses of interactivity that offers specific, measurable dimensions.

Ideologies Related to Intellectual Property. North provides a summary of neoclassic economic thought in which the system of property values is one leg of a three-legged stool upon which economic systems rest. The other two legs are an ideology and a state which enforces both the property values and the ideology (North, 1981).

Olson applies neoclassical economic theory to group behavior. He suggests that groups exist only when the individual benefits of their actions exceed the costs or when individuals can be coerced into joining such groups. In general, rational individuals will not incur the costs of participating in group action when the benefits of the group's action can be received by being a free rider (Olson, 1965). North turns the problem of free riders into a new question. Why is it that some people do NOT choose to take the free ride? For example, he asks why do some people voluntarily and anonymously donate blood? He suggests that a shared ideology is largely responsible for behaviors that can not be explained in economic terms. This shared ideology is particularly evident with respect to how the social group places value on property.

The concept of intellectual property is not an ideology accepted in all cultures. Bettig points out that in Roman society it was appropriate to attribute authorship and to compensate authors for their intellectual works. However, modern India, Bali and China do not have an ideology that embraces the concept of copyright. In those societies, the production of culture is a community-oriented, participatory process and therefore not subject to the concept of individual ownership or personal gain (Bettig, 1989). According to North's conceptualization, this suggests that a society which has adopted an ideology of intellectual property rights (and associated state sanctions against violation of those rights) would be one in which the large majority of citizens do not attempt to take a free ride by violating those rights.

Multiple ideologies related to intellectual property exist on the Internet. Anawalt suggests the dynamics of electronic media and the logic of copyright law combine to make many Internet communications potential copyright infringements (Anawalt, 1996, p. 5). Many copyright holders fear they will be unable to protect their intellectual property in a medium in which copying is as simple as a keystroke and in which copied material is virtually indistinguishable from the original (Angell & Zelkha, 1997). Yet, a large portion of Internet users share Goodwin's belief that freely available information is a central tenant of a democratic society (Goodwin, 1997). The culture of the Internet has long held that "information wants to be free." However, a growing number of content providers are coming to believe that "information wants to be worth something ("How Micropayments Will Change Our Lives!", 1993)."

This study assumes no single ideology exists about intellectual property in CMC. Rather, perspectives lie along a continuum. At one end of this continuum one may expect to find those who believe CMC is jointly held by a community with roots in volunteerism. Similarly, for those who view CMC as a selling space, intellectual property has little intrinsic value. Value is found in the transaction that occurs after a buyer attends to content. At the other end of the ideology continuum are those who believe that communication has intrinsic value and that someone (either the user or a sponsor) should pay before that information is made available to others.

Throughout this study, the term property values is used to reference this ideology continuum. High property values refer to those who have adopted an ideology that respects intellectual property; low property values refer to those who eschew such an ideology.

Audience. The audience may be another critical factor that determines who pays for content in CMC. Smythe contends that the primary business of content creators is not the creation of content but the creation of audiences. Audiences attract advertisers and advertisers are the major source of funding for most mass media in the United States (Smythe, 1977).

As early as 1978, scholars were questioning what role the buying and selling of audiences would have in the financial structure of computer-mediated communication. Erik Barnouw asked: "What role will sponsorship - and the financial control it involves - play in the system to come?" (Barnouw, 1978, p. 178) His historical analysis of the role of sponsors in broadcast media led him to suggest that advertisers' support of media results in a system where content has little or no meaning except as a tool for attracting an audience which can be bought by advertisers.

Recent developments suggest that Internet content providers are becoming aware of audience issues. Traditional audience measurement companies such as the Audit Bureau of Circulations have announced plans to provide advertisers with measures that will help them determine the value of Internet audiences ("ABC Eyes World Web Standards," 1996). And new coalitions are forming to create audience measures such as the CyberMeasurement Index (CMI) which provides media buyers with a standard way to evaluate competing online advertising possibilities ("Step Toward Global Measure," 1996).

Three hypotheses predict the main effects for each factor reviewed above.

H1: Higher levels of interactivity are expected in sites that receive the majority of their funding from volunteer efforts, non-profit organizations, government/education, and mixed funding sources;

Lower levels of interactivity are expected in sites that receive the majority of their funding from for-profit companies and sponsorship/advertising.

H2: Higher scores on property value scales are expected from content creators whose sites receive the majority of their funding from for-profit companies, and government/education;

Lower scores on property value scales are expected from content creators whose sites receive the majority of their funding from volunteer efforts, non-profit organizations, sponsorship/advertising and mixed sources.

H3: Larger audiences are expected to be found at sites that receive the majority of their funding from sponsorship/advertising, government/education and mixed sources;

Smaller audiences are expected to be found at sites that receive the majority of their funding from for-profit companies, volunteer efforts, and non-profit organizations.

Testing the above hypotheses required both a content analysis of selected instances of computer-mediated communication and a survey of content managers. The content analysis was designed primarily to determine level of interactivity at selected sites. The survey asked managers of those sites for information about funding, perceived value of intellectual property, audience size, purpose of the site, and other related factors.

This study focused on health-related topics for four reasons. First, messages about health and health-related subjects have played a central role in the early financial support for content in many media. For example, patent medicines were among the first products advertised in newspapers (Jones, 1996) and makers of health-related products were among the first sponsors of radio programming (Barnouw, 1966). Second, health-related sites are among the fastest growing topic areas in CMC (Fisher, 1996). Third, researchers have found that health-related sites in CMC are used heavily (Pingree, Hawkins, Gustafson, Boberg, Bricker, Wise, Berhe, & Hsu, 1996). Finally, health care is currently one of the fastest-growing categories of consumer advertising (Wilke, 1997) .

Sites were selected from health-related categories on the Yahoo electronic directory of Internet content. A total of 1,050 sites was selected randomly from the nearly 15,000 health-related sites listed by Yahoo in early 1997. After removal of duplicates, non-functioning Web addresses, and bad e-mail addresses, the valid sample size was 834 Web sites.

Surveys were sent and received via e-mail. A reminder message was sent one week after the initial message. A total of 395 completed surveys were returned, resulting in a response rate of 47.5%. The content analysis was conducted by three trained coders on all sites for which a completed survey was received. Intercoder reliability was checked for 10% of the sites using Holsti's reliability formula. Coders achieved 92% agreement.

Sources of Funding. A key variable for each hypothesis is the source of funding for the creation and maintenance of the content of a site. Survey respondents were asked to provide information about how sites are funded. Table 1 summarizes funding sources.

Interactivity. The level of interactivity was operationalized using the dimensions of interactivity defined by Heeter. Two measures were used for complexity of choice. First, coders counted the number of links from the first page of the site. It is presumed that a greater number of links is an indication of higher complexity of choice. Second, the presence of a search engine was considered to represent greater complexity of choice. In creating the Complexity of Choice scale, the Search Engine variable was recoded as 1 for absence and 2 for presence of the feature. This recoded variable was then multiplied by the Number of Hot Links from Front Page of Site variable to create a 10-point scale. This provides equal weight to the variables used in creating the scale. Kendall's Tau for this scale was .71 (p <.001).

The effort that users exert was operationalized by how many tools were provided to help users navigate the site. An additive scale (range = 0 to 3) was created based on how many of the following tools existed at the site: a menu bar on the first page of the site that provides a brief description of the sections of the site, with links to primary sections of the site; a menu bar on subsequent pages; and the presence of a hot link that takes the site visitor directly back to the home page. The logic of Heeter's definition suggests that this scale should be reverse-coded for analysis. She indicates that more effort is an indicator of higher interactivity. While navigational tools might make the novice user more comfortable in using the site, they actually reduce the number of choices the user makes. Thus, this variable was recoded for analysis on a scale of 0-3 where 0 is a low level of interactivity (having all three navigational tools) and 3 is a high level of interactivity (having no navigational tools). Cronbach's alpha for this scale was .62.

Responsiveness was based on whether or not the site had a "feedback form" for eliciting input from the user. Monitoring of information was measured by an additive scale (range = 0 to 2) indicating if the site has a "hit counter" that shows how many individuals visited the site and/or if the site includes a message indicating when content was updated. While these are relatively simple measures of monitoring, they suggest the level of attention site creators are paying to the audience and the content of the site. Kendall's Tau for this scale was .58 (p < .05).

The final dimension combines Heeter's concepts of "ease of adding information" and "facilitation of interpersonal communication" into a single dimension which is identified throughout this study as Interpersonal Communication. This dimension was operationalized by an additive scale (range = 0 to 2) indicating whether the site has a public bulletin board for posting messages ¹ and newsgroups which facilitate structured interpersonal communication. Kendall's Tau for this scale was .61 (p <.01).

Ideologies Related to Intellectual Property. The survey asked site managers to use a five-point scale to indicate their level of agreement with six statements. These questions use the term cyberspace to refer to computer-mediated communication because cyberspace is probably a more familiar term for these respondents ².

The literature suggests that agreement with the following statements should indicate that the respondent places a relatively high value on intellectual property.

1. Copyright laws should be enforced in cyberspace.

2. Good content creators should receive some kind of financial reward for the information that they post in cyberspace.

3. More security is needed to prevent unauthorized access to information.

The literature suggests agreement with the following statements should indicate that the respondent does not place a high value on intellectual property.

1. One should not have to pay for information if one can find the same information for free in cyberspace.

2. The concept of authorship should not exist in cyberspace.

3. Information that can be found on the Internet should be free and available to everyone.

Another factor which may impact on perceptions of value is whether content is viewed as a vehicle for facilitating transactions. Respondents were asked to use a 5-point scale to indicate their agreement with following statement: "The primary role of information in our Internet site is to facilitate online transactions (e.g., sales, donations, signing up new members, etc.)."

Audience Size. Audience size was measured in two ways. First, in the content analysis, coders looked for a hit counter; if one was found they recorded the current number. The second measure of audience size was collected via the survey. Respondents were asked to indicate the average number of individuals who access their site each week.

Respondents were asked to indicate annual cost of creating content at their sites. Table 2 summarizes responses. Creation of content is a low-cost activity for most of these sites. Overall, levels of interactivity were relatively low at these sites. For example, only 7% of the sites included newsgroups, only 19% included search engines, and only 34% provided a feedback form. All measures of interactivity were standardized for reporting purposes.

The six items used for testing property values of content creators did not scale together (Cronbach's Alpha = .29). Therefore each of the six variables were considered separately in analysis of the property values of content creators. Items associated with low property values were recoded so that all items are on a scale of 1 to 5 where 1 indicates low property values and 5 indicates high property values.

Hit counters for displaying measured audience size were found at 34.6% of the sites. Counts ranged from 25 to 574,563. The mean value of hit counters was 16,742 (std. dev. 61,084). Self-reported audience size ranged from 4 to 170,000. The mean value of self-reported visitors per week was 2,885 (std. dev. 14,024). To facilitate analysis, both measures of audience size were standardized.

Hypothesis 1. Analysis of Variance (ANOVA) was used to test hypothesized relationships between interactivity and sources of funding. Five separate ANOVA tests were run. For each, the independent variable was source of funding. Dependent variables were the five dimensions of interactivity. Table 3 reports those five ANOVA tests.

In examining hypothesized relationships, findings were considered significant if p values were .05 or less. Hypothesis 1 was partially supported. Significant F values were found for three of five measures of interactivity. Most significant relationships were in the hypothesized direction.

The dimensions of interactivity which seem to hold the most promise in relation to funding sources are Complexity of Choice and Monitoring Information. Neither Responsiveness nor Interpersonal Communication showed significant relationships with funding sources. Effort Users Exert showed significant, but problematic, relationships with funding sources. Sites that received funding from government/education organizations and those that receive funding from mixed sources were most likely to score high on significant measures of interactivity. Sites created by for-profit sources consistently had the lowest scores for most interactivity measures.

Hypothesis 2. ANOVA was also used to test hypothesized relationships between sources of funding and perceived value of intellectual property. Six separate ANOVA tests were run. For each, the independent variable was source of funding. Dependent variables were the six measures of property values. No significant relationships were found between scores on property value scales and sources of funding using these tests. However, more sophisticated ANOVA tests, using the same dependent and independent variables while adding covariates, were used to explore the possibility that other factors may influence the relationship between property values and funding sources.

First, those who see communication as a tool for facilitating a transaction may place lower value on content. Responses to the statement "The primary role of information in our Internet site is to facilitate online transactions" were included as a covariate. Second, the cost of creating content may impact on hypothesized relationships. Those who spend more on creating content may place higher property values on the content. Finally, the presence of professional communicators may have an impact on predicted relationships. These individuals may earn their living from intellectual property and therefore value it highly. A variable that indicates how many communication professionals work on the creation of content for a site was added to the ANOVA model as a covariate.

With the addition of the covariates, partial support was found for hypothesis 2. Two of the six ANOVA tests revealed significant relationships between perceived value of intellectual property and funding sources. Table 4 summarizes findings related to hypothesis 2. All variables are coded in such a way that higher means reflect higher property values. Most significant relationships were in the hypothesized direction.

While significant relationships were only found for two of the property value variables, trends in the data tend to support hypothesized relationships. In general, creators of content who get their funding from for-profit, government, or educational organizations are more likely to place high value on intellectual property than are those who fund creation of content from other sources.

Hypothesis 3. To test relationships between audience size and sources of funding, two separate ANOVA tests were run, one for each audience measure. The number on the hit counter was unrelated to funding variables (df 5, F = 1.82, p > .05). Similarly, self-reported audience size did not have a significant relationship to funding sources (df 5, F = 2.07, p > .05).

Respondents who reported on audience size were asked how they measure that audience. Five different methods were identified and used as a covariate in an ANOVA test. When these covariates were added, the main effect was still not significant (df 5, F = 1.78, p > .05). Thus the data do not support hypothesis 3.

However, in the second analysis of self-reported audience size a significant relationship was found between the covariate (method of measuring site visitors) and the number of visitors (df 1, F = 22.26, p <.000). To further explore this relationship, the method used for measuring site visitors was cross tabulated with funding sources. The c ² value for this cross tabulation was 33.8 which is significant at p < .05. Table 5 shows the relationships between these two variables. Numbers in the cells are percentages for the row.

Table 5 may provide insight into why hypothesized relationships between audience size and funding source were not found. Almost 80 percent of respondents use hit counters and/or raw log files to measure audience size. These techniques are notoriously unreliable measures of audience (O'Connell, 1995). Only about 20 percent of respondents utilize either internally or externally developed custom tools for systematically analyzing audience size.

Table 5 may also suggest some trends that are consistent with the underlying principles of Hypothesis 3. Sites which receive funding from sponsorship/advertising are the group most likely to use customized software for tracking audience size. This suggests that accurate measures of the audience are important to these site managers. This finding may indicate that these managers view the audience as a commodity to be measured, analyzed, and sold to advertisers/sponsors. By contrast, sites that receive funding from volunteer efforts are the group most likely to rely solely on hit counters as a measure of audience size. This suggests a less-calculating approach to audience measurement that may be more consistent with notions of audience as virtual community rather than audience as commodity.

Figure 1 summarizes four models of funding for computer-mediated communication which emerge from the findings in this study. These models also incorporate concepts drawn from survey respondents' description of the purpose of their sites.

Figure 1. Four Models of Funding for Computer-Mediated Communication.

Community Content characterizes 57.3% of the sampled sites. Community Content sites can exist with minimal funding from community-minded organizations and with the donated time, efforts, and expertise of volunteers. The content of such sites is an ever-changing montage of ideas, information, and intelligence created and shared by the participants. These sites are more likely than others to incorporate elements of interactivity and content creators are likely to place relatively little value on the concept of intellectual property. In this environment, a many-to-many model of communication may be most likely to exist. Some community-oriented sites may attract large numbers of participants from all over the globe. Others may be very small and focused, like many of the disease-related support groups found in this study.

Sales and Promotion, found at 20.7% of the sites, is a win-win situation for companies. The sites may cost little to create and, along with Sponsored Content, they attract qualified customers. This targeted communication reduces the need for expensive advertising campaigns in traditional publications that may deliver a high volume of waste circulation (readers who have no interest in the company's product). Interactivity at Sales and Promotion sites may be reduced to online ordering that does not take full advantage of the of many-to-many communication potential of CMC. Instead, the sites are designed to facilitate the kind of one-to-one relationship that exists in personal selling. Creators of these sites may have high property values, not because they want to protect the content of their own sites (the content of their sites has value primarily because it promotes the company and/or leads to sales) but because they come from a market-oriented perspective that accepts the ideology associated with intellectual property.

Sponsored Content characterizes 11.1% of these sites. These sites may become very much like traditional mass media forms in the way that they pre-package information and entertainment to be delivered to target audiences. As demonstrated in this study, sites that are attempting to build a business model based on advertising and sponsorship make minimal use of interactivity. As with other traditional media, a primary purpose of such sites is to attract audiences; however, audience size may not be critical. Creators of Sponsored Content who can attract small but specialized audiences may be able to sell those audiences to advertisers at a premium and enjoy profitability.

Finally, 10.8% of the sites are characterized as Public Information. The sites may display a wealth of information content. It is the intrinsic value of such content that leads to many of the concerns about the growing gap between the information rich and the information poor. Interactive characteristics of the medium facilitate search and retrieval of content and the communication is user-driven. The individual who comes to the site is likely to take an active role in deciding what type of information s/he is exposed to. In this sample, publicly-funded government and education sites are the primary providers of Public Information and, for the most part, they provide information at little or no direct cost to the citizen. However, Public Information sites also represent an opportunity for companies that specialize in preparing proprietary information. As security of Internet transactions improves, there may be a rise in the number of Public Information sites that are funded by direct consumer payment in the form of pay-per-use and/or subscription fees and thus come to be characterized as private, rather than public, information.

Suggestions for Future Research

Future research should begin by attempting to strengthen measures of interactivity and property values. Future studies should also expand beyond health-related content to include other topic areas such as sports, news, and travel. Research could also focus on forms of CMC that are more likely to include pay-per-use and subscription-based funding of content.

Future studies should continue to explore the relationship, or lack thereof, between audience size and funding sources. Is this medium immune to commodification of audiences? Has the size of the audience become less important than the composition of the audience?

Finally, research is needed on how CMC is accessed. This study focused on how content creators fund the cost of site development. It does not address the issue of how individuals gain access to that content. Currently many gateways to the Internet exist. Some, such as libraries and free nets, provide access at little or no cost. Some individuals have access through employers. Many individuals access the Internet through commercial providers who charge subscription and/or pay-per-use fees. Examination of relationships between the cost of access and individuals' actual use of various types of content are a critical next step.

As noted at the start of this paper, the question of who will pay for the content of computer-mediated communication has significant public policy implications. This study shows that for now, multiple models of funding can co-exist. For the future, public intervention may be needed to insure that costs of creating content remain low enough that multiple voices can be heard in arenas such as Community Content. On an encouraging note, this study suggests that public funds are currently being invested in highly interactive sites which provide the type of information that, at least from the perspective of site managers, has high property value. Such an investment is a necessary balance to the private interests that could seek to privatize these Public Information sites by imposing subscription or pay-per-use fees.

The four funding models described in Figure 1 have implications for communication educators. Students need to have a broad range of skills if they plan to create computer-mediated communication content. Those creating Sponsored Content sites will need many of the same skills employed by traditional journalists. Those creating Sales and Promotions content will need the skills of advertising and public relations practitioners as well as selling skills. Those creating Public Information will need to learn to create databases and information environments. Those creating Community Content will need to blend interpersonal and mass communication skills. For all students it will be important to understand the interactive nature of the medium - regardless of how much interactivity is included in the sites at which they work.

Additional implications can be suggested for communication professionals. In particular, the role of audiences is changing dramatically. No longer can content creators assume that they can attract audiences and thereby obtain advertising support for that content. Advertisers have new ways of communicating directly with target audiences. Furthermore, if individuals begin to demand higher levels of interactivity, both advertisers and traditional creators of content may find that they have been deserted as the users of CMC migrate to Public Information and/or online Community Content sites.

Generally, then, this study represents a necessary early step in understanding relationships between interactivity, property values, audiences, and funding sources for computer-mediated content. While CMC has the capacity for many-to-many communication, some content creators are utilizing it in ways reminiscent of one-to-many mass communication vehicles; others are attempting to develop forms of one-to-one communication. The present findings suggest that the computer-mediated communication environment is currently robust enough for such diversity of content, funding sources, and communication models, to co-exist.