College of Library and Information Services, University of Maryland, College Park, Maryland
Science and Technology Center, Kellogg Company, Battle Creek, Michigan
Diffusion of innovations theory offers a powerful paradigm for conceptualizing the development and acceptance of electronic publishing ventures. Investigations of the diffusion of other innovations suggest that acceptance or adoption is significantly influenced by innovation characteristics and adopter perceptions. Electronic publishing can best be understood as a cluster of related innovations which can be incorporated in different combinations. Cluster members individually and in combination influence adopter perceptions of the relative advantage, complexity, compatibility, visibility, and trialability of a particular implementation. Acceptance of electronic publishing is further shaped by the requirement of a contingent adoption decision by a community. For instance, readers cannot adopt an electronic publication without author acceptance. This paper will define the innovation cluster and provide a framework to characterize several recent publishing ventures demonstrating how the framework facilitates comparison and evaluation of individual implementations of electronic publishing ventures.
Several authors have claimed that the development of electronic publishing will produce profound changes in scholarly communication (Harnad, 1990; Harrison & Stephen, 1995; Schaffner, 1994). Advocates argue that electronic publishing will empower both authors and readers and support richer communication at lower cost. However, although its advent has been long awaited, electronic publishing is still very much in the early stages of development and adoption. In the sciences, several pilot projects have been attempted, but the impacts of these early efforts have not been felt to any great degree by the scientific community. The vast majority of scholarly publications remain in traditional paper formats.
If electronic publishing is viewed as an innovation, a powerful paradigm exists that can help explain the factors that will be involved in its development and dissemination. Researchers studying the diffusion of innovations have successfully identified methods of classifying innovations, mapped out an innovation decision process, suggested adopter categories and their characteristics, determined the factors that influence adoption decisions, and described influential social structures in relation to adoption decisions (Rogers, 1995). This approach suggests some useful ways of looking at electronic publishing as a diffusion of innovation process.
Rogers (1995) describes several ways that innovations can be characterized, but a particularly relevant insight in relation to electronic publishing is the recognition that sometimes multiple innovations must be adopted simultaneously. From this viewpoint it is clear that electronic publishing should be seen, not as representing a single innovation, but rather an innovation cluster. Innovation clusters occur when a group of innovations are adopted in some combination. In making an adoption decision, all members of the cluster do not have to be selected, but it is likely that some will be adopted together.
In the case of electronic publishing, trial developments of this innovation suggest that while a variety of technologies are typically incorporated, the exact mix may vary. Electronic distribution seems to be core to the innovation cluster since it has been incorporated into all of the prototype projects, but variations are seen in who plays the role of publisher, what kinds of novel formats are available, how research is validated, how publications are sold, and how the digital information is stored in the short and long term. Each of these cluster categories potentially can have additional facets. For instance, electronic distribution could be e-mail distribution of articles or issues, notification of the availability of new articles or issues, and individual retrieval of articles from some remote storage site.
This approach offers researchers a conceptual framework for characterizing current electronic publishing projects and analyzing hypothetical publishing scenarios. In addition, examining electronic publishing as a diffusion of innovation process could allow identification of those members of the innovation cluster which are most likely to be accepted and adopted by a wide audience. Such a method of examination contrasts with strategies currently being employed in the examination of electronic publishing. A typical approach which many writers have used is hypothesizing an outcome an approach based on the authors estimation of what would be desirable (Bailey, 1994; Harnad, 1990; Lynch, 1994; Okerson, 1991). Some authors use scenarios to examine or simply advocate for improvements of electronic publishing over print-based publishing, (e.g. Gardner, 1990 and Stodolsky, 1995). This is not a universally popular approach, and at least one author has been highly critical of the way scenarios have been used to frame discussions of electronic publishing (Schwartz, 1994). While these authors clearly recognize that electronic publishing could be implemented in a variety of ways, the viewpoint of electronic publishing as multiple, relatively independent innovations which must be adopted in a cluster has not been developed. Still other investigators have examined the development and impact of electronic publishing by querying users, potential users or other groups regarding their preferences or perceptions of the value of various effects that might result from electronic publishing (Berge & Collins, 1996; Bishop, 1995; Borghuis et al., 1996; Lancaster, 1995; Olsen, 1993; Schauder, 1994; Stewart, 1996 have used this type of approach). Often these studies simply ask users to react to what has been made available to them or to an incomplete list of possible characteristics of the new medium. Again, potential innovations are rarely considered individually in this work. In the focus on potential outcomes, the actual innovations that will create the desired benefits are rarely detailed.
The focus of previous studies may be attributable to the fact that they were performed when electronic publishing was still in the earliest stages of development and adoption was in the future. However, recent advances in technology have brought electronic publishing to the point where pilot projects are underway and some, albeit perhaps limited, adoption decisions are being made. In this new environment a typology of the members of the innovation cluster will be useful for comparing implementations and identifying characteristics affecting adoption decisions.
Such a typology of innovation cluster members is proposed in Figure 1. While it would be impossible to identify all of the possible and potential innovations uniquely, examination of previous literature  and the current environment of information technology suggest several basic categories of innovations. Within each category, innovations are identified relatively independently of specific applications. Thus electronic distribution does not assume Internet-based e-mail distribution as currently used in some instances. This type of implementation is currently being used, but distribution by other e-mail systems is conceivable. This approach was chosen to avoid limiting the utility of the typology to a specific time and mix of available application. Also, innovations included were restricted to truly novel processes. Thus the conventional role of commercial publishers does not appear in the category of publishing roles. Community acceptance of this publishing role does not require any kind of adoption decision since it already exists. Thus this publishing role is excluded from the typology.
To demonstrate the applicability of the proposed characterization of the innovation cluster two of the most prominent of recent pilot projects in electronic publishing, Online Journal of Current Clinical Trials (OJCCT) and the TULIP project, will be examined. Keyhani (1993) provides a detailed description of OJCCT, developed as a joint venture between OCLC and AAAS (the American Association for the Advancement of Science). As initially developed, this electronic journal offered innovations in the areas of distribution and retrieval, document structure, and storage. OJCCT innovated in distribution and retrieval by making articles the unit of distribution rather than issues, and by providing central storage by the publisher allowing individual retrieval by users on demand (via a network). The journal supported full text searching of all articles via a graphic user interface. Text and figures were stored electronically allowing the removal of article length restrictions and hypertext links were provided to related works such as other OJCCT articles, MEDLINE records, and Genbank records. The developers of OJCCT did not incorporate innovations in publishing roles, research validation, or pricing.
The developers of the TULIP project  made some different decisions in the development of their pilot electronic publishing ventures. Borghuis et al. (1996) and Willis et al. (1994) provide detailed descriptions of the implementations developed and also an analysis of problems encountered both in developing electronic journals and encouraging adoption by users. Although TULIP includes several different implementations of electronic formats provided by Elsevier, differences between sites were largely in interface design and integration with local systems. The fundamental types of innovations varied little between sites. The innovations developed again included distribution and retrieval innovations, document structure, and storage. However, in the case of TULIP, the electronic publications duplicated existing print publications. The journal issue remained the distribution format and distribution innovations consisted of central storage and individual retrieval. In several instances TULIP systems were integrated into existing retrieval systems such as library OPACs. Although Elsevier provided both page images and ASCII text, several sites used only page images. Thus only a few sites worked with ASCII text. Only some of these latter sites offered full text searching of content. Library storage of the electronic page images and in some instances ASCII full text was utilized. It should be noted that time lags developed in delivery of content from the publisher to libraries since distribution was ultimately via CD-ROM (network distribution was attempted and found impractical). In many ways the electronic publishing efforts of TULIP were less novel than those of OJCCT since there were so few innovations in document structure. Despite the more modest innovation cluster developed, adoption was very low. All of the participating libraries reported low usage despite widespread efforts to publicize the project and its benefits to users. Author adoption was not an issue in the TULIP approaches since publication in the electronic journals was predicated in on publication in print counterparts.
In addition to characterizing innovations, the diffusion of innovation paradigm maps out the adoption decision process. The decision process begins with knowledge of the existence of the innovation. This is followed by a persuasion period. During this phase potential adopters gather information from various sources and attempt to determine the utility of the innovation. Often, early adopters, typically innovators themselves, or in some cases change agents, attempt to convince the general user population of the benefits of the innovation. The persuasion phase ends with a decision to adopt or reject the innovation. If the decision is in favor of adoption, an implementation phase follows. Implementation is a critical period in the diffusion process since it requires overt action on the part of adopters as the new idea is put into practice. As Rogers (1995) notes, "It is often one thing for an individual to decide to adopt a new idea, but quite a different thing to put the innovation into use." At this stage, the innovation may become incorporated into the culture of the user population, or it may undergo changes (re-invention), or it may be discontinued.
It is important to note that the decision process for adoption of electronic publications is complicated by the fact that a contingent, collective innovation decision is required. Because of the interactive nature of electronic publishing, readers, authors, and publishers are interdependent. Adoption of electronic publishing thus will require a critical mass of adopters from all three groups in order to become self-sustaining. However, once general adoption has occurred (i.e. an electronic journal exists and is used), individual readers and authors must still make individual decisions to adopt the technology in order for the innovation to be maintained.
At the present time, individuals in the scientific community are generally either at the knowledge or persuasion stages. While some electronic publications such as Psycoloquy and OJCCT have been functioning for several years, they are generally still viewed by observers and many participants as trials. Current adopters may be largely characterized as innovators or early adopters using Rogers (1995) categories. If this is true, Rogers suggests that these early adopters will differ from later adopters (or by definition some sort of community average) in a variety of ways. The authors, readers, and publishers adopting first should control larger units and possess greater ability to absorb loss than later adopters. Individuals making early adoption decision should show greater empathy, less dogmatism, greater rationality, and greater ability for abstract thought than those adopting later. Early adopters should demonstrate a more positive attitude to change, have higher aspirations, enjoy greater social participation, demonstrate more interconnectedness, be more cosmopolite, have greater contact with change agents, be more exposed to interpersonal communication, more actively seek information regarding the innovation, and possess a higher degree of opinion leadership within their communities relative to later adopters.
Different segments of the scientific publishing community have diverse needs, interests and resources, which perhaps explains the diversity of trials and the slow rate of diffusion of the innovation to date. Rogers (1995) argues that innovations can be characterized along five dimensions relative to the innovation-decision process: relative advantage, compatibility, complexity, observability, trialability. Because the decision-making community consists of at least three segments, and four if libraries are included, different segments can be expected to have varying perceptions of the innovation along these dimensions. Thus it would be expected that relative influences of these will be found.
Manipulation of innovation cluster characteristics and membership in the cluster in order to optimize factors that influence the decision process is illustrated by the development decisions in the TULIP project and OJCCT. When encouraging users to adopt electronic publications, the change agents involved with the TULIP project (library staff) succeeded in increasing the observability and trialability of electronic publishing for the user segment by targeting publicity to potential user groups and offering accessible demonstrations. By working with electronic counterparts of existing print journals, they increased the compatibility of the innovation for their users. This may have worked against adoption however. Complexity remained high and relative advantage was probably decreased. The information offered was not unique and varied from print counterparts only in distribution mode and the fact that the format was electronic. Authors as a segment did not have to adopt the innovation at all as they were carried over to the new publication mode automatically.
Rogers (1995) also suggests that this social structure of the communities involved in diffusion influences the rate of adoption of innovations. Various scholarly communities differ in their social structure, and this will doubtless influence the rate of adoption that is eventually observed. A complementary body of work exists which has studied the formal and informal communication of scientists, sometimes using the diffusion paradigm (Crane, 1972; Crawford, 1971; Garvey, 1979). Much of this work has related to the description and analysis of the functioning of invisible colleges which utilize a range of largely informal communication channels to disseminate information and advance research fronts. In a reanalysis of work done by Garvey and others, Cronin suggests that research fronts differ quite strongly in the nature of their social structures and informal communication networks (Cronin, 1982). Cronin argues that the existence of strong social organization and frequent informal communication are not necessarily highly correlated. This complexity of social organization and communication structures between various fields within scientific disciplines suggests that electronic publishing may be adopted at quite different rates in different scientific communities. It may even be possible to determine the community characteristics that most strongly predict rapid adoption of electronic publishing.
Both advocates of electronic publishing and the results of some early experiments suggest that the innovation cluster represents such fundamental change in scholarly communication that many, if not most of the consequences of adoption are not yet foreseeable. In spite of this situation, all parties involved seem convinced that some form of these innovations will be adopted by all scholarly communities in the fairly near future. This acceptance of the inevitability of diffusion should not preclude study either of the adoption decision or the early impacts of the innovation. It is likely that the impacts of adoption will vary depending on continuing development within the innovation cluster and on the mix of cluster members adopted.
In discussing the characteristics of an innovation cluster, Rogers (1995) provides an illuminating description of the development of hard tomatoes for mechanical harvesting in California. Mechanical harvesting was developed in the 1960s when the availability of cheap labor was threatened by Congressional legislation. Concomitant with the development of the large machine harvester, hard tomatoes were developed to withstand the rigors of mechanical picking. The large machine harvester and the hard tomatoes formed an innovation cluster in that although growing hard tomatoes does not require mechanical harvesting, mechanical harvesting required the simultaneous adoption of hard tomatoes.
The consequences of the adoption of the innovation cluster had greater and more varied effects than anticipated. Although the mechanical harvester addressed the anticipated problem of the loss of cheap labor for harvesting, the high cost of the large picking machine forced thousands of small tomato farmers out of business. Furthermore, the hard tomatoes were less appealing to American consumers who prefer soft tomatoes to hard ones. Rogers argues that many of these consequences were determined in part during the development of the innovation prior to adoption. Perhaps if the initial researchers had considered some of these additional socioeconomic factors, the development of the technology and the diffusion of the innovation cluster might have turned out differently. As Rogers (1995) notes, "The decisions and activities occurring in the R&D phase of the technology development process directly affect the later diffusion phase." In this regard, there is a potentially large benefit in examining the development of electronic publishing before substantial and potentially irrevocable choices are made. Perhaps such an examination could be used to anticipate outcomes and address potential problems or concerns before great investments in time and money are put into projects which may have less than optimal value to the ultimate adopters.
Diffusion of innovation theory provides a powerful framework for understanding the innovation of electronic publishing as a cluster of inter-related innovations. This is helpful for explaining many of the ways in which electronic publishing projects have developed to date. However, it is provides a useful model for exploring the processes involved in accepting various instantiations of the cluster. The understanding of the decision to accept an electronic publication as contingent upon acceptance by several groups with often competing interests may make it easier to identify the interests of the groups involved: authors, readers, libraries, and publishers. This work will serve as a foundation for further exploration of the development and acceptance of specific electronic publishing projects.
Bailey, C. W., Jr. (1994). Scholarly electronic publishing on the Internet, the NREN, and the NII: charting possible futures. Serials Review, 20, 7-16.
Berge, Z. L., & Collins, M. P. (1996). "IPCT Journal" Readership Survey. Journal of the Association for Information Science, 47, 701-710.
Bishop, A. P. (1995). Scholarly journals on the net: a reader's assessment. Library Trends, 43, 544-70.
Borghuis, M., Brinckman, H., Fischer, A., Hunter, K., van der Loo, E., ter Mors, R., Mostert, P., & Zijlstra, J. (1996). Tulip Final Report : Elsevier Science.
Borman, S. (1993). Advances in electronic publishing herald changes for scientists. Chemical and Engineering News, 71, 10-24.
Crane, D. (1972). Invisible colleges: diffusion of knowledge in scientific communities. Chicago: University of Chicago Press.
Crawford, S. (1971). Informal communication among scientists in sleep research. Journal of the Association for Information Science, 22, 301-310.
Cronin, B. (1982). Invisible colleges and information transfer: a review and commentary with particular reference to the social sciences. Journal of Documentation, 38, 212-36.
Denning, P. J., & Rous, B. (1995). The ACM electronic publishing plan. Communications of the ACM, 38, 97-103.
Duranceau, E. F. (1995). The economics of electronic publishing. Serials Review, 21, 77-90.
Gardner, W. (1990). The electronic archive: scientific publishing for the 1990s. Psychological Science, 1, 333-41.
Garvey, W. D. (1979). Communication: the essence of science. Oxford: Pergamon Press.
Getz, M. (1992). Electronic publishing: an economic view. Serials Review, 18, 25-31.
Harnad, S. (1990). Scholarly skywriting and the prepublication continuum of scientific inquiry. Psychological Science, 1, 342-44.
Harrison, T., & Stephen, T. D. (1995). The electronic journal as the heart of an online scholarly communication. Library Trends, 43, 592-608.
Keyhani, A. (1993). The Online Journal of Current Clinical Trials: an innovation in electronic journal publishing. Database, 16, 14-23.
Lancaster, F. W. (1995). Attitudes in academia toward the feasibility and desirability of networked scholarly publishing. Library Trends, 43, 741-52.
Lynch, C. A. (1994). Scholarly communication in the networked environment: reconsidering economics and organizational missions. Serials Review, 20, 23-30.
Okerson, A. (1991). The electronic journal: what whence, and when? The Public-Access Computer Systems Review, 2, 5-24.
Olsen, J. (1993). Electronic Journal Literature: Implications for Scholars. Westport, CT: Mecklermedia.
Piternick, A. B. (1989). Attempts to find alternatives to the scientific journal: a brief review. Journal of Academic Librarianship, 15, 260-66.
Robison, E. C. (1996). Architecture, graphics, and the net: a short history of Architronic, a peer-reviewed e-journal. Public-Access Computer Systems Review, 7, http://info.lib.uh.edu/pr/v7/n3/roib7n3.html.
Rogers, E. M. (1995). Difusion of innovations. (4th ed.). New York: The Free Press.
Rowland, F., McKnight, C., Meadows, J., & Such, P. (1996). ELVYN: the delivery of an electronic version of a journal from the publisher to libraries. Journal of the Amercican Society for Information Science, 47, 690-700.
Schaffner, A. C. (1994). The future of scientific journals: lessons from the past. Information Technology and Libraries, 13, 239-247.
Schauder, D. (1994). Electronic publishing of professional articles: attitudes of academics and implications for the scholarly communication industry. Journal of the American Association for Information Science, 45, 73-100.
Schwartz, C. A. (1994). Scholarly communication as a loosely coupled system: reassesssing prospects for structural reform. College and Research Libraries, 55, 101-117.
Stewart, L. (1996). User acceptance of electronic journals: interviews with chemists at Cornell University. College and Research Libraries, 57, 339-349.
Stodolsky, D. S. (1995). Consensus journals: invitational journals based upon peer review. The Information Association, 11, 247-60.
Sweetland, J. H. (1992). Humanists, libraries, electronic publishing, and the future. Library Trends, 40, 781-803.
Taubes, G. (1996). Electronic preprints point the way to 'author empowerment'. Science, 271, 767-68.
Tenopir, C. (1995). Authors and readers: the keys to success or failure for electronic publishing. Library Trends, 43, 571-91.
Willis, K., Alexander, K., Gosling, W. A., Peters, G. R., Jr., Schwarzwalder, R., & Warner, B. F. (1994). TULIP - the university licensing program: experiences at the University of Michigan. Serials Review, 20, 39-47.
1 The Tulip project, The Red Sage Project, Psycoloquy, and the Online Journal of Current Clinical Trials are some of the most frequently discussed projects. These projects are described in (Borman, 1993). More recent information about Tulip is available from (Borghuis et al., 1996) and Red Sage from. (Harnad, 1990) provides a more detailed description of Psycoloquy while (Keyhani, 1993) provides the same service for the Online Journal of Current Clinical Trials.
2 (Bailey, 1994; Cronin & McKim, 1996; Denning & Rous, 1995; Duranceau, 1995; Gardner, 1990; Getz, 1992; Harnad, 1990; Harrison & Stephen, 1995; Keyhani, 1993; Lynch, 1994; Olsen, 1993; Robison, 1996; Schaffner, 1994; Schwartz, 1994; Stewart, 1996; Stodolsky, 1995; Taubes, 1996; Willis et al., 1994) explicitly described innovations included in the typology. Many additional articles discussing electronic publishing were examined but did not contribute innovations.
3 TULIP stands for The University Licensing Program and was sponsored by Elsevier Science. Nine university libraries participated in developing electronic versions of existing Elsevier journals in material science.