A social network analysis of the Journal of Supply Chain Management: knowledge generation, knowledge diffusion and thought leadership.

INTRODUCTION

The field of purchasing has been viewed as an independent function for well over 100 years (Fearon 1968) and as a distinct profession since 1915--the year in which the National Association of Purchasing Agents was founded (Leenders, Fearon and England 1989). Further, the field of purchasing has been viewed as a scholarly discipline since at least 1965 when the first article appeared in the Journal of Purchasing, a peer-reviewed scholarly journal that is now the Journal of Supply Chain Management. During this 40-year period, authors have periodically examined the evolution of the field by reviewing the subject categories, research design and methodologies of articles which have appeared in the Journal (Williams and Oumlil 1987; Carter and Ellram 2003). These articles have helped to summarize the state of the field through an examination of the research which had been published in the Journal, and in so doing have also provided guidance for future research in the field. While the work of Williams and Oumlil and Carter and Ellram investigated the content of published articles in the Journal, these papers did not explicitly examine the patterning of the citations and the changing communication patterns of the citations from those articles appearing in the Journal. The objective of this research is to fill that gap.

The broad goal of citation analysis is to better understand communication patterns within a scientific discipline (Yaru 1997). More specifically, citation analysis allows a researcher to examine citations made to and from journals which can explain and clarify flows of communication across scholarly disciplines and among the academic institutions which generate this knowledge (Phillips and Phillips 1998). This technique is often applied to the leading scholarly journal within a scientific discipline (e.g., Phillips, Baumgartner and Pieters 1999). Unlike raw frequency counts of published articles, a citation analysis can more accurately capture the value of these publications, with the idea that articles which make a greater contribution to a scientific field tend to be cited more often (Phillips and Phillips 1998).

A potential, complementary technique, called social network analysis (SNA), can be used to further understand these flows of communication and exchanges of information among actors (Everett and Pecotich 1991; Autry and Griffis 2005), where actors can be defined as individuals within an organization, organizations within a supply chain or in the case of a citation analysis, academic institutions within a scholarly discipline. In combination, this methodology can be used to model every connection between all of the institutions in an academic knowledge network, by examining the sending and receiving patterns of citations within the network (Phillips and Phillips 1998). SNA can also be used to identify what Burt (1982) refers to as the "invisible college" in a scientific community--a tightly knit sub-network of members at the center of knowledge generation within an area or field of research.

Despite the potential value of SNA for better understanding how knowledge is generated and diffused, and how supply management as a scholarly discipline has evolved, this methodology has not yet been used in the research appearing in the Journal. This study addresses this gap by examining the following research objectives:

Research Objective 1: To track the number of citations appearing in individual articles over time.

The authors' expectation is that as the field of supply management and the Journal have matured over time, research appearing in the Journal would be more theoretically rich and methodologically sophisticated, resulting in an increased number of citations per article.

Research Objective 2: To track the percent of cited articles from different fields over time.

The purpose of assessing the percent of cited articles from different fields over time is to better understand the field(s) from which the supply management body of knowledge has relied upon to develop models and integrate theories used to explain managerial behavior and supply management phenomena.

Research Objective 3: To track the percent of cited articles from the Journal over time.

The authors' expectation here is that over time articles appearing in the Journal would begin to increasingly rely upon the accumulated knowledge base of research appearing in prior issues of the Journal.

Research Objective 4: To identify those universities which are "thought leaders" in the Journal, by (a) determining not only those schools that have contributed the greatest number of articles, but (b) those universities that are most central within the Journal's citation network and (c) more broadly introducing the SNA methodology to supply management researchers, with the goal of guiding potential future applications of SNA to supply management research.

These research objectives will be accomplished through a SNA of the citations within the 40-year history of the Journal. It is important to not only identify the universities whose programs have contributed to the development of supply management knowledge, but to understand the actual influence that the research of these respective universities has had on the field. Just as journals which receive a higher proportion of citations are considered to be more influential than journals which receive a lower proportion of citations (Baumgartner and Pieters 2003), it can also be assumed that universities that are more central within a citation network of universities generating supply management research and knowledge are more influential. The behavioral research in the field of supply management has seen an over-reliance on survey and case study methodologies (Carter and Ellram 2003; Parente and Gattiker 2004). SNA can help managers to effectively map informal communication and workflow networks (Carter, Ellram and Tate 2007) and can allow organizations to better manage knowledge, information and organizational learning (e.g., Hult, Hurley, Guinipero and Nichols 2000; Hult, Ketchen and Nichols 2003). Yet despite the potential advantage of and need for complimentary research methodologies in supply management research, the use of SNA has been almost nonexistent in our field.

In the next section of the paper the authors describe the study's methodology. The results of the study's analyses based on Research Objectives 1-4b are then presented and discussed. In the paper's final sections, the authors further highlight potential applications of SNA to the field of supply management (Research Objective 4c), and provide additional thought regarding future directions of the field, based on the study's findings.

METHODOLOGY

The authors began by collecting a copy of every article published in the Journal over the 40-year period from the first issue in 1965 through the last issue in 2004. Afterwards, the article title, volume and issue numbers, authors and author affiliations for each peer-reviewed article were entered into a Microsoft Access[R] database. A total of 882 articles were entered into the database in this fashion. As the Journal has at times published non-peer-reviewed subject matter such as book reviews and executive interviews, these non-peer-reviewed articles were not included in the database. In addition, to avoid double counting, the citations contained in the article reprints from anniversary issues of the Journal were also not included in the database.

Finally, the journal titles of every citation from each of these 882 articles were entered into the database, which resulted in a total of 11,408 citations. In the event that a citation came from the Journal (e.g., an article in the Journal cited another article from the Journal), this citation was cross-referenced with the university affiliation(s) of the author(s) of the cited article. The first and second authors entered this data, and separately checked the accuracy of the data entry of the other author. This database was then used as the source for the analyses that are described next.

RESULTS AND DISCUSSION

Research Objective 1: Citations Per Article

Consistent with the methodology of Carter and Ellram (2003), the 156 issues of the Journal from the 40-year time period were divided into eight equal 5-year periods (i.e., 1965-69, 1970-74, etc.). An analysis of variance (ANOVA) was then performed using SAS, Version 9.1, to determine whether changes in the number of citations per article may have occurred over time. The independent variable, time period, ranged from one (time period 1: 1965-69) through eight (time period 8: 2000-04). The dependent variable, number of citations per article, was operationalized as the average number of citations per article per issue of the Journal. To avoid problems associated with autocorrelation, these averages were then randomly sorted within each time period. The results of the ANOVA are displayed in Table I, and suggest a statistically significant overall difference across time periods (F=66.07, p < 0.0001).

Next, Duncan's multiple range post hoc comparisons (see Bruning and Kintz 1977) were used to ascertain in which time period(s) these statistically significant differences occurred ([alpha]=0.05). As is also shown in Table I, period 8 (2000-04) has a significantly greater number of citations per article than periods 1 through 7 (1965-99) (p < 0.05), period 7 (1995-99) has a significantly greater number of citations than periods 1 through 6 (1965-94) (p < 0.05), and period 6 (1990-94) has a significantly greater number of citations than periods 1 through 5 (1965-89). There are no statistically significant differences in the number of citations across periods 1 through 5.

These results suggest that the average number of citations per article has increased over time, and provide some support for the assertion that research in the field of supply management and in particular research which has appeared in the Journal, has become more theoretically rich and methodologically complex over the 40-year period. This assertion is also supported by the work of Carter and Ellram (2003, p. 32), who find a significantly greater prevalence of hypothesis testing (as opposed to normative literature and exploratory studies) and "more advanced and inferential data analysis techniques" in the 1990s as opposed to earlier time periods in the Journal's history. As researchers have increasingly used deductive, theoretically derived models, grounded theory building that integrates existing theories, and more advanced statistical methodologies, the field has seen an increasing number of citations referencing this extant literature.

Research Objective 2: Cited Articles from Different Fields

To examine the second research objective, the authors assigned each citation to one of the 22 citation codes appearing in Table II. The first and second authors together coded the citations of each article to ensure reliability. This coding was accomplished by creating a list of journals and/or a description of the type of citation appearing under each of the coding categories. In the case of journals for which the authors were unfamiliar, the homepages of these journals and a description of the journals based on ProQuest[R] and similar scholarly databases were examined.

Column 2 of Table II displays the percent of citations attributable to each of the citation categories over the entire 40-year period. Books were by far the most common type of citation with over 27 percent of the total citations, followed by scholarly supply management journals (12.88 percent) such as the Journal, scholarly management journals (11.70 percent) such as the Academy of Management Journal, and trade publications (10.32 percent) such as Purchasing. Table II also displays the percent of citations from the 22 categories for each of the eight, 5-year time periods in columns three through 10. The researchers next used this data to examine any trends in citation patterns over the 40-year time period.

First, the authors investigated whether changes might have occurred over time in the proportion of citations from scholarly journals by performing an ANOVA in which the eight, 5-year time periods again represented the independent variable, and the average proportion of citations from scholarly journals per article per issue represented the dependent variable. As was the case with the first ANOVA, these averages were randomly sorted within each time period to avoid problems with autocorrelation. The researchers defined scholarly journals as those journals that are rigorously (generally peer, double-blind) reviewed. This included the following citation categories: accounting, economics, finance, legal-scholarly, logistics and transportation, management, marketing, operations management/operations research (OM/OR), psychology, purchasing, sociology and other-scholarly. The results of the ANOVA are shown in Table III.

The mean proportion of citations from scholarly journals over the 40-year time period is 38.42 percent. The results from the ANOVA displayed in Table III indicate that this proportion varies significantly over the eight, 5-year time periods (F=27.47, p < 0.0001). Duncan's multiple range test was used to examine the specific differences across time periods ([alpha]=.05). Period 8 (2000-04) has a significantly greater proportion of citations from scholarly journals (59.93 percent) than periods 1 through 7; both periods 7 (1995-99) at 50.90 percent and 6 (1990-94) at 48.69 percent have a significantly greater proportion of articles from scholarly journals than periods 1 through 5; period 4 (38.20 percent) has a significantly greater proportion of scholarly citations than periods 1 through 3; and periods 3 (29.36 percent) and 2 (27.88 percent) have a significantly greater percentage of citations from scholarly journals than period 1 (15.37 percent).

These findings support the earlier assertion that research published in the Journal is becoming more theoretically rich over time, and that the analyses used to conduct this research are becoming increasingly sophisticated. The results displayed in Table III also substantiate the work of Carter and Ellram (2003, p. 39) whose research indicates a "maturing of supply management as a scholarly discipline," as research which appears in the Journal is progressively more deductive in nature, and is using more advanced methodologies to test and develop theory.

In order to better understand the fields from which these scholarly citations originated, the authors next performed a multivariate analysis of variance (MANOVA). Here, the independent variable was the 5-year time period and the dependent variables were the proportion of citations for each of the twelve categories of scholarly journals listed in Table II (accounting, economics, finance, legal-scholarly, logistics and transportation, management, marketing, OM/OR, psychology, purchasing, sociology and other-scholarly). The results of the MANOVA are presented in Table IV.

The Wilk's Lambda, Pillai's Trace, Hotelling-Lawley Trace and Roy's Greatest Root test statistics were all significant at p < 0.0001, leading to a rejection of the null hypothesis that there is no overall difference in the percentage of citations for each discipline across the eight time periods. An examination of the individual ANOVAs for the scholarly disciplines, adjusted for [alpha] inflation (Flynn, Sakakibara, Schroeder, Bates and Flynn 1990), indicated significant differences across time periods in the case of logistics and transportation (p<0.0001), management (p < 0.0001), marketing (p < 0.01), OM/OR (p < 0.05) and supply management (p < 0.001) citations.

Duncan's range test was then used to investigate specific differences in the proportion of citations for each of these scholarly disciplines across time periods ([alpha]=0.05). These results, which are also displayed in Table IV, suggest a significant increase in the proportion of citations in the last five to 10 years from the fields of logistics and transportation, management and marketing. In addition, the proportion of citations from the OM/OR arena are significantly greater in periods 3 (1975-79) and periods 5-8 (1985-2004) than periods 1 and 2 (1965-74) and period 4 (1980-84). Finally, the proportion of citations from the field of Purchasing and Supply Management increased between period 1 (1965-69) and periods 2-5 (1970-89), increased again in period 6 (1990-94) and decreased in the final two periods (1995-2004).

It is not surprising to find an increase in the proportion of scholarly supply management citations after period 1, as most of these citations would have come from the Journal, and a period of five or even 10 years may be needed for a journal to establish a "critical mass" of knowledge and findings before researchers begin to cite these publications (Phillips et al. 1999). The peak in the proportion of supply management citations in period 6, followed by a decline in periods 7 and 8, can be explained by the corresponding increases in the proportion of citations from the fields of logistics and transportation, management, marketing and OM/OR. These trends correspond to the increased emphasis on a broader view of supply chain management (Mentzer, DeWitt, Keebler, Min, Nix, Smith and Zacharia 2001), and a shift from purchasing to a more holistic view of supply management (Burt, Dobler and Starling 2003; Leenders, Johnson, Flynn and Fearon 2005) and the role of supply management within the broader supply chain (Carter 1999).

Research Objective 3: Cited Articles from the Journal

An ANOVA was next performed to examine whether the proportion of citations from within the Journal changed significantly over time. The dependent variable consisted of the average proportion of citations per issue from 1965-2004. As with the prior analyses, the independent variable was operationalized by the 5-year time periods. However unlike the prior analyses, the first 5-year time period (1965-69) was omitted from the analysis in order to allow a sufficient period of time for the generation of an adequate knowledge base within the Journal (Phillips et al. 1999). The results of the ANOVA, displayed in Table V, suggest that there are significant, overall differences in the proportion of citations from the Journal across the time periods (p=0.0199). The results from the multiple comparison testing, displayed at the bottom of the table, show that there were a significantly greater proportion of citations from within the Journal during period 6 (18.66 percent) as compared with all of the periods, and that there were a significantly greater proportion of citations from within the Journal during period 4 (16.02 percent) as compared with period 5 (10.62 percent).

These analyses show that the percent of citations from within the Journal peaked during the 1990-94 time period, and thereafter decreased significantly. This finding is in line with the earlier analyses, which showed that the percent of citations from supply management scholarly journals peaked during the same sixth, 1990-94 time period and decreased significantly thereafter. At the same time, citations from management, marketing and logistics and transportation journals increased significantly after time period 6. Together, these findings suggest that the supply management field is beginning to draw from broader bases of literature, much as the field of supply chain management has evolved into a more encompassing, multi-disciplinary perspective. Similarly, the Journal has broadened its focus beyond purchasing to supply chain management (Carter 1999). While the emphasis of the Journal is still on the supply management side, articles which highlight broader supply chain management content have begun to appear (e.g., Ellram, Tate and Billington 2004).

Research Objective 4a: University Contributions to the Journal

The examination of those universities that have contributed the most to the thought leadership of the Journal began with a simple tabulation of the institutional affiliation of authors whose work has been published in the Journal. Consistent with Carter, Vellenga, Gentry and Allen (2005) and Vellenga, Allen and Riley (1981), in the case of an article with more than one author, the academic institution of each author was credited with an article. A total of 443 institutional affiliations were represented over the 40-year time period. Table VI displays the top-25 most prolific universities, in terms of the absolute number of publications that have appeared in the Journal between 1965 and 2004. Arizona State University (ASU) and Michigan State University (MSU) are by far the most prolific universities with 119 and 104 publications, respectively; afterwards, the number of publications for those universities ranked third through 23rd drop sharply. Both ASU and MSU have had strong supply management programs at the undergraduate, masters and PhD levels for many years, and thus these results are not particularly surprising. Many of the other universities appearing in Table VI have strong logistics and supply chain management curricula at the undergraduate through doctoral levels (e.g., The Ohio State University) or strong undergraduate supply management courses of study (e.g., Bowling Green State University).

While the findings displayed in Table VI offer a tangible measure of output from these universities, the data do not afford a complete picture of the impact that this research has had within the field of supply management. To provide further insights about the citation network and the diffusion of knowledge and thought leadership, the authors next performed an SNA.

Research Objective 4c: A Further Examination of University Thought Leadership

SNA is a powerful methodology for describing and analyzing the interrelationships among a group of actors within a network. These actors can be people within organizations, or the organizations themselves, and the relationships can include, for example, liking, trust and the exchange of information and goods among actors (Scott 2000). SNA therefore manifests the linkages among nodes, called actors, within a network.

Most of the existing interorganizational SNA research is found in the field of strategic management, where researchers have investigated the interrelationships among organizations in a range of social networks, including supplier relationships (Dyer 1996), interlocking directorates (e.g., Pettigrew 1992) and horizontal alliances (Nohria and Garcia-Pont 1991). The social network in which an organization is embedded can be thought of as a set of nonsubstitutable and inimitable resources (e.g., Wernerfelt 1984; Barney 1991). Gulati (1999) refers to these network ties as "network resources" and Gulati, Nohria and Zaheer (2000) propose that the capacity of a firm to use key resources through its network of strategic alliances can permit the organization to strengthen its competitive advantage. Finally, Hite and Hesterly (2001) provide a treatment of the means by which networks and organizations within networks evolve over time.

In the case of this study's application of SNA to a citation analysis, the actors are the academic institutions represented by the articles published in the Journal, and the relationships consist of citations by authors with various university affiliations of other authors at universities which have appeared in the Journal. For example, if an author from University A published an article in the Journal, and cited an article which had appeared in the Journal that was written by an author at University B, then the two nodes in this network would be University A and University B, and the link would be a unidirectional arrow from University A to University B. Similarly, if an author from University C cited the work of an author from University B which had appeared in the Journal, then another unidimensional link would be formed between the nodes of University B and University C. These links and nodes, along with links and nodes from two additional example universities, are displayed in Figure 1. An adjacency matrix of the example sociogram is presented in Table VII, and displays the number of citations from and to each institution in the five-university illustration. This table shows that University A cited University B 15 times and University D four times, but did not cite any journal publications written by authors from Universities C or E; University B cited University A 16 times but did not cite Universities C, D or E; etc.

To examine the influence that universities have had within the citation network, the researchers measured network centrality in accordance with Leavit's (1951) conceptualization as the magnitude of participation of actors in a network, and Freeman's (1979) "degree" and "betweenness" dimensions of centrality. These dimensions of centrality have been widely accepted and adopted within the SNA research (Scott 2000). The authors use the illustration in Figure 1 and the corresponding adjacency matrix in Table VII to provide an example of how these dimensions of centrality are calculated.

The term "degree" is defined as the number of ties that an actor has to other actors in a network. An actor's degree can be calculated by adding the number of links between that actor and the other actors in a sociogram, or by adding the values within a row or column of an adjacency matrix. For the five-university example, University A would have a degree (referred to in SNA parlance as "in-degree") of 23 (16 citations from University B, four from University C and three from University D) while University E would have a degree of 0, as no other university within the example network cited University E. The degree values of the remaining three universities in the example network are shown in column 2 of Table VIII.

In a social network, "betweenness" is defined as the number of paths that pass through an actor on the shortest paths connecting two other actors. From the standpoint of diffusion of knowledge, an actor (university) with high betweenness centrality has greater influence in that it can serve as a de facto liaison between isolated areas of the network (Ronchetto et al. 1989). Freeman's (1979) measure of betweenness binarizes the values in an adjacency matrix (e.g., values are equal to 0 or 1) in order to calculate the number of paths that pass through an actor. In the example network, University A is between Universities B and D, C and D and E and D (via University B), and has a betweenness score of three as shown in column 3 of Table VIII; University B is between Universities E and A, and E and D (via University A), and has a betweenness score of two; Universities C, D and E are not between any pairs of the other universities, and have betweenness scores of zero.

SNA of the Journal of Supply Chain Management

The authors began their SNA by examining the Journal's citation network for the full, 40-year period of 1965-2004. The adjacency matrix of the 443 institutional affiliations in this citation network was created from the Access database, and entered into UCINET 6--an SNA software package (Borgatti, Everett and Freeman 2002)--to calculate the measures of degree and betweenness for each institution in the matrix. In accordance with existing SNA research (e.g., Brass 1984; Borgatti et al. 2002), the degree and betweenness measures were normalized. Such normalization is particularly valuable as it allows for valid comparisons across social networks of different sizes (Freeman 1979). The researchers found a very strong, positive correlation between normalized degree and normalized betweenness (r=0.9580, p < 0.0001). The results of an exploratory factor analysis (EFA) indicated that these two measures loaded highly on a single factor that explained 97.9 percent of the variance. Thus, following the lead of earlier researchers who have used these measures, the authors calculated the sum of the measures of the normalized degree and normalized betweenness values to create a single, aggregate measure of centrality (Carter et al. 2007; Ronchetto et al. 1989). These measures of centrality ranged from 0 to 0.1441.

The top 25 universities and their aggregate centrality scores for the 1965-2004 time period are displayed in columns one and two of Table IX. While many of these universities are also represented in the top 25 universities by publication (Table VI), there are five universities which are displayed in Table IX but not found in Table VI: Louisiana State University, Oklahoma State University, University of Texas, University of Illinois and University of Nevada, Las Vegas. A more in-depth analysis revealed that these five universities had not only high degree (the extent to which a university is directly cited by other universities) but also high betweenness, meaning that these universities lay between the citation paths of other universities within the network. In addition, approximately half of the universities displayed in Table VI are not present in Table IX; while the University of Wisconsin was ranked 12th in terms of the number of publications appearing in the Journal over the 40-year time period, its centrality score was ranked 62nd. These findings suggest that institutional thought leadership is derived not simply through a large number of publications, but rather by dissemination of that research by being centrally positioned within a citation network.

Figure 2 presents a sociogram of those top 25 universities displayed in Table IX. While universities like ASU, MSU, Western Michigan University and Bowling Green State University are "in the thick of things" with many arrows leading into these universities, other schools such as the University of Oklahoma appear to be more on the fringes of the network with only one arrow leading into that university. It is important to reiterate however that for purposes of parsimony, the sociogram only displays the paths among the top-25 universities (a sociogram with all 443 universities would have been almost uninterpretable if printed on an 8 1/2 by 11 in. page of paper). Thus most nodes appearing in Figure 2 have a significantly larger number of arrows leading from and into those universities than are shown in the figure. The University of Oklahoma, for example, is cited by six additional schools that are not shown in Figure 2, while ASU is cited by a total of 42 institutions. Thus these universities are not only embedded within the more concentrated, "top-25" network displayed in Figure 2, but also play key connective roles in the dissemination of knowledge across the full citation network consisting of all 443 cited and citing universities for the 40-year time period.

The authors next subjected the citation data to sub-period analyses, in order to examine longitudinal changes that might have occurred over the 40-year time period. Consistent with Phillips et al. (1999, p. 204), the first 10 years of the time period were not considered for inclusion in this longitudinal analysis, in order to avoid, "the irregularities that might occur during a journal's initial years of publication." Following Phillips et al. (1999) the authors also chose to examine multi-year time intervals to avoid the biasing effects of short-term fluctuations within a single issue or volume of the Journal. Five-year intervals were chosen to be consistent with the earlier analyses from this study, and the work of Carter and Ellram (2003). The authors further chose to examine three separate sub-periods to provide a reasonable but not overwhelming longitudinal contrast: period 4 (1980-84), period 6 (1990-94) and period 8 (2000-04). These periods were chosen rather than data from periods 3, 5 and 7 because: (1) the analyses presented in the earlier sections of this paper often showed significant differences between periods 7 and 8 versus earlier periods, and (2) period 6 had a higher proportion of citations from within the Journal as opposed to earlier and later periods, and the researchers wanted to include data from this time period in the longitudinal analyses.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

The same methodology as described above was used to calculate the centrality rankings for the three sub-periods. The results of these sub-period analyses are shown in columns three through five of Table IX. These results display some interesting changes in the structure of the Journal's citation network over time. First, while ASU and MSU have been highly central in the network across the three sub-periods, MSU appears to have become more central over time. Conversely, ASU has become somewhat less central, yielding its top-ranked position to MSU in the final sub-period. It will be interesting to see the changes that might occur in another 10 years, given the recent movement of supply management faculty away from ASU and MSU.

Second, while the top eight ranked schools from the full, 40-year time period also appear in the 1990-94 and 2000-04 time periods, 12 of the universities which are ranked in the top-25 for the 1965-2004 time period do not appear in the top-25 for the 2000-04 time period. Many of the new entrants to the top-25 for the 2000-04 time period are schools which have had traditional logistics curricula that have been broadened to supply chain management programs, including Auburn University, Iowa State University, University of British Columbia and University of Nevada, Las Vegas. In addition, two universities from outside of North America, Copenhagen Business School and University of Warwick, also entered the top-25 for the last 5-year time period. This latter finding is consistent with Carter et al.'s (2001, p. 92) finding that "more European universities are beginning to place an increased emphasis on publications in top-tier, peer reviewed journals." Finally, several of the top-25 schools for the 40-year time period which do not appear in the top-25 for the 2000-04 time period are universities that have consistently been included in the sub-period citation networks, but not necessarily in the top-25 for any individual time period. For example, Indiana University was ranked 71st, 56th and 56th, and the University of North Carolina was ranked 5th, 47th and 44th for the 1980-84, 1990-94 and 2000-04 sub-periods, respectively.

CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS

The authors found a significantly greater number of citations per article over time, particularly in the last 15 years of the 40-year time period under investigation. In addition, there were a significantly greater proportion of citations to scholarly journals during the 2000-04 time period as compared with earlier periods, and during the 1990-99 time period as compared with earlier periods. These results are in line with Carter and Ellram's (2003) findings of a maturing of supply management as a scholarly discipline and their call for more rigorous research in terms of theory and practice. Research that simply provides summary statistics may be more understandable by practitioners. However such research usually does not make significant contributions to a scholarly field of knowledge. Further, such inquiry often lacks richness and generally does not provide tangible take-aways for managers. Even research which uses inferential statistics without a priori theory may not provide significant contributions to theory or practice, as this research can capitalize on chance when such post hoc analyses are performed (Kerlinger 1986; Flynn et al. 1990). Like Carter and Ellram, the authors advocate the need for even greater theoretical richness and methodological rigor of research that is published in the Journal in the future.

While books (27.09 percent of total), scholarly management journals (11.70 percent of total), scholarly supply management journals (12.88 percent of total) and trade publications (10.32 percent of total) comprised the majority of the categories of citations found within the Journal over the 40-year period, there was significant change over this time period in the extant theory and findings that journal contributors relied upon in developing and interpreting their research. Specifically, work which has been published in last 5-10 years of the 40-year time period has cited existing research from the fields of logistics and transportation, management and marketing to a significantly greater extent than in earlier periods, and the proportion of citations from OM/OR journals are significantly greater from 1985-2004 than in most of the earlier time periods. Conversely, the proportion of citations from scholarly supply management journals, and in particular from the Journal, reached a peak during the 1990-94 time period, and then declined during the 1995-99 and 2000-04 time periods. As was suggested earlier, these findings are consistent with the increased emphasis on a more encompassing view of supply chain management (Mentzer et al. 2001), and a shift from purchasing to a more holistic paradigm of supply management (Burt et al. 2003; Leenders et al. 2005) along with an integration of supply management within the broader supply chain (Carter 1999). In addition, the 11.27 percent of citations from within the Journal during the 2000-04 time period is largely in line with the within-journal citations of top publication outlets of other fields. For example, a review of the 2004 issues of the Journal of Operations Management and the Journal of Marketing revealed average within-journal citations of 10.12 percent and 14.41 percent, respectively.

Our use of SNA to examine the citation network within the Journal suggests that while there is a relationship between the number of published articles and network centrality, this is not a one-to-one relationship. Fourteen of the 30 universities which are ranked in the top-25 for the aggregate number of articles across the 1965-2004 time period (Table VI) are not ranked among in the top-25 institutions based on citation network centrality for the same 40-year time period (Table IX). Further, six of the universities displayed in Table VI are not included in the top-25 ranked universities for any of the sub-periods shown in Table IX. Thus SNA provides a more in-depth examination of which universities provide thought leadership. Universities which are more centrally located within the Journal's citation network are more influential in terms of disseminating their knowledge to researchers at other schools. Future research might extend these current findings by examining the potential effect that a university's Journal ranking system has on submissions to the Journal and on citations to the Journal by supply management scholars. In addition, researchers might examine how the presence of a doctoral program at a university may affect network centrality.

The Journal is being repositioned to not only "be THE journal of choice for A-level manuscripts in the supply chain field," but also to gain, "A-level status ... from top-tier academic institutions globally" (Email Correspondence, Institute for Supply Management[TM] 2006). As such, future research appearing in the Journal will need to be even more deductive and theoretically based than in the past, or use appropriate inductive, grounded theory approaches to develop new supply management theory. All of this means that the trends that were identified in this paper, including an increase in the proportion of scholarly citations, and an increased emphasis on the broader view of the supply chain in terms of the subdisciplines which are cited, will likely continue in the future. In addition, as the quality of research appearing in the Journal continues to improve, it is probable that the proportion of citations to the Journal will increase in the future. Further, as the Journal is repositioned to surpass its current status as the top-tier supply management journal, to become an A-level, general business scholarly Journal, we will likely see new entrants into future sub-period citation networks representing additional top-tier business schools.

The introduction and application of SNA in this paper also provides numerous additional avenues for future research. First, managers can use SNA to map the interactions of the members within their supply management organizations. This is particularly important, as much of the work that occurs within an organization is only partially reflected in formal organizational charts (Cross, Borgatti and Parker 2002). Further, while, "managers may be able to diagram accurately the social links of the five or six people closest to them, their assumptions about employees outside their immediate circle are usually off the mark" (Krackhardt and Hanson 2000, p. 104). SNA is a powerful tool that can allow managers to identify the most central personnel within their organizations--those who act as central connectors, boundary spanners and information brokers (Cross and Prusak 2002). It is these personnel whom managers should closely cultivate and oversee (Interview 2005).

Researchers could similarly use SNA to map the interactions of supply management personnel to better understand relationships such as those between personal attributes and positions within an informal social network of a supply management organization, and centrality within a network and an employee's influence and power. Researchers might also employ theory from organizational design and structure (Thompson 1967; Van de Ven, Delbecq and Koenig 1976; Mintzberg 1979) to compare formal structure, as represented by organizational charts, to informal social networks with the goal of better understanding the manner and effectiveness by which the supply management function is organized (Johnson and Leenders 2005).

Second, SNA could be used to examine more expansive, supply chain phenomena. From a managerial standpoint, organizations could apply SNA to map their upstream supply networks, as was executed on a smaller scale by Choi and Hong (2002). For example, a buying organization could include all of its key first-tier suppliers in the network, perhaps based on a criterion such as purchase volume or strategic importance. In turn, these first-tier suppliers could be asked to identify their key suppliers, based on the same criterion. By mapping this upstream supply chain network, the buying organization would be in a position to identify and perhaps consolidate the spend of its first tier suppliers in order to negotiate volume discounts. Researchers might also model an organization's supplier network to examine the horizontal linkages among first tier suppliers, with the goal of better understanding the diffusion of technology within an organization's supply base (Rogers and Shoemaker 1971).

There are several avenues for future research applications of SNA to the broader supply chain. SNA might be used to better understand how firms manage their supplier relationships from a portfolio perspective. For example, Uzzi (1996, 1997) proposed that firms can enhance their long-term competitiveness by developing interorganizational networks that include both collaborative and arm's length relationships. Researchers could extend Uzzi's work by investigating the potential relationship between the composition of governance structure portfolios within supplier networks and supply management performance. Researchers could also consider how collaborative, existing relationships within a supply management network could improve or conversely constrain, performance by limiting the ability of the buying organization to establish new supplier relationships.

Finally, researchers could extend extant findings regarding key interorganizational constructs such as trust (Smeltzer 1997), commitment (Kwon and Suh 2004), conflict (Morris and Carter 2005), opportunism (Moore 1998) and communication (Claycomb and Frankwick 2004), to a network perspective. Once an upstream supply chain network has been mapped, the examination of the relationships among constructs could be considered across organizations. Uzzi (1996), for example, found that on an interpersonal level, a manager in one organization who maintained links to two managers in two separate organizations would act as a go-between and help to create direct trust between these two separate managers. Researchers could similarly examine the relationships between linked suppliers in a supplier network along the above constructs, to better understand how such supplier-supplier relationships might impact buyer-supplier relationships. Researchers could thus apply SNA to extend the current unit of analysis within supply management research beyond a single firm, or even a dyad, to a network. It is the hope of the authors that this paper will stimulate supply management research that uses this more holistic, network paradigm.

REFERENCES

Autry, C.W. and S.E. Griffis. "A Social Anthropology of Logistics Research: Exploring Productivity and Collaboration in an Emerging Science," Transportation Journal, (44:4), 2005, pp. 27-43.

Barney, J.B. "Firm Resources and Sustained Competitive Advantage," Journal of Management, (17:1), 1991, pp. 99-120.

Baumgartner, H. and R. Pieter. "The Structural Influence of Marketing Journals: A Citation Analysis of the Discipline and Its Subareas Over Time," Journal of Marketing, (67:2), 2003, pp. 123-139.

Borgati, S.P., M.G. Everett and L.C. Freeman. UCINET for Windows: Software for Social Network Analysis, Analytic Technologies, Harvard, MA, 2002.

Brass, D.J. "Being in the Right Place: A Structural Analysis of Individual Influence in an Organization," Administrative Science Quarterly, (29:4), 1984, pp. 518-539.

Burt, D.N., D.W. Dobler and S.L. Starling. World Class Supply Management, 7th ed., McGraw-Hill, New York, 2003.

Burt, R.S. "Stratification in Elite Sociological Methodology," in Toward a Structural Theory of Action, Academic Press, New York, 1982.

Bruning, J.L. and B.L. Kintz. Computational Handbook of Statistics, 2nd ed., Scott, Foresman and Co., Glenview, IL, 1977.

Carter, C.R. and L.M. Ellram. "Thirty-Five Years of the Journal of Supply Chain Management: Where Have We Been and Where are We Going?," Journal of Supply Chain Management, (39:2), 2003, pp. 27-39.

Carter, C.R., L.M. Ellram and W.L. Tate. "Structure and Influence: A Logistics Management Application of Social Network Analysis," Journal of Business Logistics, (28:1), 2007.

Carter, C.R., D.B. Vellenga, B.J. Allen and J.J. Gentry. "Affiliation of Authors in Transportation and Logistics Academic Journals--Another Look," Transportation Journal, (41:2 and 3), 2001, pp. 83-93.

Carter, C.R., D.B. Vellenga, J.J. Gentry and B.J. Allen. "Affiliation of Authors in Transportation and Logistics Academic Journals: A Reassessment," Transportation Journal, (44:2), 2005, pp. 54-64.

Carter, P.L. "A Letter from the Editor," Journal of Supply Chain Management, (34:1), 1999, p. 2.

Choi, Y.T. and Y. Hong. "Unveiling the Structure of Supply Networks: Case Studies in Honda, Acura, and DaimlerChrysler," Journal of Operations Management, (20:5), 2002, pp. 469-493.

Claycomb, C. and G.L. Frankwick. "A Contingency Perspective of Communication, Conflict Resolution and Buyer Search Effort in Buyer-Supplier Relationships," Journal of Supply Chain Management, (40:1), 2004, pp. 18-34.

Cross, R., S.P. Borgatti and A. Parker. "Making Invisible Work Visible: Using Social Network Analysis to Support Strategic Collaboration," California Management Review, (44:2), 2002, pp. 25-46.

Cross, R. and L. Prusak. "The People Who Make Organizations Go--or Stop," Harvard Business Review, (80:6), 2002, pp. 104-112.

Dyer, J.H. "Specialized Supplier Networks as a Source of Competitive Advantage: Evidence from the Auto Industry," Strategic Management Journal, (17:4), 1996, pp. 271-291.

Ellram, L.M., W.L. Tate and C. Billington. "Understanding and Managing the Services Supply Chain," Journal of Supply Chain Management, (40:4), 2004, pp. 17-36.

Email Correspondence, from Holly Johnson, Senior Vice President, Institute for Supply Management[TM], June 26, 2006.

Everett, J.E. and A. Pecotich. "A Combined Loglinear/MDS Model for Mapping Journals by Citation Analysis," Journal of the American Society for Information Science, (42:6), 1991, pp. 405-413.

Fearon, H.E. "Historical Evolution of the Purchasing Function," Journal of Purchasing, (4:1), 1968, pp. 43-59.

Flynn, B.B., S. Sakakibara, R.G. Schroeder, K.A. Bates and E.J. Flynn. "Empirical Research Methods in Operations Management," Journal of Operations Management, (9:2), 1990, pp. 250-284.

Freeman, L.C. "Centrality in Social Networks: Conceptual Clarification," Social Networks, (1:3), 1979, pp. 215-239.

Gulati, R. "Network Location and Learning: The Influence of Network Resources and Firm Capabilities on Alliance Formation," Strategic Management Journal, (20:5), 1999, pp. 397-420.

Gulati, R., N. Nohria and A. Zaheer. "Strategic Networks," Strategic Management Journal, (21:3), 2000, pp. 203-215.

Hite, J.M. and W.S. Hesterly. "The Evolution of Firm Networks: From Emergence to Early Growth of the Firm," Strategic Management Journal, (22:3), 2001, pp. 275-286.

Hult, G.T.M., R.F. Hurley, L.C. Guinipero and E.L. Nichols Jr. "Organizational Learning in Global Purchasing: A Model and Test of Internal Users and Corporate Buyers," Decision Sciences, (31:2), 2000, pp. 293-325.

Hult, G.T.M., D.J. Ketchen and E.L. Nichols Jr. "Organizational Learning as a Strategic Resource in Supply Management," Journal of Operations Management, (21:5), 2003, pp. 541-556.

Interview, conducted with Corey Billington, November 7, 2005.

Johnson, P.F. and M. Leenders. Supply's Organizational Roles and Responsibilities, CAPS Research, Tempe, AZ, 2005.

Kerlinger, F.N. Foundations of Behavioral Research, 3rd ed., Holt, Rinehart, and Winston, Chicago, 1986.

Krackhardt, D. and J.R. Hanson. "Informal Networks: The Company Behind the Chart," Harvard Business Review, (71:4), 2000, pp. 104-111.

Kwon, I.G. and T. Suh. "Factors Affecting the Level of Trust and Commitment in Supply Chain Relationships," Journal of Supply Chain Management, (40:2), 2004, pp. 4-14.

Leenders, M., H.E. Fearon and W.B. England. Purchasing and Materials Management, 9th ed., Irwin, Homewood, IL, 1989.

Leenders, M., P.F. Johnson, A. Flynn and H.E. Fearon. Purchasing and Supply Management, 13th ed., McGraw-Hill, Burr Ridge, IL, 2005.

Leavitt, H.J. "Effects of Certain Communication Patterns on Group Performance," Journal of Abnormal and Social Psychology, (46:1), 1951, pp. 38-50.

Mentzer, J.T., W. DeWitt, J.S. Keebler, S. Min, N.W. Nix, C.D. Smith and Z.G. Zacharia. "Defining Supply Chain Management," Journal of Business Logistics, (22:2), 2001, pp. 1-25.

Mintzberg, H. The Structuring of Organizations, Prentice-Hall, Englewood Cliffs, NJ, 1979.

Moore, K. "Trust and Relationship Commitment in Logistics Alliances: A Buyer Perspective," International Journal of Purchasing and Materials Management, (34:1), 1998, pp. 24-37.

Morris, M. and C.R. Carter. "Relationship Marketing and Supplier Logistics Performance: An Extension of the Key Mediating Variables Model," Journal of Supply Chain Management, (41:4), 2005, pp. 32-43.

Nohria, N. and C. Garcia-Pont. "Global Strategic Linkages and Industry Structure," Strategic Management Journal, (12:Special Issue), 1991, pp. 105-124.

Parente, D.H. and T.F. Gattiker. "Innovative Data Sources for Empirically Building and Validating Theories in OM," Journal of Operations Management, (22:1), 2004, pp. 111-112.

Pettigrew, A.M. "On Studying Managerial Elites," Strategic Management Journal, (13:Special Issue), 1992, pp. 163-182.

Phillips, D.M., H. Baumgartner and R. Pieters. "Influence in the Evolving Citation Network of the Journal of Consumer Research," Advances in Consumer Research, (26), 1999, pp. 203-210.

Phillips, D.M. and J.K. Phillips. "A Social Network Analysis of Business Logistics and Transportation," International Journal of Physical Distribution and Logistics Management, (28:5), 1998, pp. 328-348.

Rogers, E.M. and F.F. Shoemaker. Communication of Innovations: A Cross-Cultural Approach, The Free Press, New York, 1971.

Ronchetto, J.R., M.D. Hutt and P.H. Reingen. "Embedded Influence Patterns in Organizational Buying Systems," Journal of Marketing, (53:4), 1989, pp. 51-62.

Scott, J. Social Network Analysis: A Handbook, 2nd ed., Sage Publications, Thousand Oaks, CA, 2000.

Smeltzer, L.R. "The Meaning and Origin of Trust in Buyer-Supplier Relationships," International Journal of Purchasing and Materials Management, (33:1), 1997, pp. 40-48.

Thompson, J.D. Organizations in Action, McGraw-Hill, New York, 1967.

Uzzi, B. "The Sources and Consequences of Embeddedness for the Economic Performance of Organizations: The Network Effect," American Sociological Review, (61:4), 1996, pp. 674-698.

Uzzi, B. "Social Structure and Competition in Interfirm Networks: The Paradox of Embeddedness," Administrative Science Quarterly, (42:1), 1997, pp. 35-67.

Van de Ven, A.H., A.L. Delbecq and R. Koenig Jr. "Determinants of Coordination Modes within Organizations," American Sociological Review, (41:3), 1976, pp. 322-338.

Vellenga, D.B., B.J. Allen and C.D. Riley. "An Analysis of Author Affiliation for Publications in Transportation and Logistics Journals, 1967-1979," Transportation Journal, (21:1), 1981, pp. 44-53.

Wernerfelt, B. "A Resource-Based View of the Firm," Strategic Management Journal, (5:2), 1984, pp. 171-180.

Williams, A.J. and A.B. Oumlil. "A Classification and Analysis of JPMM Articles," Journal of Purchasing and Materials Management, (23:3), 1987, pp. 24-28.

Yaru, D. "Structural Modeling of Network Systems in Citation Analysis," Journal of the American Society for Information Science, (48:10), 1997, pp. 946-952.

AUTHORS

Craig R. Carter is an associate professor of supply chain management in the college of business administration at the University of Nevada in Reno, Nevada.

Rudolf Leuschner is a doctoral student in the Fisher College of Business at The Ohio State University in Columbus, Ohio.

Dale S. Rogers is a professor of logistics and supply chain management and Director, Center for Logistics Management at the University of Nevada in Reno, Nevada.

The authors gratefully acknowledge the assistance of David Schultz, ISM Librarian.

Table I MEAN NUMBER OF CITATIONS PER ARTICLE

                      Analysis of Variance (a)
                      Sum of   Mean
Source           DF   Squares  Square  F Value  p > F

Model              7  17,246   2464    66.07    < 0.0001
Error            148   5,519     37
Corrected total  155  22,765

[R.sup.2]: 0.7576
Coefficient of variation: 43.9507
Root MSR: 6.1065
Mean number of citations per article: 13.89

Time Period  Mean   Duncan's Grouping*

(8) 2000-04  39.47  A
(7) 1995-99  21.78  B
(6) 1990-94  12.95  C
(5) 1985-89   8.85  D
(4) 1980-84   7.70  D
(3) 1975-79   5.26  D
(2) 1970-74   7.65  D
(1) 1965-69   8.16  D

(a) Dependent variable = average number of citations per issue.
Independent variable = 5-year time period where 1 = 1965-69,
2 = 1970-74, 3 = 1975-79, 4 = 1980-84, 5 = 1985-89, 6 = 1990-94,
7 = 1995-99, 8 = 2000-04.
*Differences noted at a significance level of p < 0.05.

Table II CITATIONS BY CATEGORY

                        Total 1965-2004  1965-69  1970-74  1975-79
Category                (%)              (%)      (%)      (%)

Accounting               0.62             0.00     0.66     0.17
Book                    27.09            38.07    38.17    32.35
Business periodical      3.05             2.49     3.96     3.29
Conference proceedings   2.21             2.71     0.88     4.15
Dissertation             1.09             1.30     2.09     2.42
Economics                1.15             1.95     0.77     0.35
Finance                  0.17             0.00     0.22     0.17
Legal                    2.32             8.79     7.04     7.79
Legal -- Scholarly       0.24             1.30     0.11     0.35
Logistics and            3.16             0.22     0.00     0.35
  transportation
Management              11.70             3.80     6.71     6.40
Marketing                6.88             1.08     2.97     5.02
OM/OR                    6.30             1.41     1.76     3.81
Other                    2.12             5.97     1.87     4.15
Other -- scholarly       1.55             1.19     0.55     0.35
Psychology               1.38             1.95     1.76     0.52
Purchasing              12.88             1.84    11.77    11.42
Report                   4.32            10.85     3.85     3.63
Report -- website        0.39             0.00     0.00     0.00
Sociology                0.49             0.33     0.33     0.17
Trade publication       10.32            14.75    14.30    12.28
Working paper            0.59             0.00     0.22     0.87

                        1980-84  1985-89  1990-94  1995-99  2000-04
Category                (%)      (%)      (%)      (%)      (%)

Accounting               0.00     0.12     0.85     0.90     0.81
Book                    30.21    28.81    22.58    25.14    22.62
Business periodical      4.81     3.12     3.46     3.15     2.26
Conference proceedings   2.01     2.64     3.07     2.21     1.66
Dissertation             0.94     1.32     1.37     0.82     0.58
Economics                0.40     0.84     0.65     0.78     1.89
Finance                  0.00     0.12     0.00     0.04     0.41
Legal                    2.94     0.60     0.65     1.52     0.03
Legal -- Scholarly       0.40     0.00     0.07     0.16     0.12
Logistics and            1.20     0.36     0.85     4.18     6.65
  transportation
Management               9.09     7.68    11.81    15.19    15.04
Marketing                4.95     6.00     5.35     8.11    10.22
OM/OR                    4.28     6.60     7.51     5.65     9.52
Other                    4.14     1.32     1.63     2.25     0.70
Other -- scholarly       0.80     0.72     2.28     1.76     2.00
Psychology               2.27     1.80     1.11     1.19     1.22
Purchasing              16.31    12.24    19.13    12.61    13.18
Report                   3.88     5.64     4.63     3.89     2.76
Report -- website        0.00     0.00     0.00     0.90     0.64
Sociology                0.40     0.84     0.13     0.61     0.64
Trade publication       10.03    17.89    11.81     8.31     6.74
Working paper            0.94     1.32     1.04     0.61     0.32

Table III MEAN PROPORTION OF CITATIONS FROM SCHOLARLY JOURNALS

                    Analysis of Variance (a)
                      Sum of   Mean
Source           DF   Squares  Square  F Value  p > F

Model              7  2.8477   0.4068  27.47    < 0.0001
Error            148  2.1917   0.0148
Corrected total  155  5.0395

[R.sup.2]: 0.5651
Coefficient of variation: 31.6706
Root MSR: 0.1217
Mean proportion of citations from scholarly journals: 38.42%

Time Period  Mean    Duncan Grouping*

(8) 2000-04  59.83%  A
(7) 1995-99  50.90%  B
(6) 1990-94  48.69%  B
(4) 1980-84  38.20%  C
(5) 1985-89  35.72%  C D
(3) 1975-79  29.36%  D
(2) 1970-74  27.88%  D
(1) 1965-69  15.37%  E

(a) Dependent variable = average proportion of citations from the
journal per issue. Independent variable = 5-year time period where
1 = 1965-69, 2 = 1970-74, 3 = 1975-79, 4 = 1980-84, 5 = 1985-89,
6 = 1990-94, 7 = 1995-99, 8 = 2000-04.
*Differences noted at a significance level of p < 0.05.

Table IV RESULTS OF MANOVA: CATEGORIES OF SCHOLARLY JOURNALS BY TIME
PERIOD

                                   Numerator  Denominator
Statistic         Value   F Value  DF         DF           p > F

Wilks' lambda     0.1458   3.72    84           847.07     < 0.0001
Pillai's trace    1.3753   2.91    84         1,001        < 0.0001
Hotelling-lawley  3.0338   4.89    84           562.35     < 0.0001
Roy's greatest    2.1485  25.60    12           143        < 0.0001
  root

Results of Duncan's range test*

Logistics and transportation: Period 8 > Period 7 > Periods 1-6
Management: Periods 7 and 8 > Period 6 > Periods 2-5 > Period 1
Marketing: Period 8 > Periods 3-7 > Periods 1-2
OM/OR: Periods 8, 7, 6, 5, 3 > Periods 1, 2, 4
Purchasing/supply management: Period 6 > Periods 2-5 and 7, 8 > Period 1

*Duncan's range test was performed for statistically significant,
univariate analysis of variances (p < 0.05, adjusted for [alpha]
inflation). Period 1 = 1965-69, Period 2 = 1970-74, Period 3 = 1975-79,
Period 4 = 1980-84, Period 5 = 1985-89, Period 6 = 1990-94, Period
7 = 1995-99, Period 8 = 2000-04.

Table V MEAN PROPORTION OF CITATIONS FROM THE JOURNAL

                         Analysis of Variance (a)
                      Sum of   Mean
Source           DF   Squares  Square  F Value  p > F

Model              6  0.1009   0.0168  2.62     0.0199
Error            130  0.8361   0.0064
Corrected total  136  0.9370

[R.sup.2]: 0.1077
Coefficient of variation: 59.97
Root MSR: 0.0802
Mean proportion of citations from the Journal: 13.37%

Time Period  Mean (%)  Duncan Grouping*

(6) 1990-94  18.66     A
(4) 1980-84  16.02     B
(7) 1995-99  13.06     B C
(2) 1970-74  12.37     B C
(3) 1975-79  11.49     B C
(8) 2000-04  11.27     B C
(5) 1985-89  10.62     C

(a) Dependent variable = average proportion of citations from the
Journal per issue. Independent variable = 5-year time period where
1 = 1965-69, 2 = 1970-74, 3 = 1975-79, 4 = 1980-84, 5 = 1985-89,
6 = 1990-94, 7 = 1995-99, 8 = 2000-04.
*Differences noted at a significance level of p < 0.05.

Table VI NUMBER OF ARTICLES BY INSTITUION: 1965-2004

University                         Number of Articles (Rank)

Arizona State University           119 (1)
Michigan State University          104 (2)
Bowling Green State University      26 (3)
The Ohio State University           23 (4)
Pennsylvania State University       21 (5)
Western Michigan University         20 (6)
University of Arkansas              20 (6)
Florida State University            17 (8)
Indiana University                  17 (8)
University of Oklahoma              17 (8)
University of Minnesota             16 (11)
University of Wisconsin             15 (12)
University of Maryland              14 (13)
Harvard University                  12 (14)
University of Georgia               12 (14)
University of Western Ontario       12 (14)
Auburn University                   12 (14)
Air Force Institute of Technology   12 (14)
University of Bath                  11 (19)
University of North Florida         11 (19)
George Washington University        11 (19)
Virginia Tech                       10 (22)
Stanford University                  9 (23)
University of Birmingham             9 (23)
University of Michigan               9 (23)
Lehigh University                    9 (23)
Clemson University                   9 (23)
University of Cincinnati             9 (23)
Old Dominion University              9 (23)
University of Pennsylvania           9 (23)

Table VII EXAMPLE ADJACENCY MATRIX OF FIVE UNIVERSITY CITATION NETWORK

              University  University  University  University  University
              A           B           C           D           E

University A              15          0           4           0
University B  16                      0           0           0
University C   4           5                      0           0
University D   3           3          0                       0
University E   0           3          0           0

Table VIII MEASURES OF CENTRALITY FOR FIVE UNIVERSITY CITATION MATRIX

Actor (University)  Degree   Betweenness

University A        23.0000  3.0000
University B        26.0000  2.0000
University C         0.0000  0.0000
University D         4.0000  0.0000
University E         0.0000  0.0000

Table IX INSTITUTIONAL THOUGHT LEADERSHIP: CENTRALITY WITHIN THE
CITATION NETWORK

               Centrality     Centrality    Centrality     Centrality
               (Ranking)      (Ranking)     (Ranking)      (Ranking)
University     1965-2004      1980-84       1990-94        2000-04

Arizona State    0.1441 (1)    0.1059 (1)     0.1431 (1)     0.1120 (2)
  University
Michigan         0.1329 (2)    0.0686 (3)     0.1429 (2)     0.1592 (1)
  State
  University
Western          0.0358 (3)   N/R             0.0676 (3)     0.0264 (4)
  Michigan
  University
Bowling Green    0.0252 (4)   N/R             0.0291 (4)     0.0202 (7)
  State
  University
The Ohio         0.0244 (5)    0.0368 (11)    0.0031 (23)    0.0166 (8)
  State
  University
University of    0.0203 (6)    0.0476 (7)     0.0044 (17)    0.0394 (3)
  Maryland
George           0.0195 (7)   N/R             0.0095 (13)    0.0134 (9)
  Washington
  University
University of    0.0191 (8)    0.0537 (6)     0.0037 (21)    0.0043 (19)
  Minnesota
University of    0.0140 (9)    0.0174 (16)  N/R            N/R
  Arkansas
Louisiana        0.0123 (10)   0.0320 (13)    0.0113 (12)  N/R
  State
  University
Oklahoma         0.0112 (11)  N/R           N/R              0.0241 (5)
  State
  University
Indiana          0.0107 (12)  N/R           N/R            N/R
  University
University of    0.0103 (13)   0.0555 (5)   N/R            N/R
  North
  Carolina
University of    0.0098 (14)   0.0398 (9)     0.0025 (24)  N/R
  Nebraska
Pennsylvania     0.0092 (15)   0.0099 (20)    0.0159 (8)   N/R
  State
  University
University of    0.0090 (16)   0.0190 (15)    0.0131 (10)  N/R
  Georgia
University of    0.0087 (17)   0.0596 (4)   N/R            N/R
  Texas
University of    0.0084 (18)   0.0822 (2)   N/R            N/R
  Illinois
Florida State    0.0081 (19)   0.0040 (25)  N/R              0.0078 (10)
  University
Harvard          0.0072 (20)   0.0389 (10)  N/R            N/R
  University
University of    0.0072 (21)  N/R             0.0092 (14)    0.0050 (15)
  Toledo
University of    0.0066 (22)  N/R           N/R              0.0238 (6)
  Nevada, Las
  Vegas
Purdue           0.0060 (23)  N/R           N/R            N/R
  University
University of    0.0059 (24)  N/R           N/R            N/R
  Oklahoma
University of    0.0058 (25)  N/R           N/R              0.0030 (25)
  North
  Florida
Size of        443            85            147            201
  network
  (number of
  university)

N/R, not ranked in the top-25 for the sub-period.