Printer Friendly

Visualizing the invisible college: community among authors in top social work journals.

SOCIAL WORK SCHOLARLY knowledge has been created for the social work community. Though some schools of social work, often those with doctoral programs, have had a disproportionately large role in the production of knowledge (Green, 2008), social work education has embraced the idea that practice informs research (Council on Social Work Education, 2008). Furthermore, some social work education programs have adopted the scholarship of teaching (Boyer, 2004) that would inform the content and methods of communication with students. So it has been expected that social work knowledge would involve communication between those who produce it and others, such as the faculty of the less research-oriented programs that educate BSW and MSW social workers, as well as practicing social workers. That communication would make a knowledge community necessary.

The present article was based on research aimed at discovering and depicting a limited part of the knowledge community, with attention to implications for the larger community. The research focused on the network of coauthors that has produced the knowledge published in top-rated social work journals, sometimes called an "invisible college" (Crane, 1972). Once I had visualized this college and explored the quality of scholarship, I undertook speculative analysis on the use of the coauthor community for knowledge production and communication more broadly. Later in this article I offer suggestions for the possible evaluation of scholarship for schools or departments of social work. Generally, I suggest that coauthoring aggregates into a networked community that facilitates the communication of knowledge and that it might be promoted as such.

Emerging Coauthorship and Social Work

Coauthoring that can lead to networks has been documented in many fields, including social work, and appears to be increasing. In several articles Newman (2001a, 2001b, 2001c) found that the average number of authors per article was about three, varying by field from about 1.5 to 3.75. Wuchty, Jones, and Uzzi (2007) used nearly 20 million publication records across many fields to document the increased use of coauthoring, and Moody (2004) documented an increase in coauthoring in sociology. Social work appears to be less aware or appreciative of the coauthor community than other fields. Baker (1992) documented a social work knowledge network based on citations, not coauthors. Discussing the structure of social work research, Williams et al. (2008) presented ANSWER (A Network for Social Work Education and Research), a formal structure comparable to what I describe in the present article. Closer to my approach, Gelman and Gibelman (1999) documented the trend toward multiple authors in social work from 1973 to 1977, and Videka, Blackburn, and Moran (2008) were aware of increasing researcher cooperation. I have pursued the issue of a coauthor network in social work scholarship, which the literature appears to have approached but not directly addressed.

Local and Nonlocal Coauthorship

Close physical location of authors might increase coauthoring. Wineman, Kabo, and Davis (2009) found that being in the same academic department predicted coauthoring, whereas long physical distances between faculty on a campus decreased coauthoring slightly. Berg-Weger and Schneider (1998) said social work faculty located near those of a discipline other than their own were more likely to collaborate with faculty of the other discipline. Zaccala (2004) found that mathematical colleagues in the same academic department sometimes had strong coauthoring ties; her observation and interviews in a mathematics institute confirmed the importance of close physical location of coauthors. On the other hand, Zaccala found that mathematicians also used travel, conferences, or graduate school to create connections. Generally, community can be formed with nonlocal networks (Castells, 2000) and through electronic means (Urry, 2003). So, coauthoring can use local or nonlocal connections as opportunities for coauthoring. The locality of the network is also addressed in the present article.

Network Emergence

Local or not, social networks emerge with identifiable patterns that build community. Barabasi et al. (2002) showed that coauthor ties emerged with preferential attachment, with bias toward the currently attached. That has produced a few highly connected coauthors that in turn tended to connect cliques, small groups in which each author usually had a tie to every other author in that group (Newman & Park, 2003). Preferential attachment combined with cliques has been shown to lead to "small worlds," communities that have short paths to more distant authors via consecutive strings of coauthors (Barabasi et al., 2002; Moody, 2004; Newman, 2001a, 2001b, 2001c; Uzzi & Spiro, 2005). Small worlds have been shown to bring people close together--usually less than six consecutive relationships, or degrees, apart (Watts, 2003). Ultimately, such connections have been found to form "giant components" (Newman, 2002), or networks containing a majority of all authors in related areas of study that make the community more inclusive. Based on this literature, the question of an emerging social work literature giant component containing small worlds is also one I have addressed here.

Benefits and Support of the Coauthor Network

Various benefits of coauthor communities have been suggested. Zaccala (2004) said the mathematicians she studied used their connections to share ideas and develop new ones. Uzzi and Spiro (2005) wrote that small-world networks enabled the creative material to develop, circulate, and become credible. Moody (2004) said sharing expertise in quantitative methods has been associated with coauthoring in sociology. Gelman and Gibelman (1999) saw shared expertise and mentoring as advantages of the increased coauthoring in social work. Guimera, Uzzi, Spiro, and Nunes Amaral (2005) said that coauthor networks produced more citations, a measure of publication quality, than less connected coauthors. The relationship of connections to benefits, however, may not be linear. Uzzi and Spiro (2005) said that although the benefits of small worlds increased into the midrange of connectivity, they turned negative with too much connectivity, possibly because they homogenized the pool of material and led to the sharing of common rather than novel information. Thus, the literature has suggested the question of the optimum level of networking that would be beneficial for social work, also addressed in the present article.

Rewarding Coauthoring and Content

If coauthoring is beneficial, it should be rewarded. Seipel (2003) studied mostly tenured and social work educated faculty in graduate social work education programs and found that being listed as first author or single-authoring refereed articles was most valued. Alternatively, coauthoring might be rewarded by giving substantial credit to coauthors beyond the first author, perhaps using wikis to keep track of authors' contributions, as Dobkin (2009) has suggested. In addition, many articles in the Journal of Social Work Education (JSWE) have discussed the scholarship of teaching. Boyer (2004) asserted that the scholarship of teaching should emphasize the content of the discipline as well as pedagogy. If practice is the content of social work education and, in fact, is social work, how should the scholarship of pedagogy be connected to the scholarship of practice, and should it be rewarded differently than the scholarship of practice? If so, what kind of connections and rewards should be used in what kind of social work education programs? I have included some speculative thinking on these last questions in this article.

Implications of the Literature

The literature reviewed suggested six questions for the research reported in the present article. I brought data to bear on four of these questions.

1. To what degree has a social work invisible college developed?

2. If it has developed, has it been local or cosmopolitan?

3. Has it reached some middle level of clustering that might produce quality publications?

4. Does coauthoring improve publication quality (tentatively answered here with some initial data on citations)?

5. What content should the network communication emphasize?

6. How should the communication and content of scholarship be rewarded?

The latter two questions are not related to data included in the present article but are related to the scholarship of teaching and practice.

To address these questions, I sought the discovery of a skeletal network structure within social work's invisible college. The rationale, based on the literature reviewed, was that prestigious literature would likely have an invisible college improving its quality compared with less prestigious literature; therefore, that was the place to look for community. Furthermore, a well-connected coauthor skeletal network would provide clear visualization compared with a less connected if more complete part of the network. Then, as I observed a discovered skeletal network over time, growth of the network was documented. Furthermore, citations of networked authors could suggest quality achieved by networked scholars, so I identified networked authors' citations as the network grew. Finally, questions of quality and content of scholarship implied policy for evaluating social work faculty in relation to connections and content, so I speculated on these issues.


The general method used in this article was social network analysis (Freeman, 2004). Of the four elements of social network analysis indicated by Freeman, three were included here, namely (1) a structural approach that links actors, or authors in this study; (2) systematic empirical data; and (3) graphic imagery. However, I augmented the structural approach addressed by Freeman with a process-oriented approach that examined network change over time.

The program I used to find and visualize community was a template for Microsoft Excel spreadsheets called NodeXL (Smith et al., 2009). The primary data entered into the NodeXL program were, in the language of the program, ties and vertices. Vertices, also known as nodes, were the authors themselves, and ties were collaboration relationships. Generally, I used the term coauthoring to refer to the coproduction in a single work by any pair of authors, whereas collaboration refers to combined instances of a pair of authors coauthoring in multiple works. Ties were entered into NodeXL by simply listing the two authors of collaboration dyads in horizontally adjacent cells. More elaborately, the ties between the authors studied varied depending on the number of authors for a given article, the cumulative number of times the authors worked together, and the date of their collaboration, all depicted here via NodeXL capabilities.

Identifying Connected and Top-Rated Journal Authors

Many social work journals have been established, and social workers often have published outside the field (Green & Baskind, 2007). Consequently, I operationalized top-rated social work journal literature and narrowed the literature studied to obtain a network small enough to visualize clearly. Sellers, Mathiesen, Perry, and Smith (2004) provided an initial basis, identifying frequently cited and prestigious journals. Highly ranked social work journals on one or both of Sellers et al.'s measures included Social Work (SWK), Journal of Social Work Education (JSWE), Social Work Research (SWR), and Social Service Review (SSR). Cnaan, Caputo, and Shmuely (1994) rated these four among the top five social work journals, Baker (1992) included three of the four in his core structure, Gelman and Gibelman (1999) examined three of the four, and Hodge and Lacasse (2011) included them in the top 10 social work journals. Based on this literature, I deemed these four journals as top-rated.

I used ties between coauthors who had published refereed articles in the four top journals, including instrument development articles in SWR, as the initial data for the present research. I excluded ties of authors of commentary, editorials, viewpoints, book reviews, and notes in any journal. Of course, all single-author articles published in the top-rated journals could not provide data on coauthoring because there were no coauthors; thus they were omitted from the network data. In addition, collaborators with very weak relationships, below .02 on the collaboration tie measure described below, were excluded from the study. Also, I expanded the authors identified in the initial selection to include their lower-rated or unrated publications by initially entering all of their coauthor ties from that additional literature and then limiting that selection as described below. For a given coauthored article or other publication, I entered a tie for every coauthor to every other coauthor, mostly from Internet sources. NodeXL provided a check of duplication of ties, but differences in recording authors' names or initials produced errors. Though Newman (2001c) found such errors in less than 2% of the data, I undertook substantial manual checking of initials and names to ensure author identity. Although this should have eliminated errors, if errors remained in the data they would have resulted in either the inclusion of nonexistent coauthor ties or the absence of existing coauthor ties.

Although I thought it was important to extend the network of authors published in top-rated journals to those published in lower-rated journals as well as books and monographs, the inclusion of too many authors and ties would have hampered visualization of the networks. Eigenvector centrality, computed by NodeXL, was used to assign scores to authors based on direct and indirect ties to other highly connected authors, and I used it to include the most central authors. Eigenvector centrality traditionally has been calculated by arranging all the authors as both row and column marginal headings of a matrix. Any value of one in the matrix, therefore, indicated a coauthor relationship, and a value of zero indicated no such relationship. For each author, eigenvector centrality was based on the sum of those ones for the author and was made a proportion of the average of all neighboring authors, resulting in a score for each author. When the multiple-degree ties were between other well-connected authors, these ties became influential in the value of eigenvector centrality for a given author. This score can be interpreted as the importance of the author in terms of both direct, or first-degree, ties and indirect, or multiple-degree, ties for an author. Degree, a simple count of the direct ties an author has, was also calculated by NodeXL for each author. Borgatti (1995) has discussed degree and eigenvector centrality. It should be noted that NodeXL did not use the weighting described in the visualizations below in the calculations of degree and eigenvector centrality.

After assigning eigenvector scores to top-rated journal authors, I retained those with the highest eigenvector centrality for study. Then a search via Google Scholar identified other coauthors with those initially selected with high eigenvector scores, whether or not they had published in top-rated journals. I added these coauthors to the list and again calculated eigenvector centrality. If any of the additional coauthors moved into the top 25 of the original list as measured by eigenvector centrality, they were included, and the procedure was repeated. This conservative approach added authors only if they were quite well connected. Thus, for the identified authors, the diagrams displayed below omit dozens of nearby connected authors and hundreds of more distant but indirectly connected authors. However, the selection method retained the authors who were likely to be part of the skeleton that held the network together.

Weighting Ties for Visualization

For visual display in the diagrams, I developed weighting procedures for authors who coauthored more than once, and I combined and weighted their efforts to produce collaboration ties. Values for weights other than journal articles were based on publication length, except in the case of reissued or edited books. All journal articles took on equal values, and an edited volume of articles or reissued textbooks were assumed equal to an article. Original books or research reports were credited as one article per 25 pages. It should be noted that the weights were intended to represent the quantity of information produced in a coauthor relationship, not the quality of an author's contributions, which was suggested by citations as described below.

I weighted authors and used author weights to calculate tie weights. I numbered individual authors in reverse order, totaled the author numbers to derive the number of pieces, and allocated pieces according to each author's number to get weights for each. For example, an article with four authors would have authors numbered 4, 3, 2, and 1. Summing those numbers yields 10 pieces of the article. The first author, who has the number 4, gets 4 x .1, or .4, of the article; the second, .3; the third, .2; and the last, .1. The products of the author weights were used to weight the strength of coauthor ties. So in the above example, the fie between the first and fourth authors got a value of .040 (.4 x .1), and the tie between the second and third got a .060 or (.3 x .2) value. These weights were rounded to three places. This resulted in a scale for coauthor tie weights for a single article that began near zero for about the 10th and 11th coauthor of an article with 11 or more authors and ended with .222 for first and second coauthors of an article with two authors.

Ties illustrated in the diagrams were weighted in two ways. First, using the method described above, collaboration was made cumulative. If authors had coauthored more than once in the time period studied, the coauthor weights were added. For example, if the same two authors had written two articles together without additional authors, they were allocated a tie weight of .222+.222=.444 for their total collaboration. The ties were weighted without regard for direction; for example, the tie between a first and second author was seen as the same as that between the second and first. Therefore, the above two authors who had coauthored a two-author article twice got a tie weight of .444 whether or not they reversed their listed order in the two publications. In the network diagrams of the present article, the collaboration weights of ties are represented by the width of the lines. In a second weighting procedure, ties were weighted according to the date of the publication. Presumably, ties decayed over time if they were not renewed with additional publications, so ties were also weighted by the year of the most recent publication. In the network diagrams of the present article, this is represented by the shade of the lines representing ties. Black lines represented the most recent years of publication, whereas older ties were represented by lines of increasingly lighter shades of gray, fading with time.

Diagramming and Delimiting the Networks

NodeXL has good network visualization capability, but development of a visually clear diagram required hiding many coauthors. As described above, the network diagrams presented in this article were filtered by NodeXL to visualize the ties between authors with relatively high eigenvector centrality and degree. In the final visualization, authors who had any eigenvector centrality value smaller than .001 or degree less than 2 were excluded. Another visualization issue was the arrangement of the authors in the diagram. Initially the Fruchterman and Reingold (1991) algorithm, available in NodeXL, was used to force authors apart unless they were directly connected by ties, in which case they were drawn together. That procedure left some collaboration cliques obscured by close clustering. NodeXL allows the manual click-and-drop movement of vertices, and I used it to separate those cliques for clear visualization.

Based on existing literature, the data used here were divided into two overlapping 5-year time periods, 2001-2005 and 2004-2008. Longer time periods would include more connections and form a denser network than shorter periods. Barabasi et al. (2002) indicated that in traditional fields such as mathematics, a giant component connecting a majority of authors emerged only when several years were considered. At the extreme, Zaccala's (2004) 26-year giant component of a mathematics network was dense and visually opaque. In addition, Chen (2004) suggested the use of overlapping time periods to demonstrate turning points. Therefore, the 5-year time periods selected here were expected to be long enough to show any existing network, and the overlap was expected to reveal development of the network over time.

I selected the coauthors included in the later 2004-2008 time period first, beginning with the inclusion of all top-rated journal coauthor dyads in the analysis but narrowing them to authors who had eigenvector centrality values of .001 or greater. Then first-degree non-top rated coauthors identified via snowball sampling and qualified via eigenvector centrality of .001 extended the original list. A repeated snowball technique for second-degree extensions revealed no further additions, a fortunate outcome for visualization and resources. Thus, the 2004-2008 visualized author list was limited to a one-degree extension into non-top rated authoring and totaled 57 authors, of which 34 (60%) were originally included because they had published in top-rated journals. The 2004-2008 visualized author list was then searched for publications in the 2001-2005 time period and also extended one degree from that search to match the 2004-2008 procedure. For 2001-2005 this resulted in the inclusion of 20 authors who had been identified in the 2004-2008 period and another eight who did not appear in the 2004-2008 network. Seventeen (61%) of the resulting 28 authors in the 2001-2005 network had published in top-rated journals, similar to the 60% in the 2004-2008 network.

To provide tentative data on the effect of the network on citations, I gleaned and recorded the number of citations for each author in the network during the last year and the following year from Google Scholar. That is, I recorded the number of all of the citations for each author in the 2001-2005 network cited in the years 2005-2006, and for each author in the 2004-2008 network, the number citations in the years 2008-2009. I included references to any literature by the author, whether it was included in the development of the network or was a singly or multiply authored work. Because the citations follow a rich-get-richer pattern, the appropriate average is the median. The mode would have been biased toward the many small values of the distribution, whereas the mean would have been biased toward the few very high values of the distribution. Consequently, I located the median number of citations for authors for both time periods.


Table 1 provides descriptive statistics of the complete list of authors who published in top-rated journals, as well as those included in the network diagrams. As the top half of Table 1 indicates, 1,196 authors published in the four top-rated journals in the 2001-2005 period, and 1,234 published in these journals in the 2004-2008 period. Of the total top-rated journal authors, 465 (39%) published individually in the first period and 415 (34%) in the second period. Numerous top-rated journal authors wrote with one collaborator and thus one degree; therefore, many such authors failed to meet the selection criteria for diagramming. As coauthoring increased between 2001-2005 and 2004-2008, the numbers with few degrees declined, and the average number of collaboration ties per top-rated journal author increased from 1.53 in 2001-2005 to 1.90 in 2004-2008. The proportion with eigenvector centrality and degree sufficient to include them in the network diagrams was quite small--17, or about 1.4%, of the total in the first period--but had grown to 33, or about 2.7%, in the second period. The diagrammed top-rated journal and non-top rated authors are described in the lower half of Table 1.

The diagrammed authors also showed increases in coauthoring in all areas except the 10- to 14-degree range. Therefore, both the top-rated journal authors as well as those diagrammed increased their coauthoring across the periods studied. It should be noted, however, that diagrammed coauthors were a small fraction and were not representative of the larger population. Finally, the median number of citations for the diagrammed authors for the two time periods is presented in the last line of Table 1. The median for 2001-2005 was 1, and the median for 2004-2008 was 5.5, indicating that the typical diagrammed author in the latter period was much more likely to be cited than one in the first period. However, it should be noted that no attempt was made to link the citations to the articles on which the network was based, nor were alternative explanatory variables examined.

Figures 1 and 2 compare the discovered skeletal networks in the two time periods. Figure I contains 28 authors and seven cliques that were found in the 2001-2005 period. The largest clique, with 11 people, consisted of Green and his mostly student coauthors, found in the lower right of the diagram. Proceeding clockwise around Green, two small and fading cliques most notable for the presence of Baskind are diagrammed. Another small clique including Hamlin was revealed, and at the top of the diagram a heavily used clique including Walsh was made visible. Corcoran had coauthored with Walsh but was not otherwise connected by the selection criteria, and two small cliques were made evident to the right of Walsh. Green and Walsh, who were connected by several direct and indirect ties and were both members of multiple cliques, held the Figure 1 network together. Despite a tie to Ghaemi at Harvard, Figure 1 demonstrates a highly local network centered primarily at Virginia Commonwealth University (VCU) and secondarily in greater Virginia. Many of the ties in Figure 1 were based on one coauthoring experience.


Figure 2 depicts the 57-author skeletal network discovered for 2004-2008. Professor Green with his numerous connections remained present, largely by virtue of the overlapping time period, in the lower left of the diagram. Many of Green's ties were from 2004-2005 and had faded, but Green and Baskind had renewed and strengthened their tie. Kiernan-Stern had developed current and weighty ties from book coauthoring. Walsh had become less central than he was in the earlier period. Furthermore, as social network analysts say, the network had become more robust because Baskind's connections offered alternative routes to many of the cliques, and the network was less dependent on Green. The network was still based in Virginia and at VCU, but Baskind's ties had spread the network wider, primarily into New York State and St. Louis University. Briar-Lawson and Rizzo became central in the non-Virginia part of the network. Although Briar-Lawson's ties were of similar strength to Baskind's, they were more current. Most of Rizzo's ties were less current but were well used, and Rizzo's ties to Mizrahi were both current and well used.


In brief, Figure 2 was twice the size of Figure 1. The growth was largely a result of Baskind's emerging connections, which were less local, more cosmopolitan, and more--to reiterate social network language--small world.


This exploratory study is hardly conclusive, but it shows that at least one limited and locally based community emerged among top-rated social work journal authors, and that network was growing and becoming more cosmopolitan. Although much of that community was hidden to provide visualization in the diagrams, even the visualized and most connected pieces probably have not reached a level that would produce the best work because connections are less numerous than in other fields, and the most cited authors are associated with the most connected and most recent network. Some of the tentative findings on the limited existence of community may result from shortcomings of the study. First, an assumption that all coauthoring is comparable might have introduced measurement error. In addition, beginning with top-rated social work journals and expanding the network conservatively has surely led to limited discovery of community when the full network is certainly more extensive. For example, some initial exploration suggested that substantial community in specialized practice such as child welfare extended strongly into allied professions. Finally, the citations recorded are a crude representation of network effects, with no attempt to relate the number of citations to the specific articles or the status of the authors. These limitations, as well as the intentional limitation of the selection of authors, means that Figures I and 2 are not representative of the larger network, nor are they conclusive. Nonetheless, statistics on all authors of top-rated journal articles appear to suggest limited but growing collaboration and networking in social work scholarship; the diagrams and their statistics suggest that this growth is somehow associated with greater numbers of citations.

Setting the limitations aside and assuming a nascent community among top-rated social work journal authors that leads to increasing citations, the question of continued growth is suggested. Although the complex systems approaches usually applied to networks would suggest unpredictability, it seems likely that growth will continue. Increasing numbers of PhD-producing social work programs will probably push more authors to publish, and increasingly complicated research methods will encourage specialization and cooperation between scholars. Also, cooperation is perhaps more consistent with social work values than alternative competitive approaches. Assuming the number of articles in the top-rated journals remains about constant and present growth rates of coauthoring continue, increased coauthoring to about 2.4 authors per article would result in an increase in the total number of published top-rated journal authors to about 1,275 in the 2007-2011 five-year period. A review of the four top-rated journals published in 2009, the midpoint of the 2007-2011 period, suggested that top-rated social work publishing has reached an average that places it in the middle of the range that Newman (2001c) found in other fields. Because initial growth usually takes the form of an accelerating "S" curve (Crane, 1972), faster future growth is possible. If connected cliques such as the ones described here are emerging elsewhere, then bridges between cliques may be producing small worlds, and a giant component may be emerging in a rapid "phase transition" (Guimera et al., 2005). Such bridges are exemplified by Baskind's increasing connections between 2001 and 2008.

Should coauthoring be encouraged? Citations might be considered a measure of the quality of scholarship. The large difference between the number of citations associated with Figures 1 and 2 suggests fertile ground for associating increased networking with publication quality. In addition, the finding that the higher level of citations is associated with the higher level of networking suggests that social work coauthoring is far from the point of being too networked for the best quality. If that tentative finding holds up in further research, and the results of Uzzi and Spiro (2005) apply to social work publishing, social work scholars might be encouraged to do more coauthoring because it could lead to community that produces better work.

However, that would suggest another question related to the present research, namely, how schools of social work should evaluate coauthored work. If schools need to adjust the value of coauthoring for faculty evaluation, perhaps they should do it with weighting similar to that used in this article, or more precisely by writing papers with wikis that keep track of contributions, as suggested by Dobkin (2009). Then coauthors could be expected to have more publications but might do so with help. In this way, schools of social work and the invisible college might both improve their quality. In addition, the connections to young faculty members, or to faculty in programs that do not routinely produce prestigious publications, would encourage the integration of valued knowledge into the larger social work education community. Significant value of authorship beyond the first author in decisions of promotion and tenure might encourage such connections and thus the distribution of knowledge for teaching and practice. In addition, the well-published senior researchers might mentor distant or local proteges through such connections, and the coauthor network could expand its influence. However, based on Uzzi and Spiro's (2005) findings, connectivity could go too far and reduce quality, so in the more distant future caution may be needed. Questions for future research therefore involve the appropriate level of connectivity for the production of quality research.

Of course, some aspects of the quality of scholarship will not improve just because there is an appropriately connected scholars' community. I have addressed what Watts (2003) called the dynamics of the network, that is, the creation and dissolution of coauthor ties. Also important is what Watts called dynamics on the network, or the content of the communication between the coauthors. If coauthoring is to increase the quality of scholarship, it requires in-depth communication on important topics, so a growing social work scholars' community must ask, community for what? For example, one might argue that the scholarship of teaching often addressed in JSWE needs to give way to the scholarship of practice to provide appropriate content for teaching and practice. Although academic social workers' study of teaching is a way to improve our research and teaching effectiveness, perhaps our top-rated scholars' community should communicate more heavily about practice and related research as it appears in SWR and SWK. Additionally, whether the scholarship of teaching is valued might depend on the nature of the school's or department's emphasis on teaching. The nature of good pedagogy also may differ relative to the content. It might be appropriate for faculty of social work education programs emphasizing teaching to publish on the best teaching practices for different kinds of knowledge--for example, teaching human behavior as opposed to teaching practice. Although these may be value issues as well as research issues, the establishment of a scholarly community prompts these questions.

In summary, the initial discoveries reported in this article suggest a limited but growing collaboration network that is associated with better-quality scholarship. That growth suggests that social work may be poised for rapid development of a large and integrated network of scholars. An awareness of that network, the promotion of it, the purposeful use of it, and attention to the content of the information it transmits might all be of benefit to social work.


Baker, D. R. (1992). A structural analysis of the social work journal network: 1985-1986. Journal of Social Service Research, 15(3-4), 153-168.

Barabasi, A. L., Jeong, H., Neda, Z., Ravasz, E., Schubert, A., & Vicsek, T. (2002). Evolution of the social network of scientific collaborations. Physica A: Statistical Mechanics and Its Applications, 311, 590-614. Retrieved from

Berg-Weger, M., &. Schneider, F. D. (1998). Interdisciplinary collaboration in social work education. Journal of Social Work Education, 34, 97-107.

Borgatti, S. P. (1995). Centrality and AIDS. Connections, 18(1), 111-113.

Boyer, E. L. (2004). Enlarging the perspective. Physical Therapy, 84, 571-575.

Castells, M. (2000). Toward a sociology of the network society. Contemporary Sociology, 29, 693-699.

Chen, C. (2004). Searching for intellectual turning points: Progressive knowledge domain visualization. Proceedings of the National Academy of Sciences, 101, 5303-5310.

Cnaan, R. A., Caputo, R. K., & Shmuely, Y. (1994). Senior faculty perceptions of social work journals. Journal of Social Work Education, 30, 185-199.

Council on Social Work Education. (2008). Educational policy and accreditation standards. Retrieved from

Crane, D. (1972). Invisible colleges. Chicago, IL: University of Chicago Press.

Dobkin, B. H. (2009). Collaborative models for translational neuroscience and rehabilitation research. Neurorehabilitation and Neural Repair, 23, 633-640.

Freeman, L. C. (2004). The development of social network analysis. North Charleston, SC: Booksurge.

Fruchterman, T. M. J., & Reingold, E. M. (1991). Graph drawing by force directed placement. Software--Practice and Experience, 21, 1129-2264.

Gelman, S. R., & Gibelman, M. (1999). A quest of citations? An analysis of and commentary on the trend toward multiple authorship. Journal of Social Work Education, 35, 203-212.

Green, R. G. (2008). Tenure and promotion decisions: The relative importance of teaching, scholarship and service. Journal of Social Work Education, 44, 117-127.

Green, R. G., & Baskind, F. R. (2007). The second decade of the faculty publication project: Journal article publications and the importance of faculty scholarship. Journal of Social Work Education, 43, 281-295.

Guimera, R., Uzzi, B., Spiro, J., & Nunes Amaral, L. A. (2005). Team assembly mechanisms determine collaboration network structure and team performance. Science, 308, 697-702.

Hodge, D. R., & Lacasse, J. R. (2011). Ranking disciplinary journals with the Google Scholar H-Index: A new tool for constructing cases for tenure, promotion, and other professional decisions. Journal of Social Work Education, 47, 579-596.

Moody, J. R. (2004). The structure of a social science collaboration network: Disciplinary cohesion from 1963-1999. American Sociological Review, 69, 231-236.

Newman, M. E. (2001a). Scientific collaboration networks: I. Network construction and fundamental results. Physical Review E, 64, 016131.1-016131.8.

Newman, M. E. (2001b). Scientific collaboration networks: II. Shortest paths, weighted networks, and centrality. Physical Review E, 64, 016132.1-016132.7.

Newman, M. E. (2001c). The structure of scientific collaboration networks. Proceedings of the National Academy of Sciences, 98, 404-409.

Newman, M. E. (2002). Assortive mixing in networks. Physical Review Letters, 89, 20870.1-20890.0.

Newman, M. E., & Park, J. (2003). Why social networks are different from other types of networks. Physical Review E, 68, 036122.1-036122.13.

Seipel, M. O. (2003). Assessing publication for tenure. Journal of Social Work Education, 39, 79-88.

Sellers, S. L., Mathiesen, S. G., Perry, R., & Smith, T. (2004). Evaluation of social work journal quality: Citation versus reputation approaches. Journal of Social Work Education, 40, 143-160.

Smith, M. A., Shneiderman, B., Milic-Frayling, N., Rodrigues, E. M., Barash, V., Dunne, C., ... Gleave, E. (2009). Analyzing (social media) networks with NodeXL. Procedings of the Fourth International Conference on Communities and Technologies. New York, NY: Springer.

Urry, J. (2003). Social networks, travel and talk. British Journal of Sociology, 54, 155-175.

Uzzi, B., & Spiro, J. (2005). Collaboration and creativity: The small world problem. American Journal of Sociology, 111, 447-504.

Videka, L., Blackburn, R. A., & Moran, J. R. (2008). Building research infrastructure in schools of social work: A university perspective. Social Work Research, 32, 294-301.

Watts, D. (2003). Six degrees. New York, NY: Norton.

Williams, J. B. W., Tripodi, T., Rubin, R., Hooyman, N., Allen-Meares, P., Padgett, D. K., & Fortune, A. E. (2008). A historical account of the Society for Social Work and Research: Presidential perspectives on advances in research infrastructure. Social Work Research, 32, 208-219.

Wineman, J. D., Kabo, F. W., & Davis, G. R (2009). Spatial and social networks in organizational innovation. Environment and Behavior, 41, 427-442.

Wuchty, S., Jones, B. F., & Uzzi, B. (2007). The increasing dominance of teams in production of knowledge. Science, 316, 1036-1038.

Zaccala, A. A. (2004). Revisiting the invisible college: A case study of the intellectual structure and social process of singularity theory research in mathematics. (Unpublished doctoral dissertation). University of Toronto, Canada.

Accepted: 05/11

Ralph Woehle is professor emeritus at the University of North Dakota.

Address correspondence to Ralph Woehle, 40518 447th St. SE, Lengby, MN 56651; e-mail:

Ralph Woehle

University of North Dakota

DOI: 10.5175/JSWE.2012.201000088
Table 1. Characteristics of Authors in Relation to Networks

                                     2001-2005   2004-2005

Top-rated journal article authors
  Total published                    1,196       1,234
  Undiagrammed 0 degree                465         415
  Undiagrammed 1 degree                263         241
  Average degree                         1.53        1.90
  Diagrammed 2-26 degree                17          33
Diagrammed authors
  Degree 2-4                            15          33
  Degree 5-9                             1          12
  Degree 10-14                          11           9
  Degree 15-26                          11           3
  Median citations                       1           5.5
COPYRIGHT 2012 Council On Social Work Education
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Woehle, Ralph
Publication:Journal of Social Work Education
Article Type:Report
Geographic Code:1USA
Date:Sep 22, 2012
Previous Article:Teaching writing as a professional practice skill: a curricular case example.
Next Article:Assessing BSW student direct practice skill using standardized clients and self-efficacy theory.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |