Science has developed from a solitary pursuit into a team-based collaborative activity and, more recently, into a multidisciplinary research enterprise. The increasingly collaborative character of science, mandated by complex research questions and problems that require many competencies, requires that researchers lower the barriers to the creation of collaborative networks of experts, such as communities of practice (CoPs).
The aim was to assess the information needs of prospective members of a CoP in an emerging field, dental informatics, and to evaluate their expectations of an e-community in order to design a suitable electronic infrastructure.
A Web-based survey instrument was designed and administered to 2768 members of the target audience. Benefit expectations were analyzed for their relationship to (1) the respondents’ willingness to participate in the CoP and (2) their involvement in funded research. Two raters coded the respondents’ answers regarding expected benefits using a 14-category coding scheme (Kappa = 0.834).
The 256 respondents (11.1% response rate) preferred electronic resources over traditional print material to satisfy their information needs. The most frequently expected benefits from participation in the CoP were general information (85% of respondents), peer networking (31.1%), and identification of potential collaborators and/or research opportunities (23.2%).
The competitive social-information environment in which CoPs are embedded presents both threats to sustainability and opportunities for greater integration and impact. CoP planners seeking to support the development of emerging biomedical science disciplines should blend information resources, social search and filtering, and visibility mechanisms to provide a portfolio of social and information benefits. Assessing benefit expectations and alternatives provides useful information for CoP planners seeking to prioritize community infrastructure development and encourage participation.
Over the centuries, science has developed from a solitary pursuit into a team-based collaborative activity and, more recently, into a multidisciplinary research enterprise [
Biomedical research follows this trend closely, due in large part to federal funding initiatives such as the National Institutes of Health (NIH) Roadmap, which encourages the formation of multidisciplinary research teams as outlined in its “Research Teams of the Future” theme [
However, the emergence of e-communities is not limited to multidisciplinary research teams but can be observed in many different contexts. E-communities have long been used to support collaboration among professionals and researchers [
E-communities can be characterized according to social, commercial, or professional orientation [
Characterization of e-communities (derived from [
CoPs focus on one domain of knowledge and the accumulation of knowledge and expertise in this domain over time [
Compared with the research performed on social and commercial e-communities and on professional e-communities focused on product development or services or on learning, research on CoPs lags behind. A thorough search for literature evaluating how well these systems facilitate the initiation of collaborations yielded no results. Judging from anecdotal evidence, systems of this type currently do not play a significant role in helping researchers establish collaborations. However, it is this type of e-community that is crucial for the transformation of biomedical research. Little is known about how socially embedded benefits can be exploited for the formation of CoPs. However, this is what programs like the CTSA aspire to, advancing science through communication among scientists from different fields with disparate primary research agendas. The research described in this paper focuses on the role e-communities can play in the genesis and growth of new or loosely formed fields or disciplines.
The field examined in this case study is dental informatics (DI), which, unlike its parent discipline, biomedical informatics, can still be characterized as a nascent discipline [
To these ends, a global e-community, the Dental Informatics Online Community (DIOC), is being established (
Screenshot of Dental Informatics Online Community (DIOC) home page
The DIOC, like any other new CoP, first needs to attract and retain a critical mass of participants by, for instance, widely advertising the expected benefits of participation. Unlike traditional information systems, a CoP depends on volunteers to provide content. Thus, after attracting participants, CoPs need to foster active participation. Studies of participation demographics in multi-user communities and social networks have found that between 46% and 82% of users are lurkers who never contribute [
The first step in attracting participants to a new CoP and then transforming many of them into active contributors is to determine the information needs of the target audience. There is general recognition that a needs assessment is the first step in any project that aims at providing useful information for a specific target audience [
This analysis leads to three main research questions:
Which information resources do researchers currently prefer to use?
How can their current professional relationships be described?
What are their expectations of a CoP, and how are these influenced by factors such as amount of participation necessary for a sustainable e-community and level of involvement in funded research?
The answers to these questions can assist with outlining the basic requirements for an e-community whose goal is to accelerate the emergence of a new discipline. While other successful e-communities could partially be used to model the DIOC, creating a community for a field in its formative stages requires more than just copying and pasting features and functions of e-communities for well-established disciplines. Thus, a needs assessment of prospective members was undertaken.
A review of the literature did not identify an existing instrument suitable for determining information needs and expected benefits. Thus, our first task was to develop such an instrument. Informal interviews with a convenience sample of four active DI researchers suggested some common information needs and revealed a strong desire for peer communication. Problems they identified with finding information sources as well as information needs identified in published studies were used as the starting point for an original survey instrument. These initial items were then developed and refined using Dillman’s Tailored Design Method [
An expert group (three DI faculty [TKS, HS, TPT], three medical librarians [PMW plus two others], one business school faculty member [BSB], and one business school doctoral student [XW]) provided qualitative feedback. As a result of their evaluation, two questions were dropped, 12 were revised, and the texts of the preamble and email invitation were altered.
Nine volunteers from the target population participated in an evaluation using the Retrospective Thinkaloud protocol as suggested by Sudman at al [
asked them to answer one survey question at a time
engaged them in a short follow-up discussion after each answer
inquired about the methods used to arrive at each answer
logged their answers, problems, or comments
solicited final comments and general suggestions
Evaluation of phone interview data resulted in further revision of seven of the 20 survey questions: in four cases, wording was not sufficiently comprehensive; in three, questions were too specific; in two, questions were misinterpreted. In addition, two more questions were eliminated, and two questions were combined into one.
The final version of the survey instrument included 17 items that were presented on one screen: five demographic questions, including current position; one question on expectations regarding the DIOC; six questions regarding professional relationships; and four questions about information-seeking behavior. There was also a general comments section at the end of the survey.
Three question formats were used. Two questions were open-ended, asking for extended text input; five questions were open-ended, with short answers such as age; and nine questions provided multiple-choice options. The question regarding participants’ expectations branched differently depending on whether or not they had already signed up for the DIOC; those who had signed up were also asked how they had learned about it (see the Multimedia Appendix).
The study was approved by the University of Pittsburgh’s Institutional Review Board in May 2006.
To increase the likelihood that the survey would provide representative data encompassing the needs of all people interested in DI, the composition of the prospective target audience was first determined. In addition to including clusters of people easily accessed through established gatherings such as the American Dental Education Association (ADEA) TechnoFair, an annual teaching technology showcase event of dental educators, we wanted to cover the possibility of unanticipated subgroups that might have their own membership or meeting organizations. To that end, we analyzed a set of 620 Medline abstracts identified for a 2003 study [
Distribution of target population across interest/source groups
Group Description* | Email |
Survey Respondents, No. (%) |
Personally approached at AADR, ADEA 2006 | 113 | 28 (24.8) |
Authors of 620 DI papers | 910 | 58 (6.4) |
AMIA DI working group member list | 44 | 13 (29.5) |
IMIA DI working group member list | 133 | 24 (18.0) |
Bioinformatics researchers with dental interest | 11 | 3 (27.3) |
ADEA TechnoFair authors (2004, 2005, 2006) | 369 | 48 (13.0) |
Current DIOC members | 211 | 92 (43.6) |
2003 DI conference participants | 82 | 15 (18.3) |
MLIS community | 110 | 6 (5.6) |
MLA (randomly selected 385 of the 3850-member directory) | 385 | 6 (1.6) |
280 funded informatics researchers (randomly selected 100) | 100 | 1 (1.0) |
9000 funded dental researchers (randomly selected 300) | 300 | 14 (4.7) |
Total | 2768 | – |
Total after eliminating duplicates | 2609 | – |
Total after eliminating duplicates and validating | 2303 | 256† (11.1) |
*AADR, American Association for Dental Research; ADEA, American Dental Education Association; DI, dental informatics; AMIA, American Medical Informatics Association; DIOC, Dental Informatics Online Community; MLIS, Master of Library and Information Science; MLA, Medical Library Association.
†Total number of respondents is smaller than sum of group respondents because some individuals belong to more than one group.
Email addresses for individuals in the groups were obtained using two main approaches. Where member directories for organizations such as the American Medical Informatics Association (AMIA) DI working group were accessible, addresses were extracted directly from them. If member directories were not accessible, names and institutional affiliations were extracted from other publicly available sources; for example, current email addresses of the authors of 620 known DI papers from Medline [
At the time of the survey, the DIOC website had been operating and accepting registrations for 6 months. Although many DIOC features were not yet functional, 211 people had registered after finding the site either through publicity or their own search. These individuals were also invited to participate in the survey.
All 2768 email addresses in the combined groups (see
It was incidentally noted that after merging the 12 groups, there were only 158 duplicate addresses, indicating a very shallow overlap among target audience sectors. The majority of people who had not signed up for the DIOC (2354, 98.1%) belonged to just one of the sampled organizations; 40 were members of two organizations, and 5 belonged to three organizations. Among the 211 DIOC members, 136 (64.4%) did not belong to any other organization; 61 belonged to one other organization, 11 belonged to two others, and 3 belonged to three others. These observations are consistent with the common characterization of DI as a diverse but somewhat fragmented community. Overall, then, the sample seemed to include both a very small core of widely active participants and a large body of peripherally involved individuals.
In order to calculate a more accurate response rate, we tried to filter out nonexisting email addresses by programming an add-on to Sendmail (version 8.13.1; Sendmail Inc, Emeryville, CA, USA) and emailing the invitations from the server it ran on (Linux 2.6.9, Red Hat 3.4.5/Apache 2.2.0; Red Hat Inc, Raleigh, NC, USA; Apache Software Foundation, Forest Hill, MD, USA). The add-on program recorded and flagged 306 email addresses as nonexistent. After this process, 2303 unique email addresses remained. However, it was not possible to detect email accounts that, while technically operational, had been abandoned by users. As a result, the response rates reported here are biased low.
A Web-based format was chosen for the survey instrument because it significantly reduces turnaround time compared with mail surveys [
Invitations to complete the survey were emailed and included a unique access code to prevent both duplicate entries and completion by people who were not part of the target audience. Prospective participants were informed of how long the survey would take, who the investigators were, and that the data would be used for scholarly purposes only. Incomplete surveys could be submitted by respondents since no validation of user entries was performed. Thus, response rate for each question was different, as reported in the survey results below.
The initial invitation was emailed on June 1, 2006. A reminder was sent on June 14, 2006, and a final reminder was sent on July 10, 2006. No incentives were provided to any respondents.
After the survey closed on August 10, 2006, all response data in the MySQL database were exported to an MS Excel (Microsoft Corporation, Redmond, WA, USA) spreadsheet stored on a secure local file server. The majority of the survey questions required quantitative responses and could thus be analyzed with little or no additional manipulation. The open-ended questions regarding expected benefits of the CoP were coded into categories by two raters [BB, HS]. After agreeing on a 14-category coding scheme, both raters independently coded all individual responses. Disagreements on coding for specific items were resolved through discussion.
Analysis of the data included descriptive characterization of information-seeking and collaboration-related needs, examination of differing expectations within meaningful subsets, and identification of respondent clusters with distinctive expectations for a research-oriented online community. Comparison of the subsets was based on chi-square tests of difference in the relative proportions of the reported expectations. A two-step cluster analysis (implemented in SPSS version 15.0; SPSS Inc, Chicago, IL, USA) was used to determine the degree of homogeneity in benefit expectations. This exploratory procedure uses comparisons of individual responses (in this case, the benefits expected by each respondent) to identify sets of similar individuals. Examination of relative scores and
The response rate of 11.1% (256/2303) is based on the validated, unique email addresses. Of the 211 individuals already signed up as DIOC participants, 92 (44% of group and 36% of all respondents) completed the survey (see
On average, respondents were 46.4 years old, had held their current title for 7.9 years, and had been at their current institution for 11.6 years. The 249 respondents to the question on country of residence reported living in 30 different countries (
Distribution of respondents’ country of residence (partial list, only countries mentioned at least three times)
Country | No. (%) |
United States | 139 (54.3) |
Germany | 15 (5.9) |
Canada | 10 (3.9) |
United Kingdom | 7 (2.7) |
Netherlands | 7 (2.7) |
India | 6 (2.3) |
Australia | 4 (1.6) |
Sweden | 4 (1.6) |
Italy | 4 (1.6) |
Japan | 3 (1.2) |
Missing responses | 7 (2.7) |
Total number of respondents | 249 (97.3) |
Total | 256 (100) |
Distribution of respondents’ academic positions (partial list, only positions mentioned at least twice)
Academic Position | No. (%) |
Full professor | 36 (14.1) |
Associate professor | 35 (13.7) |
Department chair/CEO/director | 25 (9.8) |
Postgraduate student | 21 (8.2) |
Dental practitioner | 18 (7.0) |
Scientist | 17 (6.6) |
Consultant | 13 (5.1) |
Administrator | 11 (4.3) |
Librarian | 7 (2.7) |
Dean | 6 (2.3) |
Predoctoral student | 3 (1.2) |
Dental hygienist | 2 (0.8) |
Missing responses | 25 (9.8) |
Total number of respondents | 231 (90.2) |
Total | 256 (100) |
Use of information sources*
Information Source | Frequently, No. (%) | Sometimes, No. (%) | Never, No. (%) | Total |
Medline (via Ovid, |
196 (80.3) | 35 (14.3) | 13 (5.3) | 244 |
Internet search engines |
186 (83.4) | 35 (15.7) | 2 (0.9) | 223 |
Online journals (e-print, full-text archives of print journals, etc) | 184 (78.6) | 48 (20.5) | 2 (0.9) | 234 |
Print journals | 114 (47.5) | 117 (48.8) | 9 (3.8) | 240 |
Books from your personal collection | 103 (44.4) | 113 (48.7) | 16 (6.9) | 232 |
Conferences, lectures, etc | 94 (40.7) | 134 (58.0) | 3 (1.3) | 231 |
Researchers within my institution | 89 (38.7) | 115 (50.0) | 26 (11.3) | 230 |
Researchers from other institutions | 70 (30.7) | 143 (62.7) | 15 (6.6) | 228 |
Books from/in libraries | 61 (26.3) | 137 (59.1) | 34 (14.7) | 232 |
Bibliographic databases such as… |
61 (26.5) | 93 (40.4) | 76 (33.0) | 230 |
Newsletters | 60 (26.0) | 127 (55.0) | 44 (19.0) | 231 |
National or local media (newspapers, television, etc) | 51 (22.0) | 114 (49.1) | 67 (28.9) | 232 |
Other information source: which? | 48 (60) | 32 (40) | N/A | 80 |
IEEE Xplore | 20 (9.4) | 41 (19.3) | 151 (71.2) | 212 |
*Responses to the following question: “How often do you use the following information sources when trying to find professional information?”
Asked about the existence and use of an institutional library, 213/251 respondents (84.9%) indicated that they have access to one, and 194 (91.1% of those indicating access) do use it either physically or virtually.
There were 162 responses to an open-ended question regarding the manner in which the respondents find out about research funding. Funding resources were identified mostly through visits to known funding agencies’ websites, frequently those of NIH. Next in frequency were various forms of intra-institutional notification; personal communication, including not only formal contact but also informal word of mouth; and use of general Web search engines. Among the 16 resources that were categorized as aggregating services, Community of Science was mentioned most often.
Respondents were asked about collaboration, with collaborator defined as “co-author, co-investigator, consultant to a specific project” (Question 3). During the previous 12 months, 193 respondents had, on average, worked with 10 collaborators.
Origin of collaborators during the past 12 months (multiple selections were permitted)
Options for Origin of Past Collaborators | No. (%) |
Come from my department | 179 (92.7) |
Come from other institutions with faculty specializing in my area of interest | 173 (89.6) |
Come from my institution, outside my department | 172 (89.1) |
Are people with whom I have collaborated in the past | 170 (88.1) |
Are people with whom I have conducted relevant research | 133 (68.9) |
Are people whom I met at conferences, conventions, etc | 119 (61.7) |
Are people to whom I was introduced to by a colleague | 111 (57.5) |
Other | 38 (19.7) |
When asked where they usually find research assistants (Question 4), most of the 248 respondents reported getting help from inside their institution (mentioned 86 times, 34.7%), from past helpers (mentioned 74 times, 29.8%), or from inside their department (mentioned 69 times, 27.8%) rather than from recruitment services within (mentioned 43 times, 17.3%) or outside (mentioned 26 times, 10.5%) their organization.
On average, respondents attend five professional meetings per year (based on 245 respondents to Question 7). Relevance of the meeting agenda to one’s general research interests, relevance to particular research projects, and potential for networking with fellow researchers were the crucial criteria used in deciding meeting attendance (
Factors influencing conference attendance*
Factor | Very Important, No. (%) | Somewhat Important, No. (%) | Not Important, No. (%) |
Relevance of agenda to my general research interests | 168 (65.6) | 51 (19.9) | 7 (2.7) |
Relevance of agenda to a particular research project | 122 (47.7) | 85 (33.2) | 14 (5.5) |
Conference features an esteemed researcher | 48 (18.8) | 121 (47.3) | 49 (19.1) |
Likelihood of attendees’ research interests coinciding with my own | 82 (32.0) | 108 (42.2) | 31 (12.1) |
Networking with fellow researchers | 109 (42.6) | 90 (35.2) | 21 (8.2) |
Availability of funding to support attendance | 88 (34.4) | 73 (28.5) | 60 (23.4) |
Ability to present my own work | 92 (35.9) | 91 (35.5) | 37 (14.5) |
Other | 34 (13.3) | ||
Missing responses | 7 (2.7) | ||
Total number of respondents | 249 (97.3) | ||
Total | 256 (100) |
*Responses to the following question: “To what degree do the following factors influence whether you attend a particular conference or not? (Rate the factors.)”
Respondents were asked if they belonged to specific dental and informatics organizations (Question 9). They could augment their response by entering up to three additional organizations; 130 respondents (56.3%) belonged to the International Association for Dental Research (IADR), 97 (42.0%) to the American Dental Education Association (ADEA), and 77 (33.3%) to the American Dental Association (ADA). A total of 88 respondents were members of one of the listed organizations, 59 of two organizations, and 30 of three organizations. The most common write-in choices were European dental research and medical specialty organizations.
Participants who had already signed up for the DIOC were asked about what kinds of benefits they expected from their involvement. Those who had not signed up were asked how they thought an e-community might help them with their research; 64% (164/256 respondents, both groups combined) reported at least one type of expected benefit. The two raters coded the individual responses on a 14-category coding scheme (Kappa = 0.834), concentrating on how benefit expectations related to (1) the respondents’ willingness to participate in the DIOC and (2) how this willingness was related to involvement in funded research (
Individuals who had already signed up for the DIOC tended to expect more specific benefits from the community than those who were not yet registered, including general information, identification of experts, networking with peers, advocacy support, and career development (
Comparison of expected benefits mentioned by different groups
Benefit Category | DIOC Membership | Research Funding | Total, No. (%), (n = 164) | ||||
Non-Member, No. (%), (n = 97) | Member, No. (%), (n = 67) |
|
Funded, No. (%), (n = 115) | Not Funded, No. (%), (n = 49) |
|
||
Information Benefits | 72 | 63 | 93 | 42 | 135 | ||
General information | 38 (39.2) | 47 (70.1) | < .001 | 56 (48.7) | 29 (59.2) | .15 | 85 (51.9) |
Funding information | 18 (18.6) | 4 (6.0) | .02 | 20 (17.4) | 2 (4.1) | .02 | 22 (13.4) |
Specific topic | 10 (10.3) | 7 (10.4) | .59 | 12 (10.4) | 5 (10.2) | .60 | 17 (10.4) |
Teaching materials | 4 (4.1) | 2 (3.0) | .53 | 3 (2.6) | 3 (6.1) | .25 | 6 (3.7) |
Data sharing | 2 (2.1) | 3 (4.5) | .33 | 2 (1.7) | 3 (6.1) | .16 | 5 (3.1) |
Social Benefits | 45 | 69 | 74 | 40 | 114 | ||
Peer networking | 21 (21.6) | 30 (44.8) | .00 | 35 (30.4) | 16 (32.6) | .46 | 51 (31.1) |
Identification of potential collaborators and/or research opportunities | 19 (19.6) | 19 (28.4) | .13 | 30 (26.1) | 8 (16.3) | .12 | 38 (23.2) |
Advocacy support | 2 (2.1) | 9 (13.4) | .01 | 5 (4.3) | 6 (12.2) | .07 | 11 (6.7) |
Expert identification | 1 (1.0) | 6 (9.0) | .02 | 2 (1.7) | 5 (10.2) | .03 | 7 (4.3) |
Participation in the field | 2 (2.1) | 5 (7.5) | .10 | 2 (1.7) | 5 (10.2) | .03 | 7 (4.3) |
Instrumental Benefits | 3 | 8 | 7 | 4 | 11 | ||
Career development | 1 (1.0) | 7 (10.4) | .01 | 4 (3.5) | 4 (8.2) | .19 | 8 (4.9) |
Recruiting | 2 (2.1) | 1 (1.5) | .64 | 3 (2.6) | 0 (0.0) | .34 | 3 (1.8) |
Other Benefits | 23 | 11 | 26 | 8 | 34 | ||
Uncertain | 17 (17.5) | 3 (4.5) | .01 | 18 (15.6) | 2 (4.1) | .03 | 20 (12.2) |
Unclassifiable | 6 (6.2) | 8 (11.9) | .16 | 8 (7.0) | 6 (12.2) | .21 | 14 (8.5) |
Average number of benefits cited per respondent | 0.87 | 1.64 | 1.24 | 0.99 |
*Determined by chi-square analysis.
Benefit clusters
The approximately 70% of respondents who participate in funded research were significantly more likely to expect the DIOC to be a source of funding information and opportunities (see
Active researchers were significantly more likely than non-researchers to express uncertainty concerning the potential benefits of participation in the DIOC. Number of collaborators was also positively correlated with the likelihood of a respondent reporting uncertainty (Spearman correlation = 0.229,
Overall, the most frequently expected benefits from participation in the DIOC were general information (eg, exchange of ideas, keeping well informed), mentioned by 51.9% of respondents; peer networking (eg, finding colleagues with same interests), mentioned by 31.1%; and identification of potential collaborators and/or research opportunities, mentioned by 23.2%. Two-stage cluster analysis revealed five identifiable clusters, each associated with a distinctive collection of benefit expectations (see
Cluster 1: General information
Cluster 2: General information and social benefits (collaboration, peer networking, etc)
Cluster 3: General information and peer networking
Cluster 4: Uncertainty
Cluster 5: General information and collaboration opportunities
General information benefits were widely mentioned across all clusters, but responses regarding social benefits varied. While 58% mentioned some type of social benefit, the cluster analyses suggest that some individuals seek general information alone, while others expect general information combined with peer networking and collaboration opportunities.
In addition to reflecting specific combinations of benefits, the clusters were also distinguished by the characteristics of the individuals associated with them. Individuals in Clusters 1 and 2 tended to have fewer collaborators, be less likely to be doing funded research or using online search resources (Medline, Cochrane Library), and be more likely to have signed up for the DIOC. By contrast, members of Cluster 4 were proportionately more likely to be participating in funded research and to have a higher number of collaborators. Members of Cluster 5 were more likely to have a higher number of collaborators, more likely to be doing funded research, and less likely to have signed up for the DIOC.
Of the individuals who had already signed up for the DIOC, 36/91 respondents (40%) learned of it via an Internet search engine, 26 (29%) received an electronic announcement, 19 (21%) heard about it during a conference, and 22 (24%) specified other sources. Respondents were allowed to select multiple responses for this question.
Any online community must attract a critical mass of involved participants if it is to be sustainable. Individual researchers develop expectations about the benefits of involvement, and these benefit expectations play a significant role in their satisfaction with, commitment to, and, ultimately, participation in an e-community [
Up-to-date information resources are a foundational element of any planned CoP. Access to a variety of timely information was often mentioned as a desirable benefit of involvement in the DIOC by individuals across all clusters. The DIOC’s planned information stores, including general information about DI as well as more specific resources such as a project directory, address this need.
The ideal is for community participants to generate a significant proportion of information resources themselves in such forms as detailed personal profiles, postings to the project directory, and tags, comments, and other annotations. But it may be difficult to quickly attain and then sustain such a goal to a degree that satisfies researchers accustomed to immediate access to plentiful and readily available traditional library resources—not to mention the abundant, if unvetted, resources of the Web. In addition, a CoP needs to offer an attractive breadth and depth of material without creating an undue content creation burden on each participant. Thus, DIOC planners may need to allocate ongoing funds for creation and maintenance of information resources to augment content created by participants, such as a mix of searchable databases and interactive features that can accommodate the anticipated range of user expectations and behavior. Whether this challenge exists for research-oriented CoPs in general is a question for future research.
Just as respondents judge the value of a conference or meeting by how well its topic matches their interests or has particular relevance to a specific research project, potential CoP participants see information resources as an indication of the fit between community activities and their own needs and interests. However, since any one individual is likely to be interested in only a fraction of the available material, CoP architectures and interfaces must include targeting and filtering capabilities. For example, CoPs should aggregate timely information about funding opportunities relevant to their prospective audiences and automatically alert users to new funding opportunities in a targeted manner. These notifications need to match user subject interests and accommodate user preferences [
The high degree of reliance on personal communications and word of mouth (mentioned 34 times out of 162 responses) indicates that even with electronic alerts and Internet searches, personal communication remains a significant source of information about funding opportunities for our respondents. This finding matches the results of earlier studies regarding the information-seeking behavior of dentists [
To support social information seeking and sharing, CoPs need infrastructure for both direct communication (such as document sharing and referrals) and indirect information sharing (via collective tagging or public annotation of informational items). CoPs also should provide contexts such as message boards and forums in which individuals who lack well-developed interpersonal networks can observe and participate in group discussion. Allowing CoP members to annotate, comment on, and discuss information will not only add value to the CoP, but will also encourage the building of trust and knowledge in the community, which are important elements in the development of computer-mediated interaction [
Discipline- and research-oriented CoPs need to support professional relationships among members, enabling individuals to find potential collaboration partners and to form and maintain relationships. Our respondents’ collaborations originated almost equally from inside and outside their own departments and institutions, substantiating the findings of Griffith and Miller [
One key aspect of relationship formation is visibility. Increasing the visibility of individuals, their interests, and their intentions helps catalyze effective professional relationships. Each CoP member should be able to create and maintain a profile accessible to all, enabling subscribers to construct and develop verifiable identities within the community [
In addition to forming collaborative relationships with other individuals or other participant subgroups, individuals also want to develop and maintain awareness of what the overall community is doing. Emerging disciplines usually do not support a standing professional meeting, but CoPs can provide at least a partial substitute for that aspect of scholarly activity and for the networking opportunities generally available at traditional professional meetings. As mentioned above, it is hoped that the DIOC will substitute for a standing DI conference and serve as a professional home for researchers who primarily dedicate their career to this emerging discipline, allowing virtual affiliation without travel. Again, closely linked project and people directories that let members learn about ongoing projects and who is responsible for them are key resources.
Respondents with higher numbers of collaborators and involvement in funded research were more likely to express uncertainty about the benefits of participation. They were more likely to mention general information and collaboration opportunities as expectations, while those with fewer collaborators and no funded research participation mentioned social benefits such as expert identification and advocacy support. These differing profiles, coupled with the significant negative correlation between tenure in an organization and the expectation of general information benefits, underscore the fact that academic online communities such as the DIOC are competing with individuals’ own environments—their networks, institutions, and other immediately available resources.
Unlike traditional information systems, which are typically seen as the only, or at least the primary, source of information of a particular type within an organization, CoPs operate within a much broader, highly competitive social-information ecology. CoPs compete with individuals’ own local resources, so persuading time-pressured researchers to move from habitual exclusive reliance on known resources to exploring new tools and techniques in the interest of improving long-term productivity is a key challenge [
Individuals uncertain about benefits were proportionately more common among those who had not signed up for the DIOC (
However, in complex ecological systems, attempting to “win” simply by direct competition can be a costly approach that often fails. The CoP planner should look for ways in which the presence of related resources and systems supports the goals of the community. For example, the use of online information sources by DI researchers, the emergence of the Internet as an important tool for dentists [
Taken together, these results characterize both the promise and the challenge of academic online communities. On the one hand, CoPs present clear benefits for individuals who are more isolated, less connected, and lacking in access to local institutional resources; these participants can, in return, increase the diversity and impact of an otherwise fragmented discipline such as DI [
Yet the structure of the clusters in the DI community suggests a possible solution. By building a base of commonly valued information resources and providing individuals with the ability to pick and choose the nature of their social engagement with the community, the DIOC can provide an infrastructure that brings together a diverse group of individuals with complementary needs. Identifying the interlocking contribution-benefit pairs allows them to be addressed, and leveraged, during implementation [
A response rate of 11.1% is low but within the expected outcome range [
Data about current position and country of residence show that respondents were well distributed across the spectrum of the intended target audience. The results seem to reflect the fact that interest in DI is spread among many different countries and pursued by people in various academic and clinical positions. However, it is possible that the selection of 12 target audience groups might not be entirely inclusive.
Some of the general comments made on the concluding survey question (“Is there anything else you’d like to tell us?”) criticized our US-centric view. While it is true that most of the professional organizations listed as choices for membership were US-based, the results of our pilot tests did indicate predominance of US respondents. However, a pro-US bias might have influenced question constructs and results.
This study relied on self-reported data, which may be incomplete and/or incorrect. For instance, respondents might have unperceived information needs that they did not report [
We were able to assess the information needs of dental informaticians, researchers, educators, clinicians, and other interested parties. Data on expected benefits of a CoP for DI were collected and evaluated, allowing compilation of requirements for the creation of the DIOC.
The survey itself has increased the awareness of the DIOC project. Casual observation has shown that DIOC registration spiked in the wake of the various survey invitations and reminders.
Future work should focus on validating the instrument used in this study as well as carefully applying our findings to other emerging biomedical research fields such as consumer health informatics.
The authors would like to thank the National Library of Medicine (NLM) for the funding for the Dental Informatics Online Community project (1 G08 LM008667-01 A1). Further acknowledgment goes to Colleen Dugan, predoctoral dental student at the University of Pittsburgh, School of Dental Medicine, for undertaking the task of identifying author email addresses for 620 abstracts; to Noelle Peters, prospective predoctoral dental student, for supporting the extended literature search; and to Marcos Kreinacke, consultant at the Center for Dental Informatics, for programming the add-on to Sendmail. In addition, our thanks go to Patricia F Anderson, Dental Library at the University of Michigan, and Ellen G Detlefsen, School of Information Sciences at the University of Pittsburgh, for input during instrument development. Steven E Poltrock, Boeing Phantom Works, and Jonathan Grudin, Microsoft Research, provided input to the design of
None declared.
Survey instrument, including branching differences
community of practice
Clinical and Translational Science Awards
dental informatics
Dental Informatics Online Community
National Institutes of Health