budd.p65 308 College & Research Libraries July 1999 Increases in Faculty Publishing Activity: An Analysis of ARL and ACRL Institutions John M. Budd This study builds on a previous one that focused on publishing produc­ tivity of faculty at ARL institutions for the period 1991–1993. The present research expands the analysis to the period 1995–1997 and adds an examination of faculty activity at selected ACRL institutions for the two time periods. Measures of total publications and per capita publications per institution increased significantly for both groups over the two time periods. The increases indicate that, for what is likely to be a complex set of factors, faculty feel the need to communicate more and are turn­ ing to traditional print outlets for at least some of the communication. Deliberations on the future of scholarly communication should incorpo­ rate the views and practices of faculty. here is concern across all areas of higher education about what is frequently termed the “serials crisis,” but what is ac­ tually a crisis of access to information. Li­ brarians and faculty share the concern regarding institutions’ ability to afford access (through purchase, subscription, or some kind of licensing) to the products of research and scholarship. The serials dilemma is analyzed as part of a larger study by Stephen J. Bensman and Stanley Wilder, who offer an extensive review of the literature on the subject.1 One large- scale effort to respond to the challenge of access is the Scholarly Publishing and Academic Resources Coalition (SPARC), sponsored by the Association of Research Libraries (ARL) and endorsed by several other associations.2 This effort is aimed at promoting a more effective means of sharing research and scholarship, thus enhancing institutions’ ability to provide information access for their members. Another approach to the problem fo­ cuses on the production of information. Stephen D. Gruning reports that research is a principal factor in assessing both the undergraduate and graduate reputations of a sample of public and private univer­ sities.3 Meredith Jane Ludwig maintains that faculty research has the potential to contribute to an integrated and holistic fulfillment of institutions’ missions.4 John M. Braxton and Joseph B. Berger review the literature on faculty productivity and find that publishing and research activi­ ties have beneficial (improved course con­ tent and teacher effectiveness) and detri­ mental (lessened rigor of examinations and less prompt feedback to students) effects on teaching.5 The debate on the proper role for faculty and the proper emphasis on research and publication has John M. Budd is an Associate Professor in the School of Information Science and Learning Technologies at the University of Missouri-Columbia; e-mail: libsjmb@showme.missouri.edu. 308 mailto:libsjmb@showme.missouri.edu Increases in Faculty Publishing Activity 309 led some to reconceive the definition of productivity. Ernest L. Boyer ’s Scholarship Reconsidered and a more recent follow-up to that proposal, Scholarship Assessed, question the efficacy of reward systems that perpetuate the production of numer­ ous publications. 6–7 The central question asked here is: Has there been any observable change in recent years in publishing patterns by faculty? Studies of the effects of faculty publish­ ing may be enlightening, but they address only a portion of the issue. All commen­ tators on the subject can agree that the matter of access to information for schol­ arly purposes is complex. Although the efforts to investigate matters of access are necessary, there also is the need to exam­ ine the production of information objects (such as journal articles). The present study analyzes publishing patterns of fac­ ulty at selected universities. The central question asked here is: Has there been any observable change in recent years in pub­ lishing patterns by faculty? Description of the Study This study builds on an earlier one con­ ducted by John M. Budd.8 That project, prompted by observations that quantity of publication is important as an assessment of productivity, focused on measures of faculty publishing. It included an exami­ nation of publishing by faculty at ARL in­ stitutions for the period 1991 through 1993 and used Science Citation Index, Social Sci­ ences Citation Index, and Arts and Humani­ ties Citation Index as sources of data. The present study extends that earlier one in two ways. First, it looks to compare the earlier findings to a more recent time pe­ riod, 1995 through 1997. Second, it adds a study of publishing by faculty at institu­ tions included in the ACRL university li­ brary population. The latter group is stud­ ied over the two time periods. As was the case with the earlier study, this examination also uses the citation indexes as sources of data. The indexes allow for searching by corporate source, so each university can be searched to de­ termine the number of publications by its faculty. The same disclaimers and cave­ ats that applied to the previous study also hold here. The citation indexes do not al­ low for the measure of numbers of books, book chapters, and conference proceed­ ings. Further, they do not cover the uni­ verse of journal titles; rather, they index a small and select subset of that universe. For these reasons, the figures that are re­ ported here do not represent total publi­ cations but, rather, a portion of total pub­ lications. Because the citation indexes are selective in their coverage, it is inferred that publications in the indexed journals are some, albeit imperfect and limited, indication of quality. Other limitations of the earlier study apply to the present one. The main cam­ pus of each university is searched as a cor­ porate source. Branch campuses are not included. In some particular instances, such as the University of California, sepa­ rate campuses that are individual mem­ bers of ARL are searched separately. In some other instances, medical schools are not housed with the main campus, but may be in a different city. Attempting to include geographically remote medical schools remains problematic, so they are not included as part of the universities’ publication totals. If the medical school is included within the corporate source in­ dex for the main campus, the publications by medical school faculty are counted. As a result of this anomaly, institutional rankings are not absolutely accurate and representative. Rankings should be taken as relative measures of productivity. The same definition of publication is used in this study as in the earlier one. Anything that is designated as an “article” in the citation indexes is included. This means that edi­ torials, book reviews, bibliographies, and so on are not counted. Findings: ARL Institutions As was reported in the earlier study, the mean number of publications per institu­ tion for the 1991–1993 period is 4,595.8. 310 College & Research Libraries July 1999 number of publi-TABLE 1 cations per insti-Top Twenty ARL Institutions Compared: tution for the twoTotal Number of Publications time periods. 1991-1993 1995-1997 H 2 There is no Harvard 16,945 Harvard 21,913 statistically sig-UCLA 12,566 UCLA 13,620 nificant differ-MIT 11,788 Michigan 13,006 ence between theMichigan 10,907 UC Berkeley 12,237 mean per capitaU. of Washington 10,645 U. of Washington 12,117 number of publi-Cornell 10,518 Minnesota 11,369 cations for theUC Berkeley 10,378 Stanford 11,169 two time periods.Minnesota 10,304 Wisconsin 10,952 Stanford 9,723 Cornell 10,918 For purposes ofWisconsin 9,663 Johns Hopkins 10,576 testing and analysis,Johns Hopkins 9,636 Pennsylvania 10,247 the type I error level Pennsylvania 8,636 UC San Diego 10,059 (the probability of Illinois 7,884 Pittsburgh 9,148 rejecting a true null Columbia 7,824 Yale 8,938 hypothesis) used isYale 7,779 Columbia 8,886 0.05.UC San Diego 7,732 MIT 8,732 In each instance,UC Davis 7,621 Ohio State 8,552 a paired t-test is em-Ohio State 7,155 Penn State 8,543 ployed to determinePittsburgh 7,155 Illinois 8,400 statistical signifi-Penn State 6,925 UC Davis 8,389 cance. In the first case, the mean num- The range is 669 to 16,945. For 1995–1997, ber of publications per institution, the cal­ the mean number of publications is culated probability is less than 0.001. In 5,493.5, an increase of nearly 900 publica- the second case, the mean per capita num­ tions per institution. The range is 659 to ber of publications, the calculated prob­ 21,913. A comparison of the top twenty ability is also less than 0.001. Both of the institutions by total publications for the null hypotheses are rejected; there are sta­ two time periods is presented in table 1. tistically significant differences between A similar trend is apparent for per capita the two time periods. publications. For the earlier period, the The mean figures for total and per mean per capita number of publications is capita publications suggest the magni­ 3.56, with a low of 0.50 and a high of 12.71. tude of the difference over the two time In the more recent period, the mean rises periods. Because the same data source is to 4.20, with a low of 0.53 and a high of used for the two time periods, the reasons 12.94. Table 2 presents a comparison of the for the differences can only be speculated top twenty institutions by per capita pub- on. The pressure on faculty at research lications for the time periods. universities to publish is certainly great These data may hold some interest in today, but it also was substantial in the themselves but, more important, they al- early 1990s. It may be that there has been low for testing of some hypotheses. Two some turnover in the faculties; some in- hypotheses regarding the data from the dividuals in the senior ranks may have two time periods (stated as null hypoth- retired in recent years. The faculty who eses) are posed here: have replaced those who have retired may be more sensitized to the need to H1 There is no statistically signifi- publish. Further, the faculty who have cant difference between the mean been hired more recently may be in jun­ Increases in Faculty Publishing Activity 311 TABLE 2 Top Twenty ARL Institutions Compared: Per Capita Publications 1991-1993 1995-1997 Johns Hopkins Harvard MIT Washington U. (M.) UCLA UC San Diego UC Berkeley Stanford Minnesota Cornell Brown Princeton Chicago So. California UC Davis Virginia Utah Michigan Maryland Pennsylvania 12.71 11.46 11.26 10.24 7.51 7.34 7.06 6.92 6.90 6.81 5.79 5.46 5.16 5.04 4.96 4.82 4.79 4.64 4.61 4.61 Harvard Johns Hopkins Washington U. (M.) MIT Duke UC Berkeley Rochester UC San Diego UCLA Stanford Minnesota Cornell Brown Emory UC Davis Princeton Iowa Pittsburgh Chicago UC Riverside ior positions, and tenure and promotion may depend on their publication records. In addition, there have been shifts at some universities to hire individuals who are not on the tenure track. These individu­ als may be in nonpermanent positions. If there have been such moves in recent years, the remaining faculty may be more productive. This factor may affect per capita publications but may not explain increases in total publications. To reiter­ ate, these are only speculations. Other analyses of the ARL data are pos­ sible. Rank-order correlations can be cal­ 12.94 12.03 11.14 10.39 10.32 9.87 9.85 9.38 7.93 7.79 7.58 7.36 7.12 7.10 6.49 6.20 6.04 5.88 5.83 5.72 culated, comparing the rankings of the ARL institutions for the two time periods. When rank-order correlation is com­ puted for total publi­ cations, the correla­ tion coefficient is .98. This is a very strong positive correlation, suggesting that there is little variance in the rankings from 1991– 1993 to 1995–1997. The correlation coef­ ficient for per capita publications is .81, a slightly less strong positive correlation. The publication data can be compared with some other mea­ sures, two of which are selected for analy­ sis here. Rank-order correlation is calcu­ lated for total publications and volumes held by the ARL institutions’ libraries. The resulting coefficient is .70, which suggests that there is some variance in the two ranked lists. The coefficient for per capita publications and volumes in the libraries is .40, a rather weak correlation. When the ranked list of institutions by total publica­ tions is compared with that for numbers of doctorates awarded, the coefficient is .83. The coefficient for per capita publications compared with doctorates awarded is .45. Table 3 presents a comparison of the corre­ lation coefficients for the two time periods. TABLE 3 Rank-Order Correlations (ARL Institutions): Com(arisons over Time 1991-1993 1995-1997 Total Publications by Volumes in Libraries .69 .70 Per Capita Publications by Volumes in Libraries .42 .40 Total Publications by Doctorates Awarded .79 .83 Per Capita Publications by Doctorates Awarded .48 .45 312 College & Research Libraries July 1999 TABLE 4 Top Twenty ACRL Institutions Compared: Total Number of Publications 1991-1993 1995-1997 Tufts 2,883 Tufts 3,811 Virginia Commonwealth 2,521 Virginia Commonwealth 2,816 Carnegie Mellon 2,372 Carnegie Mellon 2,744 South Florida 2,359 South Florida 2,678 Kansas State 2,006 Kansas State 2,313 Vermont 1,702 West Virginia 1,965 Louisville 1,557 Vermont 1,851 West Virginia 1,515 St. Louis U. 1,846 St. Louis U. 1,485 UC Santa Cruz 1,832 UC Santa Cruz 1,447 Louisville 1,663 Clemson 1,358 Clemson 1,528 Wisconsin-Milwaukee 1,296 Northeastern 1,528 Northeastern 1,259 Wisconsin-Milwaukee 1,515 Rhode Island 1,255 Rhode Island 1,399 Baylor 1,224 Utah State 1,363 Utah State 1,147 Nevada-Reno 1,347 Arkansas 1,057 Arkansas 1,233 Nevada-Reno 1,024 North Texas 1,148 Wyoming 1,017 Lehigh 1,139 Akron 1,014 New Mexico State 1,102 In the previous study, it was found that chi-square tests performed on pairs of variables—publications by volumes in the libraries, publications by total library expenditures, per capita publications by volumes, per capita publications by to­ tal expenditures, and other library mea­ sures—resulted in no statistically signifi­ cant differences. Because of the earlier findings and the similarities between the trends evident in the two time periods, the chi-square tests are not performed here. Findings: ACRL Institutions ACRL institutions are included in the present study in order to gain insight into whether the trends that are evident among ARL institutions occur in other universi­ ties. Because the ACRL institutions are not research universities, it can be expected that the level of publishing productivity is not as high as it is at the ARL universities. This expectation is realized, but there is still an increase in publishing activity from one time period to the other. The mean num­ ber of total publications per institution in 1991–1993 is 874.0 (the range is 149 to 2883). The mean for 1995–1997 is 1074.9 (with a range of 165 to 3811). The top twenty insti­ tutions for each time period are presented in table 4. An upward trend also is apparent with regard to per capita publications. The mean for the earlier period is 1.59 (the range is .49 to 9.55); the mean for the later time frame is 1.78 (with a range of .42 to 12.33). Top twenty institutions are pre­ sented in table 5. As is the case with the ARL data, these figures allow for the testing of some hy­ potheses. The two hypotheses regarding the data from the two time periods (stated as null hypotheses) are similar to those expressed above: H 1 There is no statistically signifi­ cant difference between the mean number of publications per institu­ tion for the two time periods. Increases in Faculty Publishing Activity 313 TABLE 5 Top Twenty ACRL Institutions Compared: Per Capita Publications 1991-1993 1995-1997 Tufts Carnegie Mellon Clarkson UC Santa Cruz Virginia Commonwealth Vermont Texas-Dallas Lehigh Kansas State Baylor Utah State Nevada-Reno Rhode Island Colorado School of Mines Idaho Drexel Missouri-Rolla St. Louis U. Wyoming Clark 9.55 4.30 3.83 3.37 3.16 2.90 2.48 2.32 2.31 2.08 2.02 2.00 1.91 1.85 1.83 1.79 1.77 1.77 1.70 1.69 Tufts Carnegie Mellon UC Santa Cruz Clarkson St. Louis U. Alaska-Fairbanks Lehigh Nevada-Reno Vermont Drexel Rhode Island Montana State Colorado School of Mines Northeastern Texas-Dallas Kansas State Utah State Virginia Commonwealth Wisconsin-Milwaukee SUNY Binghamton 12.33 4.79 4.51 3.96 3.61 3.18 2.85 2.72 2.52 2.37 2.22 2.12 2.10 2.05 1.95 1.92 1.90 1.88 1.84 1.81 H 2 There is no statistically signifi­ cant difference between the mean per capita number of publications for the two time periods. The type I error level (the probability of rejecting a true null hypothesis) again is set at 0.05. Once again, a paired t-test is used to de­ termine statistical significance. In the first case, the mean number of publications per institution, the calculated probability is less than 0.006. In the second case, the mean per capita number of publications, the calcu­ lated probability is less than 0.01. The re­ sults are in keeping with those for the ARL data. Both of the null hypotheses are re­ jected; there are statistically significant dif­ ferences between the two time periods. The speculations offered above with regard to publishing by faculty at the ARL institutions also may hold with this population. There also may be increasing pressure to publish at universities that do not have a histori­ cally strong research emphasis. Rank-order correlations also can indi­ cate some relationships within the publi­ cation data and between those data and other measures. When the two time peri- Faculty constitute a large and politically influential segment of the campus community. ods are compared, there tends to be a fairly strong positive correlation. The co­ efficient for total publications is .88; the coefficient for per capita publications is .72. The indications of these tests are that the patterns exhibited by the ARL insti­ tutions also are present with the ACRL institutions, although the ACRL correla­ tions are a bit less strong. Further com­ parisons can be made between the pub­ lishing data and the numbers of volumes in the libraries and numbers of doctor­ ates awarded. These comparisons are pre­ sented in table 6. As is apparent from the table, the cor­ relations are not strong, suggesting that 314 College & Research Libraries July 1999 TABLE 6 Rank-Order Correlations (ACRL Institutions): Comparisons over Time 1991-1993 1995-1997 Total Publications by Volumes in Libraries .28 .36 Per Capita Publications by Volumes in Libraries .05 .00 Total Publications by Doctorates Awarded .42 .44 Per Capita Publications by Doctorates Awarded .11 .12 there is little connection between the mea­ sures of library holdings and doctorates awarded and those of publishing produc­ tivity. Discussion It is evident from the results presented here that faculty members at research universities and at universities without a traditional research emphasis are pub­ lishing greater numbers of items. The reasons for the increases, as stated above, are matters for speculation. However, the increases are of interest to academic li­ brarians. Faculty constitute a large and politically influential segment of the campus community. Their behavior pat­ terns almost inevitably have an impact on the focus of the library. If those be­ havior patterns are changing, it behooves academic librarians to consider the na­ ture of those changes and their possible effects on services, collections, and ac­ cess mechanisms. Even a change in mag­ nitude, such as the one suggested in this paper, may be important to the libraries on these campuses. However, there are other, more com­ plex, potential implications of these find­ ings. At the present time, everyone en­ gaged in research and scholarship—and the communication of the fruits of that research and scholarship—is concerned about the means of communication. Pub­ lication in print has some inherent limi­ tations, as well as some definite costs. On the other hand, faculty are publishing in print in increasing numbers. Everyone involved in the scholarly communication system needs to investigate whether the increase signals an affinity for the tradi­ tional means of communicating or whether it signals a set of behaviors that are constrained by the avenues of com­ munication which are currently open and by the rewards structures of individual campuses. There is no doubt that exist­ ing and emerging technologies can offer potential outlets for formal communica­ tion of research and scholarship. How­ ever, some essential points still need to be resolved, including: • Can the technology provide suffi­ cient access to the products of scholarly work? • Is the technological solution more economical than print? • Is the technological solution more effective than print for all academic dis­ ciplines? • Can the institutions transform their rewards structures to accommodate a variety of publication media? These are not the only questions that need to be addressed, but they provide a starting place. As deliberations and con­ versations progress, these questions should be addressed within the context of increasing publishing productivity by fac­ ulty. As principals involved in the produc­ tion of publications, the faculty should be involved in the decision-making processes that will shape scholarly communication. The data and findings presented here can provide something of a context in which the discussion can continue. Notes 1. Stephen J. Bensman and Stanley Wilder, “Scientific and Technical Serials Holdings Optimization: Increases in Faculty Publishing Activity 315 A LSU Serials Redesign Project Exercise,” Library Resources & Technical Services 42 (July 1998): 147–242. 2. See the SPARC Web site at: http://arl.cni.org/sparc/. 3. Stephen D. Gruning, “Research, Reputation, and Resources: The Effect of Research Activity on Perceptions of Undergraduate Education and Institutional Resource Acquisition,” Journal of Higher Education 68 (Jan.–Feb. 1997): 17–52. 4. Meredith Jane Ludwig, “Framing the Public Policy Debate on Faculty: What Is the Role of Research?” New Directions for Institutional Research, no. 90 (summer 1996): 67–77. 5. John M. Braxton and Joseph B. Berger, “Public Trust, Research Activity, and the Ideal of Service to Students As Clients of Teaching,” New Directions for Institutional Research, no. 90 (sum­ mer 1996): 79–91. 6. Ernest L. Boyer, Scholarship Reconsidered: Priorities of the Professoriate (Princeton, N.J.: Carnegie Foundation for the Advancement of Teaching, 1990). 7. Charles E. Glassick, Mary Taylor Huber, and Gene I. Macroff, Scholarship Assessed: Education of the Professoriate (San Francisco: Jossey-Bass, 1997). 8. John M. Budd, “Faculty Publishing Productivity: An Institutional Analysis and Compari­ son with Library and Other Measures,” College & Research Libraries 56 (Nov. 1995): 547–54. Data and findings related to ARL institutions for the 1991–1993 time period are taken from this source. http://arl.cni.org/sparc