An underlying aim of research in information systems is to aid the creation of better systems. In this context the word 'system' means the relationships and synergy between people, practices and technology to achieve a goal. Such research is often approached using combinations of research paradigms. In my work I am investigating the usability and social structures needed to make online communities successful. The intersection of these two aspects - usability and social structures - leads to fundamental questions about the nature of communities and communication and applied questions derived from the 'build and test' paradigm of engineering. For instance, the behavior and needs of people in scholarly discussion groups are different from patient support groups. Consequently, understanding the focus and culture of each kind of community is important. However, there is a difference between the need for understanding culture in order to build tools and researching the fundamental basis of online culture in order to shape a social framework for the field. The former is a role for information systems researchers and the latter for sociologists and anthropologists. My work is situated in the information systems paradigm but I draw upon work from other disciplines.

My studies are naturalistic, ethnographic and formative in nature. They involve observation of online behavior, analysis of message content and demographic analysis. I have worked on this topic for two years (see publication list). In collaboration with graduate students, I have built 6-8 online web communities comprising web pages, communications software (chats, bulletin boards, listservs) and other resources. This construction activity is fostering research with a long-term aim to manipulate different designs and social structures in these communities without damaging them. Three Ph.D. students contribute to this research effort.

Research summary

I take a broad view of the term 'community'. Listservs and bulletin boards can support communities provided they are used for discussion and not just for distributing announcements. My work started with these rudimentary environments and is progressing to communities supported by the combinations of software mentioned above. While I consider it important to define terms, I will build on work by sociologists e.g. (Wellman, 1997), and psychologists (Roberts, 1998) who are characterizing the notion of 'community'. The current use of 'online community' to describe all kinds of communities including computer virtual environments (Bruckman, 1994), networked communities (Cohill & Kavanaugh, 1997), and groups supported by listservs and bulletins boards (Preece, 1999a) is too general. But the resolution of terminology, while desirable for the field, is not essential for the work described below.

I build on the work of researchers who have observed expressions of emotion in online textual communication. Initially they were surprised by people's honesty and openness (e.g. Sproull & Kiesler, 1991; Rheingold, 1993; Turkle, 1995). Numerous anecdotes describe examples of people supporting each other in education (Hiltz, 1994); medical (Brennan, Moore, & Smyth, 1991; Brennan & Ripich, 1994; Rappaport, 1996), and emotional support communities (King, 1994). Others describe the strong friendships that develop online (Park & Floyd, 1996; Haythornwaite & Wellman, 1998) and the role of online communities for personal and social exploration,(Turkle, 1995; Bruckman, 1993). My work attempts to identify and classify different types of content in bulletin board and Usenet communications. The first study examined empathy in a single community. Later work seeks to establish how generalizable the findings from the single community study are to other communities. Current and future work is developing demographic profiles, describing moderators' tasks and developing tools to support them, and understanding lurker behavior and needs.

In the first study I analyzed 500 messages from an online support community for people suffering knee injuries (Preece, 1999a; Preece & Ghozati, 1998b) in order to answer the question: 'does empathy occur in this group?' I observed this community for several months prior to doing the study and was surprised to see a high level of empathy and no hostility even though there was no moderation. People seemed to know what the other person is feeling, feel what the other person is feeling; and respond compassionately (Levenson & Ruef, 1992 p. 234). Using this definition and drawing on work on empathy (Ickes, 1993; Etchegoyen, 1991) I was able to provide a theoretical definition for the empathic communication that I witnessed. While empathy can involve negative emotion (i.e. empathizing about anger), I focused upon positive emotions. I developed 4 other category descriptions and did a content analysis (Krippendorff, 1980; Robson, 1993; Weber, 1990) similar to Worth and Patrick's analysis of discussion lists (Worth & Patrick, 1997). In order to achieve good inter-researcher reliability my analysis focused on the predominant content of the message. This methodology is discussed thoroughly in a paper accepted for publication (Preece, 1999a).

I found that 44.8% of the messages were predominantly empathic, 17.4% were factual and 32.0% were personal stories. There were no hostile responses. 5.8% of the messages did not fit any of the categories and contained advertisements or announcements. These results suggest that empathy, as well as factual information exchange, is important in this community. I also found some evidence of gender-related differences in the content of the messages (Preece, 1999a). Women made more empathic comments and the men's comments were more factual, which is supported by a chi-square test (chi-square: 12.18, DF 4, p = 0.05). This result differs from research in face-to-face communication, which is inconclusive (Eisenberg et al., 1989). Although I can not be absolutely certain of the gender of the posters in this study, I was careful to categorize messages that were suspect as 'unclassified'. For example, log-in names that were unusual were discounted and all the messages were reviewed in context to look for signs of gender masking or swapping. Having observed this community for several months prior to doing the study I am confident of this approach but because gender swapping is common online (Bruckman, 1994; Turkle, 1995), further studies are needed in which the gender of posters is known before drawing hard conclusions.

In a follow-up exploratory study of 100 bulletin board and Usenet communities (Preece & Ghozati, 1998a) (Preece, 1999b) we investigated the following questions:
(i) To what extent does empathic communication occur? Do the findings of our first study generalize?
(ii) Is empathy influenced by the community's focus of interest? Is the level of empathy similar in all communities, or are some communities more empathic than others?
(iii) Does the presence of a moderator appear to influence empathy in the community?

2000 messages in total were analyzed, 20 from each community. All the communities were open and did not impose membership regulations. 24 were moderated, 42 were not moderated and 34 could not be classified for moderation. Approximately half were bulletin board communities and the other half were Usenet groups. This is a small sample from the thousands of communities that now exist online, and twenty messages is a tiny slice of the activity going on in each community, but our results help to give a broader picture of online communication. 81% of the communities contained some empathic messages. Over half of the messages in 18% of the communities were empathic.

Although empathic communication was present in all but 19/100 communities, it was much stronger in most of the patient and emotional support communities than in other communities. 78% of patient and emotional support communities have five or more empathic messages, whereas only 7% of the other communities have this many empathic messages. Only 1/19 communities in which no empathy occurred is a patient support community. Most of the communities in which no empathic messages occurred are religious, scientific, sports and cultural communities.

Flaming, equated here with hostility, is also an expression of emotion, but it is not empathic. 36% of the communities contained at least one hostile message. Over a quarter of the messages in 8% of communities were hostile. A chi-square test comparing the types of messages sent in patient and emotional support communities with those sent in the other communities indicates that there is more empathy and less hostility in patient and emotional support communities (Chi- square: 398.55, DF 4, p = 0.001). Anecdotal evidence suggests that hostile communication is associated more strongly with unmoderated communities (Sproull & Kiesler, 1991). In general less hostility and more empathy occurred in the moderated communities, except in patient and emotional support communities, which were strongly empathic even though they contained hostile messages.

Overall, it appears that empathy is common in many communities and so is hostility. The occurrence of each is related to both the topic of interest and the presence of moderators. The surprising finding is how much empathy occurs in patient and emotional support communities regardless of whether they are moderated.

Future directions

The questions we are addressing are related to Kling's 'social informatics' - 'a body of research ... that examines social aspects of computerization' (http://www.slis.indiana.edu/SI/). This term conveys design of technology and social mechanisms.

• What is the role empathic communication in support communities? Does it build a sense of lasting community? Do people communicate privately or off-line? Does trust develop? How threatening is hostility? Does it damage the community? If so, how long does it take to recover? We are studying these and related questions ethnographically and using questionnaires in a Parkinson's disease community.
• What design features are needed in tools to support moderators in different types of communities? Using a taxonomy of moderator roles (Berge, 1992) (Collins & Berge, 1997) we are investigating text retrieval techniques which will form the basis of a tool for identifying flames, spam, advertising etc. We are also examining communication characteristics. For example, preliminary findings suggest that empathic discussion generates broad, shallow threads and factual discussion generates narrow, deep threads (Seabrook & Preece, 1999). Controlled experiments will be conducted in which the tool(s) developed will be tested with experienced moderators of scholarly discussion groups and patient support communities?
• Are there ways of discouraging hostility and other unsociable behavior that do not involve moderators? An alternative way of deterring hostility maybe 'training by example' based on techniques from psychotherapy (Zimmer & Alexander, 1996). Anecdotal reports suggest that techniques for ensuring that participants acknowledge each other's points of view, and check for common understanding appear successful, and are in harmony with common ground theory (Clark & Brennan, 1993). We plan to investigate the effectiveness of this approach in an educational community.
• What is the demography of participants and lurkers in listservs? In this study we are examining posting and lurking behavior using data logging techniques. This work is being complemented by an ethnographic study that aims to answer questions about who lurks and why they are silent. An interesting preliminary finding indicates the deep commitment of many lurkers and their strong sense of 'belonging to a community' (Nonnecke & Preece, 1999).

• What is the accepted terminology in this field? We all know more or less what we mean but agreed terminology is important in any academic field?
• What are the useful theoretical frameworks for understanding online communities? For example, how do common ground theory (Clark & Brennan, 1993), social network analysis (Wellman, 1997; Wellman & Gulia, 1998) and other theories apply in different types of communities?
• What are the characteristics of software design and the social structures that will support the millions of people who will join online communities during the next five years?
• What are the usability criteria, measures and guidelines for supporting software design for different types of communities?
• How does trust develop and what kind of software design and social structures are needed to encourage trust?
• What ethical codes should researchers in this field apply?

Berge, Z. L. (1992). The role of the moderator in a Scholarly Discussion Group (SDG). Available: http://star.ucc.nau.edu/
Brennan, P. F. Moore, S. M., & Smyth, K. A. (1991). ComputerLink: Electronic Support for the Home Caregiver. Advances in Nursing Sciences , 13(4), 14-27.
Brennan, P. F. & Ripich, S. (1994). Use of a Home-Care Computer Network by Persons with Aids. International Journal of Technology Assessment in Health Care, 10(2), 258-272.
Bruckman, A. (1993). Gender Swapping On The Internet. Paper presented at the The Internet Society (INET '93) Conference , San Fransisco, CA.
Bruckman, A. (1994). Panel: Approaches to managing deviant behavior in virtual communities. Paper presented at the CHI'94 Conference Companion, Boston, MA.
Clark, H. H. & Brennan, S. E. (1993). Grounding in communication. In R. M. Baecker (Ed.), Groupware and Computer-Supported Cooperative Work (pp. 222-233). San Francisco, CA: Morgan Kaufmann Publishers.
Cohill, A. M. & Kavanaugh, A. L. (1997). Community Networks: Lessons from Blacksberg, Virginia. Norwood, MA: Artech House.
Collins, M. P. & Berge, Z. L. (1997, March 24-28.). Moderating online electronic discussion groups. Paper presented at the 1997 American Educational Research Association (AREA) Meeting, Chicago, IL.
Eisenberg, N., Fabes, R.A., Schaller M., Miller, P.A. (1989) Sympathy and personal distress: Development, gender differences and inter-relations of indexes. In N. Eisenberg. Empathy and related emotional responses . Jossey-Bass: San Francisco, CA.
Etchegoyen, R. H. (1991). The Fundamentals of Psychoanalytic Technique. New York: Karnac Books.
Haythornwaite, C. & Wellman, B. (1998). Work, friendship, and media use for information exchange in a networked organization. Journal of the American Society for Information Science, 49(12), 1101-1114.
Hiltz, S. R. (1994). The Virtual Classroom: Learning Without Limits via Computer Networks. Norwood, New Jersey: Ablex Publishing Corporation.
Ickes, W. (1993). Empathic Accuracy. Journal of Personality, 61(587-610).
King, S. (1994). Analysis of Electronic Support Groups for Recovering Addicts. Interpersonal Computing and Technology: An Electronic Journal for the 21st Century (IPCT), 2(3), 47-56.
Krippendorff, K. (1980). Content Analysis: An Introduction to its Methodology. Newbury Park: Sage Publications.
Levenson, R. W. & Ruef, A. M. (1992). Empathy: A physiological substrate. Journal of Personality and Social Psychology, 63, 234-246.
Nonnecke, B. & Preece, J. (1999). Shedding light on lurkers in online communities. Paper presented at the Ethnographic Studies in Real and Virtual Environments: Inhabited Information Spaces and Connected Communities . 24-26 January. Edinburgh. (Ed.) K. Buckner. 123-128.
Park, M. R. & Floyd, K. (1996). Making friends in cyberspace. Computer Mediated Communication, 4.
Preece, J. (1999a). Empathic Communities: Balancing Emotional and Factual Communication. Interacting with Computers (Accepted).
Preece, J. (1999b). Empathy Online. Journal of Virtual Reality Research and Design (accepted).
Preece, J. & Ghozati, K. (1998a). In search of empathy online: A review of 100 online communities. Paper presented at the Proceedings of the 1998 Association for Information Systems Americas Conference, Baltimore, USA. 92-94.
Preece, J. & Ghozati, K. (1998b). Offering Support and Sharing Information: A Study of Empathy in a Bulletin Board Community. Paper presented at the Computer Virtual Environments, Manchester, England.
Rappaport, S. H. (1996). Supporting the 'Clinic without Walls' with an event-directed messaging system integrated into an electronic medical record. Paper presented at the Proceedings of the American Medical Informatics Association .
Rheingold, H. (1993). The Virtual Community. Reading, MA: Addison-Wesley Publishing.
Roberts, T. L. (1998). Are newsgroups virtual communities? Paper presented at the CHI 98 Human Factors in Computing Systems, Los Angeles.
Robson, C. (1993) Real World Research. Blackwell: Oxford and Cambridge.
Seabrook, R. & Preece, J. (1999). Techniques for detecting different types of content in textual communication. Information Systems Dept. UMBC 1999 Draft Report.
Sproull, L. & Kiesler, S. (1991). Connections: New Ways of Working in the Networked Organization. Cambridge, MA: MIT Press.
Turkle, S. (1995). Life on the Screen. New York: Simon & Schuster.
Weber, P. W. (1990). Basic Content Analysis . Newbury Park, London: Sage.
Wellman, B. (1997). An electronic group is virtually a social network. In S. Kiesler (Ed.), Culture of the Internet (pp. 179-205). Mahwah, N.J.: Lawrence Erlbaum Associates.
Wellman, B. & Gulia, M. (1998). Virtual Communities as Communities: Net surfers don't ride alone. Chapter 7. In P. Kollock & M. Smith (Eds.), Communities in Cyberspace . Berkeley, CA: Routledge.
Worth, E. R. & Patrick, T. B. (1997). Do electronic mail discussion lists act as virtual colleagues. AMIA, 325-329.
Zimmer, B. & Alexander, G. (1996). The Rogerian Interface: For open, warm empathy in computer-mediated collaborative learning. Innovations in Education and Training International, 33(1), 13-21.

Additional papers about my work

Preece, J. and Ghozati, K. (1999) A Snapshot from 100 Communities: Observations and Explorations of Empathy Online. (Submitted January 1999).
Lazar, J & Preece, J. (1998) Designing and Implementing Web-based Surveys. The Journal of Computer Information Systems. (Accepted)
Lazar, J., Tsao, R. & Preece, J. (1998) One Foot in Cyberspace and the Other on the Ground: A Case Study of Analysis and Design Issues in a Hybrid Virtual and Physical Community. WebNet Journal: Internet Technologies, Applications & Issues (Accepted)
Van Oostendorp, H., Preece, J. and Arnold, B. (1998) Multimedia and Human Empowerment, Interacting with Computers: The Interdisciplinary Journal of Human-Computer Interaction . (Accepted)
Preece, J. (1998) Empathic Communities: Reaching out Across the Web. Interactions Magazine, 32-43, March 1998.
Lazar, J. and Preece, J. (1999) Sampling Methodologies for Using Electronic Surveys to Evaluate Networked Information Resources. ASIS Midyear Conference, May 1999, Los Angeles (submitted)
Preece, J. (1998) Empathy on-line: Is It Different for Men and Women? American Psychological Association Conference, LA, August. Panel.
Lazar, J. and Preece, J. (1998). Classification Schema for Online Communities. Proceedings of the 1998 Association for Information Systems, Americas Conference, 84-86.
Preece, J. (1998) Offering Support and Sharing Information: A Study of Empathy in a Bulletin Board Community, UK Computer Virtual Environments Conference, June 17-19.