Budd.indd Faculty Publishing Productivity: Comparisons over Time John M. Budd Concerns about higher education abound, and these include concerns about productivity.The present study extends two previous examinations of faculty publishing productivity covering the years 1991 to 1993 and 1995 to 1997. Both members of ARL and a group of institutions included in ACRL’s data set are included. For both groups there are some increases in mean total numbers of publications, although the rate of increase has decreased since the second time period. Per capita rates of publication demonstrate an even flatter pattern. In recent years, there have been some changes in the dynamics of universities’ faculties; there are more part-time faculty and more faculty who are not on the tenure track.These factors, coupled with the publishing data, point to activities that all aca- demic librarians should be aware of. he self-directed motivation and the institutional impetus in higher education to inquire and to publish the fruits of inquiry have not changed materially for a number of years. Tenure and promotion decisions, by and large, are still based on quantitative measures of productivity. Although the culture has remained quite constant, the information landscape has continued to shi . Over the past several years, serials publishers and aggregators have made thousands of journal titles available to scholars electronically. Indi- vidual academic libraries and consortia have access to more serials titles as a result of the licensing of services and packages. A result is that more libraries, and thus the scholars at the universities, have access to more serials than ever before. Further, the entire academic community has greater access to information than ever before. The realization of the magnitude of ac- cess can prompt some questions. Given the numbers of titles and the scope of the contents of those titles, where does the content come from? Universities are a primary source for the researchers who write and publish. This is not a surprise as research is one of the missions of uni- versities. Rankings by US News & World Report, the National Research Council, and others frequently include metrics that account for numbers of publications, among other variables. In short, more publications can lead to higher rank- ings of academic programs and entire institutions. For example, Robert G. Green, Melissa H. Bellin, and Frank R. Baskind reported that, in social work programs, twelve of the sixty-one doc- John M. Budd is Professor and Associate Director in the School of Information Science & Learning Tech- nologies at the University of Missouri-Columbia; e-mail: buddj@missouri.edu. 230 mailto:buddj@missouri.edu Faculty Publishing Productivity: Comparisons over Time 231 toral programs’ faculties are responsible for 43 percent of the journal articles published in the field. Further, the rank- ings of the programs correlated highly with publishing productivity.1 Robert K. Toutkoushian and others conducted an examination of faculty publications (total and per capita). They ranked the universities by the two measures and then compared those rankings to overall National Research Council and US News & World Report rankings. The authors concluded that: “Given the increasing emphasis on performance indicators, the measure of the ratio of publications to full-time faculty member can fill an important gap in how institutions are evaluated and compared.”2 The nature of rankings and the em- phasis on research and publication at some universities can be the subject of inquiry and commentary, but they are not the focus of the present examination. The principal purpose of this examina- tion can be simply stated. It follows up on two previously conducted studies and provides some comparison over time. The two central questions asked here mirror those of the previous works: (1) What is the publishing output of faculty affiliated with members of the Association of Re- search Libraries and with the data set pro- vided by the Association of College and Research Libraries? and (2) What changes in publishing pa erns have occurred over the years of the three studies? The Study This work builds on two previous stud- ies by John M. Budd. The first of the two published works examined faculty pub- lishing activity at Association of Research Libraries (ARL) institutions from 1991 through 1993.3 It used as data sources the Science Citation Index, the Social Sciences Citation Index, and the Arts & Humanities Citation Index. The second study also used the citation indexes and added institutions included in Asso- ciation of College and Research Libraries (ACRL) data reports. The time period for the later study was 1995 through 1997.4 By way of brief summary, it can be said that all publishing trends increased from the first to the second time period. For both the ARL and the ACRL populations, the mean institutional and the mean per capita publications rose from 1991–1993 to 1995–1997. The conclusion of the sec- ond study warned that: “As principals involved in the production of publica- tions, the faculty should be involved in the decision-making processes that will shape scholarly communication.”5 To date, faculty have had li le influence over the rewards structure and have reacted to university policies by increasing publica- tion rates. Also, technological applications have served to alter access mechanisms, but not to affect fundamental communica- tion practices, especially publication. The present study extends the previ- ous two to the period 2002 through 2004. The principal reason for investigating the matter of faculty publishing once again is to enable comparisons with the past. As is the case with the two previous studies, the citation indexes, available online as Web of Science, is the source of publication data for this investigation. As is stated above, the time period in ques- tion is 2002 though 2004. The coverage of Web of Science is not exhaustive; it includes many titles (nearly 9,000) in the sciences, the social sciences, and the arts and humanities. This is actually a fraction of the journals published; because of this limitation, a subset of potential publishing venues is included here. Web of Science covers journal literatures, so publication of book or book chapters is not included. The resource does cover many types of serial publication, but, as in the previous studies, only the publications designated as “articles” are counted. Book reviews, editorials, and so on are excluded. Web of Science enables searching by organization, so each university can be searched individually. In the cases of mul- ticampus systems, such as the University of Texas or the University of Wisconsin, specific cities also can be searched. Thus, 232 College & Research Libraries May 2006 institutions can be combined with cities so that the publications emanating spe- cifically from the University of Texas at Austin or the University of Wisconsin, Madison, can be obtained. Another caveat from the previous studies bears repeating here. The scope of a university’s activities may vary from institution to institution. To be specific, in some cases a medical school is a ached to a university’s main campus, so the publications by medical school faculty are counted. If, on the other hand, a medical school is located in a different city from the main campus, its publications are not counted. The numbers, then, contain some discrepan- cies, but adhering to this means of data gathering more readily enables compari- son over time. Findings: ARL Institutions In the two previous studies, the mean publications per institution were 4,595.8 and 5,493.5, respectively. Given the numbers of journals being produced and the continuing pressures on faculty to publish, it may be expected (all other things being equal) that the institutional mean would rise and, in fact, it has. For the 2002–2004 period, the mean number of publications per institution is 6,078.2. The range is 776 to 23,728. To offer an- other perspective on the magnitude of publishing activity, the total number of articles by faculty at the ARL member universities during the three years is more than 550,000. Comparison of the top twenty institutions for all three periods is presented in table 1. TABLE 1 Top Twenty ARL Institutions Compared: Total Number of Publications 1991–1993 1995–1997 2002–2004 Harvard 16,945 Harvard 21,913 Harvard 23,728 UCLA 12,566 UCLA 13,620 UCLA 15,083 MIT 11,788 Michigan 13,006 Washington 14,335 Michigan 10,097 UC Berkeley 12,237 Michigan 13,857 Washington 10,645 Washington 12,117 Johns Hopkins 13,760 Cornell 10,518 Minnesota 11,369 UC Berkeley 13,055 UC Berkeley 10,378 Stanford* 11,169 UC San Diego 12,947 Minnesota 10,304 Wisconsin 10,952 Pennsylvania 12,274 Stanford* 9,723 Cornell 10,918 Wisconsin 11,427 Wisconsin 9,663 Johns Hopkins 10,576 Columbia 10,990 Johns Hopkins 9,636 Pennsylvania 10,247 Cornell 10,795 Pennsylvania 8,636 UC San Diego 10,059 MIT 10,083 Illinois 7,884 Pittsburgh 9,148 Penn State 10,018 Columbia 7,824 Yale 8,938 Ohio State 9,589 Yale 7,779 Columbia 8,886 Florida 9,577 UC San Diego 7,732 MIT 8,732 Minnesota 9,479 UC Davis 7,621 Ohio State 8,552 Yale 9,377 Ohio State 7,155 Penn State 8,543 Pittsburgh 9,343 Pittsburgh 7,155 Illinois 8,400 Duke 8,952 Penn State 6,925 UC Davis 8,380 UC Davis 8,945 *Stanford withdrew from ARL between the second and third studies. Faculty Publishing Productivity: Comparisons over Time 233 TABLE 2 Top Twenty ARL Institutions Compared: Per Capita Publications 1991–1993 1995–1997 2002–2004 Johns Hopkins 12.71 Harvard 12.94 Harvard 11.88 Harvard 11.46 Johns Hopkins 12.03 Johns Hopkins 11.46 MIT 11.26 Wash. U. (MO) 11.14 Duke 9.86 Wash. U. (MO) 10.24 MIT 10.39 Wash. U. (MO) 9.72 UCLA 7.51 Duke 10.32 UC Berkeley 9.41 UC San Diego 7.34 UC Berkeley 9.87 UC San Diego 8.64 UC Berkeley 7.06 Rochester 9.85 Pennsylvania 8.60 Stanford* 6.92 UC San Diego 9.38 Case Western 8.36 Minnesota 6.90 UCLA 7.93 UCLA 8.06 Cornell 6.81 Stanford* 7.79 Columbia 7.47 Brown 5.79 Minnesota 7.58 UC Santa Barbara 7.35 Princeton 5.46 Cornell 7.36 Princeton 6.81 Chicago 5.16 Brown 7.12 Brown 6.61 So. California 5.04 Emory 7.10 Cornell 6.57 UC Davis 4.96 UC Davis 6.49 Minnesota 6.08 Virginia 4.82 Princeton 6.20 Yale 5.89 Utah 4.79 Iowa 6.04 Georgia Tech 5.81 Michigan 4.64 Pittsburgh 5.88 UC Irvine 5.69 Maryland 4.61 Chicago 5.83 Iowa 5.67 Pennsylvania 4.61 UC Riverside 5.72 Wisconsin 5.55 *Stanford withdrew from ARL between the second and third studies. It should be noted that Stanford’s ab- sence from the present study is a result of the university’s decision to withdraw from ARL. It also bears repeating that the inclu- sion of only the publications produced by faculty at a university’s main campus has an impact on rankings. For example, if the University of Texas medical school at Galveston were included, the University of Texas at Austin would rank eighteenth for 2002–2004 (the base number of 7,027 plus 2,410 for the medical school). Other institutions also would be affected; Loui- siana State University’s total would rise from 4,006 to 5,245 if the New Orleans medical school were counted. It also is apparent from the table that some insti- tutions moved (upward or downward) over time. The University of California, San Diego, not only increased its number of publications, but its ranking improved with each time period. The University of Illinois, on the other hand, saw its ranking fall to the point where it no longer ranked in the top twenty by the third study. In addition to total numbers of publica- tions, the mean per capita publications by institution can be calculated. The source for faculty numbers for each university is ARL’s interactive statistics.6 For the 1991– 1993 period, the per capita mean was 3.56; that mean rose to 4.20 in 1995–1997. For 2002–2004, the per capita mean is 4.24, only a slight increase over the previous time period. The range is 0.71 to 11.88. The top twenty institutions over the three periods is presented in table 2 (again, with Stanford absent for 2002–2004). 234 College & Research Libraries May 2006 As mentioned above, recent years have seen some changes in the composition of university faculty. The impact of the shi is not possible to determine, but the reported numbers of institutional faculty reported by ARL may include both ten- ured/tenure-track and non-tenure-track faculty. To use one institutional example again, it is apparent that the University of Missouri-Columbia reported the total number of faculty to ARL. An accurate count of faculty could influence the per capita figures. The impact of changing proportions of tenure-track versus non- tenure-track is discussed below. In Budd’s second study, a couple of hy- potheses were tested, based on compari- sons of the data from the 1991–1993 and 1995–1997 time periods. These hypotheses also can be tested with the 1995–1997 and 2002–2004 sets of data. The hypotheses, stated as null, are: H1 There is no statistically signifi- cant difference between the mean numbers of publications per institu- tion for the two time periods. H2 There is no statistically signifi- cant difference between the mean per capita numbers of publications for the two time periods. Also as in the previous work, the type I error level (the probability of rejecting a true null hypothesis) used is 0.05. To test the two hypotheses, the paired t-test (a test of differences between means) is employed. In the cases of both mean publications per institution and mean per capita publications, the calculated probability is greater than 0.05. Neither null hypothesis can be rejected; there are no statistically significant differences between the two periods. Some library- and institution-based data also can be compared to the publica- tion data. In other words, some additional examination can indicate whether there are strong correlations between publica- tions and such measures as volumes held by the libraries, total materials expendi- tures by libraries, and PhDs awarded by the universities. The first and third of the measures were analyzed in the second study by Budd; the second is added here to determine if there are any additional indications of connection. Rank-order correlation is used here to compare the rankings of universities by total publica- tions and per capita publications and the ranking by the three measures stated above. The correlations do not indicate any causal relationships, but the con- nections between publications and other measures may provide some indicators of linkages. The correlations are presented in table 3 (along with the correlations for volumes held and for PhDs awarded for the previous two time periods). As is evident from the table, where comparison is possible over time, the cor- relation coefficients are generally slightly higher. Focusing on the per capita mea- sures, the comparison between these and total volumes held by the library indicates TABLE 3 Rank-Order Correlations (ARL Institutions): Comparisons over Time 1991–1993 1995–1997 2002–2004 Total publications by volumes in library .69 .70 .72 Per capita publications by volumes in library .42 .40 .46 Total publications by PhDs awarded .79 .83 .83 Per capita publications by PhDs awarded .48 .45 .45 Total publications by materials expenditures .74 Per capita publications by materials expenditures .65 Faculty Publishing Productivity: Comparisons over Time 235 TABLE 4 Top Twenty ACRL Institutions Compared: Total Number of Publications 1991–1993 1995–1997 2002–2004 Tufts 2,883 Tufts 3,811 Alabama, Birm. 4,889 VCU 2,521 VCU 2,816 Tufts 3,864 Carnegie Mellon 2,372 Carnegie Mellon 2,744 Carnegie Mellon 3,552 South Florida 2,359 South Florida 2,679 Wake Forest 2,842 Kansas State 2,006 Kansas State 2,313 UC Santa Cruz 2,818 Vermont 1,702 West Virginia 1,965 VCU 2,779 Louisville 1,557 Vermont 1,851 South Florida 2,769 West Virginia 1,515 St. Louis U. 1,846 Kansas State 2,548 St. Louis U. 1,485 UC Santa Cruz 1,832 Louisville 2,327 UC Santa Cruz 1,447 Louisville 1,663 Clemson 2,010 Clemson 1,358 Clemson 1,528 Northeastern 2,000 UW, Milwaukee 1,296 Northeastern 1,528 St. Louis U. 1,988 Northeastern 1,259 UW, Milwaukee 1,515 Vermont 1,902 Rhode Island 1,255 Rhode Island 1,399 West Virginia 1,862 Baylor 1,224 Utah State 1,363 Drexel 1,598 Utah State 1,147 Nevada-Reno 1,347 Central Florida 1,531 Arkansas 1,057 Arkansas 1,233 Arkansas 1,506 Nevada-Reno 1,024 North Texas 1,148 Nevada-Reno 1,470 Wyoming 1,017 Lehigh 1,139 Georgia State 1,451 Akron 1,014 New Mexico St. 1,102 UW, Milwaukee 1,380 a mixed, but overall upward, trend. The coefficient for 1991–1993 is .42; the one for 1995–1997 is .40; the coefficient for the last time period is .46. The trend for per capita publications compared with PhDs awarded is similarly flat: .48 to .45 to .45. When comparing total publications and total volumes held by the library, the co- efficient rises over the three time periods from .69 to .70 to .72. A similar trend oc- curs with the comparison between total publications and PhDs awarded: .79 to .83 to .83. Again, these correlations do not indi- cate causation, but they may be somewhat indicative of some conjoined variance. That is, the factors that affect the number of volumes held by the library and doc- toral students that can matriculate and the factors that affect publication activity may be similar. The numbers of volumes held can be an indication of sustained financial support for the libraries so that a history of acquisitions can be chronicled. If the financial support for faculty activi- ties is similar to that for libraries and for doctoral students, publication also may be enhanced. To extend the point, the abil- ity of the institution to support doctoral students may be linked with the increased potential for collaborative publication (be- tween faculty and students). Because of the relatively high correlation coefficients related to total volumes, a measure is added to the present study—total materi- als expenditures. The coefficient for total publications compared to total materials expenditures is .74. The coefficient for per capita publications compared to total materials expenditures is .65. 236 College & Research Libraries May 2006 Findings: ACRL Institutions In the study of 1995–1997 data, Budd stat- ed that: “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 universities.”7 The institutions included in that investigation are not as extensively engaged in research as are the ARL mem- bers. As is the case with the ARL popula- tion, the ACRL institutional data can be examined over time. The mean for total publications per university in 1991–1993 was 874.0; for 1995–1997, the mean was 1,074.9. For the 2002–2004 period, the mean is 1,158.8 (the range is 122 to 4,889). The top twenty institutions by total pub- lications are presented in table 4. The trend for per capita publications is a bit less clear. The mean for the first time period was 1.59; the mean for the second time period was 1.78. In each of those stud- ies Tu s University had the highest per capita mean, with 9.55 and 12.33. Tu s’s total publications are in line with the past, but the reported number of faculty is higher at this time than in the past. The source of faculty sizes is the 2003 edition of Academic Trends & Statistics, published by ACRL.8 If the previous high per capita fig- ures for Tu s are controlled for, the means for the previous sets of years come down a bit, to be almost flat across all three time periods. The top twenty universities for the three time periods by mean per capita publications are presented in table 5. Once again, these data allow for the testing of two hypotheses. The two hy- potheses, again stated as null, are very similar to those used with the ARL data: TABLE 5 Top Twenty ACRL Institutions Compared: Per Capita Publications 1991–1993 1995–1997 2002–2004 Tufts 9.55 Tufts 12.33 Alabama, Birm. 6.39 Carnegie Mellon 4.30 Carnegie Mellon 4.79 Tufts 5.02 Clarkson 3.83 UC Santa Cruz 4.51 UC Santa Cruz 4.99 UC Santa Cruz 3.37 Clarkson 3.96 Carnegie Mellon 4.06 VCU 3.16 St. Louis U. 3.61 VCU 3.33 Vermont 2.90 Alaska-Fairbanks 3.18 South Florida 3.19 Texas-Dallas 2.48 Lehigh 2.85 Idaho 2.63 Lehigh 2.32 Nevada-Reno 2.72 Louisville 2.63 Kansas State 2.31 Vermont 2.52 Wake Forest 2.47 Baylor 2.08 Drexel 2.37 Northeastern 2.41 Utah State 2.02 Rhode Island 2.22 Lehigh 2.39 Nevada-Reno 2.00 Montana State 2.12 Vermont 2.39 Rhode Island 1.91 CO School Mines 2.10 West Virginia 2.63 CO School Mines 1.85 Northeastern 2.05 Kansas State 2.21 Idaho 1.83 Texas-Dallas 1.95 Central Florida 2.16 Drexel 1.79 Kansas State 1.92 Nevada-Reno 2.16 Missouri-Rolla 1.77 Utah State 1.90 Utah State 2.00 St. Louis U. 1.77 VCU 1.88 Georgia State 1.98 Wyoming 1.70 UW, Milwaukee 1.84 Wyoming 1.98 Clark 1.69 SUNY-Bing. 1.81 Clemson 1.96 Faculty Publishing Productivity: Comparisons over Time 237 H1 There is no statistically signifi- cant difference between the mean numbers of publications per institu- tion for the two time periods. H2 There is no statistically signifi- cant difference between the mean per capita numbers of publications for the two time periods. Also once again, the type I error level is set at 0.05. The paired t-test is employed as the means to test the hypotheses and to de- termine if there are statistically significant differences between means. As is the case with the ARL data, there are no statistical- ly significant differences between the two sets of means. This means that the two null hypotheses cannot be rejected. There is some increase in the mean total publica- tions, but, as noted above, the mean per capita publication rates are flat. Rank-order correlations do demon- strate some changes in the relationships of publication activity and some library and institutional measures. The correla- tion coefficients for the three time periods are presented in table 6. The correlations for mean total publica- tions and total volumes in libraries have not moved a great deal over the years. However, the correlations for mean total publications and PhDs awarded have ris- en over time: .42 to .44 to .51. The changes in the relationships between mean per capita publications and the two measures are more pronounced. Comparison with total volumes in libraries moved from .05 to .00 to .19. The changes with regard to PhDs awarded are even clearer, from .11 to .12 to .52. As is also the case with the ARL universities, comparisons with total materials expenditures reveal relatively strong correlations. The coefficient for mean total publications and expenditures is .65, and the coefficient for mean per capita publications and expenditures is .59. The indication is that the factors contributing to materials expenditures also may be contributing to publication activities. Discussion Although the form of journal literature has been altered somewhat (traditional print journals also are available online through a number of sources), faculty at universities continue to publish in substantial numbers. The increases have moderated, but increases in total pub- lications are still evident. The tripartite mission of universities—teaching, service, and research—continues to contain an emphasis on the third part. Given the magnitude of the activity, not only on ex- tensive research institutions, but on others as well, a well-supported infrastructure is needed. This infrastructure includes laboratories, equipment, and information services and access. The research and publishing activities of faculty rely on a level of institutional support that extends to all elements of the process of inquiry. The correlation coefficients, reported above, that relate to libraries’ materials TABLE 6 Rank-Order Correlations (ACRL Institutions): Comparisons over Time 1991–1993 1995–1997 2002–2004 Total publications by volumes in library .28 .36 .34 Per capita publications by volumes in library .05 .00 .19 Total publications by PhDs awarded .42 .44 .51 Per capita publications by PhDs awarded .11 .12 .52 Total publications by materials expenditure .65 Per capita publications by materials expenditure .59 238 College & Research Libraries May 2006 expenditures indicate a connection be- tween the financial support provided by the institutions and the productivity of the faculty. This financial support may be complex at this time. If there are large- scale alterations in the mode, media, and financing of publication in the coming years, the past trends may not be an adequate indicator of future activity. An explosion of Open Access journals could have an impact, for example, but only if the peer-review processes also are trans- formed. If, as Richard C. Atkinson has argued, the purpose is greater accessibil- ity to research results and scholarship,9 the findings presented here may not look foreign at some future date. The comparison presented here may be particularly important in light of some changes in higher education in recent years. Pressures and motivation to conduct research and to publish may not have diminished recently, but some shi s should be pointed out. Although tenure and promotion criteria do not seem to have altered appreciably, some other dynamics related to have shi ed. For example, according to the National Center for Education Statistics (NCES), the percentage of part-time faculty, as a proportion of the total, rose from 32.3 percent in 1995 to 44.5 percent in 2001.10 Other indicators should also be noted; the percentage of all faculty at four-year institutions with tenure fell from 53.5 percent in 1997–1998 to 48.7 percent in 2001–2002.11 Further, according to the IPEDS, the percentage of full-time faculty not on the tenure track increased from 21.9 percent in 1995 to 28.1 percent in 2001.12 By 2003, the percentage of full-time faculty at public doctoral universities was 30.3 percent.13 To take a single institution as one more example, the percentage of “nonregular” (non-tenure-track) faculty at the University of Missouri-Columbia increased from 17.5 percent in 1997 to 23.5 percent in 2002.14 Extrapolation from these figures is a ma er of speculation, but if we do assume that the reduction in the proportion of faculty on the tenure track is concomitant with publishing activity, the institutional average for the ARL members might jump from 6,078 to 6,455 (a 6.2% difference). If this difference also holds for the ACRL institutions, their mean might shi from 1,159 to 1,231. The aforementioned trends may not be surprising to anyone who has followed higher education activities over the past several years, but they do have relevance to this study. If a smaller proportion of an institution’s faculty is tenured or on the tenure track, it might be assumed that the total number of publications produced by a university would not increase over time. Given the downturn in the percentage of a university’s faculty being tenured or on the tenure track, curiosity about publish- ing pa erns begs for examination. For in- stance, if the trends in faculty demograph- ics continue, there could eventually be a downturn in the total numbers of articles published. Such trends might have an impact on the futures of specific journal titles. That outcome could be a simple one compared to others. If the shi from tenure-track to non-tenure-track faculty is uneven across disciplines, the intellectual and research bases of entire fields could be affected. In short, both supply (the numbers of kinds of publications being produced) and demand (the faculty who are active researchers and users of pub- lished research) could be altered in ways that affect the purposes of the university library. It is essential that our profession watch all of these interrelated phenomena closely so that libraries are not in a merely reactive stance. Notes 1. Robert G. Green, Melissa H. Bellin, and Frank R. Baskind, “Results of the Doctoral Faculty Publication Project: Journal Article Productivity and Its Correlates in the 1990s,” Journal of Social Work Education 38 (Winter 2002): 135–52. http:percent.13 http:2001�2002.11 Faculty Publishing Productivity: Comparisons over Time 239 2. Robert K. Toutkoushian and others, “Using Publications Counts to Measure an Institution’s Research Productivity,” Research in Higher Education 44 (Apr. 2003): 143. 3. John M. Budd, “Faculty Publishing Productivity: An Institutional Analysis and Comparison with Library and Other Measures,” College & Research Libraries 56 (Nov. 1995): 547–54. Data and findings from the 1991–1993 period are taken from this source. 4. ———, “Increases in Faculty Publishing Activity: An Analysis of ARL and ACRL Institu- tions,” College & Research Libraries 60 (July 1999): 308–15. Data and findings from the 1995–1997 period are taken from this source. 5. Ibid., 314. 6. Association of Research Libraries, “WWW Interactive Edition of ARL Statistics Data. Avail- able online from h p://www.arl.org/stats/arlstat/index.html. [Accessed 3 March 2005]. 7. Budd, “Increases in Faculty Publishing Activity,” 312. 8. Association of College & Research Libraries, 2003 Academic Library Trends & Statistics for Carnegie Classification: Doctorate-granting Institutions (Chicago: ACRL, 2004). 9. Richard C. Atkinson, “A New World of Scholarly Communication,” Chronicle of Higher Education 50 (Nov. 7, 2003): B16. 10. U.S. Department of Education, National Center for Education Statistics, Digest of Education Statistics: 2003 (Washington, D.C.: NCES, 2004), 291. 11. Ibid., 309. 12. U.S. Department of Education, Integrated Postsecondary Education Data System, “Chang- ing Composition of Faculty at Public and Private Four-year Institutions, 1989–2001, Two-year Intervals.” Available online from h p://www.aaup.org/surveys/zrep.htm. [Accessed 10 March 2005]. 13. U.S. Department of Education, National Center for Education Statistics, 2004 National Study of Postsecondary Faculty: Report on Faculty and Instructional Staff in Fall 2003 (Washington, D.C.: NCES, 2005), 12. 14. University of Missouri-Columbia, “Number of Full-time Ranked Faculty.” Available online from h p://www.missouri.edu/~ir/pages/personnel/faculty/ rnkten88.pdf. [Accessed 10 March 2005].