INTRODUCTION

This Study was undertaken to provide a basic understanding of the environment for research in the natural sciences at predominantly undergraduate colleges and universities. These institutions have served as a national resource for a significant proportion of students who undertake professional careers in the sciences (ref. 1) and a primary reason cited for their output has been the research experiences of undergraduate students with faculty mentors.(ref. 2) However, prior to the Study there was a growing perception that the number and quality of proposals from their science faculty was declining and that external funding opportunities for research were shrinking. Furthermore, several sources were reporting that faculty time was being stretched beyond its limits, and there was evidence from selected institutions of moderation in their number of science baccalaureates and a decline in the number of peer-reviewed publications from science faculty.

Concern over these perceived trends, expressed at a meeting in May of 1999 with representatives of private foundations having programmatic interests in the sciences, was the origin of this Study. Since the sampling of schools and programs that suggested these trends was limited, there was agreement that collection of relevant information would be required and that presentation of this data to an assembly of college presidents would be desirable. A conference at Oberlin College in 1985 of presidents from 48 selected private colleges and universities had previously received information that included self-reported data about research activity and the baccalaureate origins of PhDs in science and mathematics.(ref. 1) That conference and subsequent developments had impact on a number of programmatic and attitudinal changes that subsequently benefited the sciences at liberal arts colleges.

Five private foundations, The Camille & Henry Dreyfus Foundation, The Welch Foundation, the W. M. Keck Foundation, the M. J. Murdock Charitable Trust, and Research Corporation, have funded this project. Representatives of these foundations provided the intellectual guidance and oversight for the design and evolution of the Study into its present form. This evolution resulted from two subsequent meetings in 1999, during which three consultants were actively involved in assisting the foundation representatives to focus the Study. A meeting in March of 2000 with the presidents of ten private and public undergraduate institutions advised the foundation representatives that the direction being taken in the Study would be of benefit to their institutions and that they were supportive. The Study was operated from Research Corporation because of space and personnel considerations, the latter relating to the availability of two talented scientists who have managed the Study, the data acquisition, and the presentation of Study results.

Overview of the Study Forms, Surveys and Institution Selections. Two Survey forms were created to access information not readily available in order to assess the environment for research at predominantly undergraduate institutions. The Institutional Survey would provide critical information about the flow of funding into the natural sciences, information on internal budgets, faculty and staff sizes, library budgets and holdings, student involvement in research, and perceived changes in time allocations and institutional trends. The Faculty Survey would provide critical activity on individual faculty activities in obtaining external grants, publishing peer-reviewed papers, student involvement in research as evidenced by coauthorship of these papers, invited talks by faculty, and perceived changes in time allocations and institutional trends. Both surveys invited essay responses to a broad range of considerations. Separately the surveys were intended to provide different vantage points with which to assess the environments for research in the natural sciences at the surveyed institutions. Together, we anticipated that the surveys could provide a multidimensional assessment that would assist foundations, government, and especially the institutions, both administration and faculty, to understand the complex environments that exist in the natural sciences as well as their singular impact and placement relative to other institutions.

From the start we recognized that there was an abundance of information relating to the Study that was publicly available. This included enrollment and degree demographics, baccalaureate origins of PhDs in the sciences, grant awards from several private foundations and government agencies, and institutional endowments. Some of this data was readily accessible, but most was not. One of the great challenges of this endeavor was to obtain this information for the surveyed institutions and, whenever possible and appropriate, for the entire community of predominantly undergraduate institutions.

The Study was restricted to the natural sciences, and the primary reasons for this were the similarities of the natural sciences in requirements (facilities, laboratories, research performance, for example) combined with the interests of the sponsoring foundations. For the vast majority of the surveyed institutions, the departments of biology, chemistry, earth science or geology, and physics constitute the natural sciences. At other institutions natural sciences also include agricultural sciences, astronomy, biochemistry, environmental sciences, or neuroscience or a combination of these. The Study did not include mathematics, computer or information science, engineering, or any other departments or departmental units that could not be identified with the natural sciences. However, interdisciplinary natural science units such as neurobiology that existed in a department outside of the survey were included.

Following the March 2000 meeting with representative college presidents and receipt of their advice, the sponsoring foundations set out to identify those institutions, public and private, that would be included in the Survey. The only significant criteria were that there would be perceived research activity and that there be representation from a broad array of private and public, bachelors degree-granting and advanced degree-granting institutions. A few doctoral institutions were included, but no Research I (Carnegie Classification) institution was selected. Geographical considerations were also relevant. Foundation representatives added and removed institutions known to them and to their organizations. In the end the presidents of 191 institutions were invited to participate in the Study and the attendant Conference, and 159 responded affirmatively. When deadlines for submission of survey information could no longer be extended, 136 institutions had completed the institutional survey; and 132 institutions had submitted faculty surveys.

The institutions that were not able to participate rarely gave reasons, but those that did informed us that there was an absence of requested information at their institution or that they did not have sufficient time or personnel with which to conduct the surveys. One of the institutions informed us that "Much of the data requested are not on file... In some instances, we only keep records for a period of one or two years..." However, there was no obvious pattern from the schools that declined to participate, and the reasons for their declination must be considered to be due to individual circumstances and interests.

The Institutional Survey and its Reliability. Invitations to presidents of 191 institutions were sent at the end on May in 2000, with follow-up correspondence in July and August, and the surveys were sent to them following their agreement to take part in the Study. Return of the completed Institutional Survey was requested for mid-September, and many were received by this time, but the bulk arrived during the ensuing month. There is no doubt that this was a laborious process for each participating institution. Unfortunately, institutional size and activity increased the workload, but as highly productive faculty who provide annual reports also understand, this is the price paid for prominence.

Various interpretations were used by institutions in completing the data requested in the Institutional Surveys, and not all information requested was available to the institution. Extensive cross checks were used to determine the completeness of contributed data. In some cases, for example, the narrative mentioned a renovation or construction that was not included in the section on grants and awards. In such instances contact was made with the institution to clarify this matter. In addition, a letter was sent to all participating institutions requesting additional information that might not have been included in the original submission. The additional information received (about 5 % of the total entered) added to our sense of confidence in the completeness of the data. However, we observed that "General Institutional Support for Physical and Life Sciences" was not available to several institutions prior to 1994.

How reliable is this data? We are certain that some institutions have not included all of the appropriate funding made available to the natural sciences and their faculty, and that there were other institutions that included funding more appropriate to other instructional units. Overall, however, we are confident that we have accessed the vast majority of available resources. As cross checks on reporting, reported institutional data for faculty research grants obtained from the Petroleum Research Fund of the American Chemical Society as well as from the Research Corporation was compared with public records from these funding sources; there was greater than 90% correspondence. Reporting for awards received from the Camille & Henry Dreyfus Foundation and from the Howard Hughes Medical Institute offered similar levels of confidence.

All of the data was checked and rechecked by those managing the Study as well as through computational assessments that could identify "outliers." When there was such a discrepancy, the institution was contacted to ascertain the validity of the information. Still, we remain skeptical about the uniform validity of some categories of information. This is especially true for categories 3-7 of General Institutional Support for Physical and Life Sciences and for the number of junior and senior majors and the number of research students (academic year) from Instructional Effort. In cases where the data requested are not routinely accessed or for which there are multiple interpretations, there is decreased reliability.

The vast majority of respondents provided one-to-three page narratives that addressed issues related to the Study. As a primary goal, the Study sought the institution's perspective on "recent trends and evident needs in the physical and life sciences." In some cases we were informed that the narratives represented consensus impressions of several persons or groups; in most they are the views of the institutional respondent. The narratives were abstracted for the SourceBook according to the primary focus of the comment with any reference that could identify the source expunged to ensure confidentiality. The number of times a trend or need was listed is taken as a reflection of the intensity of the concern expressed, and this is captured in Section 7 by noting the volume of comments in the selected categories. Most institutions provided advice to answer the question of "How could external public and private funding sources best structure their programs to adjust to recent trends and evident needs?"

Faculty Perspectives and their Reliability. Faculty surveys were due one month following the Institutional Survey, and considerable flexibility was provided to institutions to assist them in the reporting process. Overall nearly 70% of natural science faculty at the surveyed institutions responded - nearly 3,000 - and their efforts are greatly appreciated. Percent response varied from institution to institution, from 20% to over 90% with four of the 136 surveyed institutions having either no or an unusable faculty response. We are aware of heroic efforts made by several institutions to insure a high response. We are also aware of institutions at which faculty were concerned about unintended uses of the surveys, at which faculty resented this encroachment on their limited time, and at which there was little incentive or instruction given to faculty to prepare their survey forms.

Here the labor of the Study increased by at least an order of magnitude. Few institutions provided faculty responses using the recommended formats, and often the faculty included their information in a form associated with them individually or that was peculiar to their institution. Consequently, there was a significantly higher degree of interpretation required for the faculty survey than for its institutional counterpart. In many instances, the appropriate category for grant awards was not provided, and institutional or departmental awards were listed by faculty as individual grants. This was particularly true of large awards from foundations such as HHMI and in listing grants received for major instrumentation. In some cases, faculty included publications that could not be judged as having been peer-reviewed; if this judgement could be confirmed, these publications were not included in the survey results. Finally, our intent to designate invited presentations was missed by most faculty who generally listed every place and time that they gave a talk, poster, or workshop.

Institutions with a high percentage of faculty response provide a good indication of faculty activity and attitudes. The institutional data drawn from their faculty have a validity that is high. For institutions that had low faculty response there is a corresponding uncertainty with their composite data. We do not have a rigorous measure of which faculty, if any, are underrepresented in providing their Faculty Perspectives. There were very few faculty surveys submitted from departments that did not fit the Study criteria for natural sciences.

Data Entry. The principal task confronting those directly involved with the Study surveys was the correct coding of the data and their entry into the data base that was prepared for future access. Early in the period of data collection from Institutional Surveys we recognized that the information being provided could be categorized and subcategorized to optimize search characteristics. An extensive coding system was developed for the Institutional Survey to designate the source and purpose of grants and funding. The coding system was consistent with the initial categories on the survey forms but extended beyond them to specify the nature of the activity (for example, the specific funding source, whether the award was for basic research or education, if the purpose was the acquisition of instrumentation, curriculum development, or basic/applied research). Another coding system was devised for the Faculty Perspectives to allow entry of such items as faculty rank, number and dollar amount of grants received, total publications and research publications, and invited talks. Overall, more than 110,000 individual items were separately coded. Each was checked and rechecked for accuracy and consistency by persons other than the one making the entry, and we would estimate that there is a high uniformity in these designations.

Certain categories could not be used effectively because of different meanings that they have in different disciplines. One of these was the "on campus/off campus" designation. Originally intended to allow differentiation between research at the home institution and research carried out at another academic institution, a company, or a national facility, we were rapidly made aware that field research was generally considered to be "off campus." Consequently, although the on campus/off campus differentiation may have disciplinary relevance, overall significance related to research performed on campus or off campus cannot be judged.

In addition to the categories that required specific coding, there were extensive efforts to ensure that all of the data provided could be entered into accessible tables.

Innumerable items were entered, and these entries were checked and then searched computationally for "outliers" and inconsistencies. Following entry into individual institutional tables in Excel, the composite was then entered into the FoxPro database management system. The resulting database was the source from which the outcomes of composite institutional and faculty information have been taken.

Cluster Analyses. As data entry progressed, we were made aware of the potential of the information in the Study to identify institutional similarities and intrinsic characteristics of the participating schools. Were there patterns and potential predictors that we were missing? Accustomed to two-dimensional analyses, we were concerned that critical information imbedded in a multidimensional data set could be missed entirely. Thus as data entry was being completed we contracted advanced analytical and modeling efforts with East Pond Analytics, Inc. With knowledge of the operating systems we proposed two institutional models - one based on enrollment without reference to factors directly related to research activity and the other based on research activity at the institution. Referred to as the Enrollment Model and the Research Model, the composite sets of institutional data were entered into a multidimensional matrix, indexed, rotated, and conjoined to produce statistically relevant clusters of institutions whose relevant characteristics are presented in Section 6. Clusters are derived from separate analyses of all institutions, private institutions, public institutions, bachelors degree granting institutions, and advanced degree granting institutions. Each institution belongs to three clusters in each model, and each model has more than ten relevant characteristics averaged for each cluster. The outcome is an institutional profile with more than 60 statistically relevant characteristics.

The overall intent of this Study was to learn more about the natural sciences at predominantly undergraduate institutions. The information in the SourceBook serves as a resource for decision-making in predominantly undergraduate institutions, private foundations and government agencies, and professional associations. If there is sufficient interest, a more complete report will be published in the future.

Michael P. Doyle

_____________

1. David Davis-Van Atta, Sam C. Carrier, and Frank Frankfort, "Educating America's Scientists: The Role of the Research Colleges," report of The Future of Science at Liberal Arts Colleges Conference (Oberlin College, Oberlin, OH, 1985).
2. Academic Excellence, Michael P. Doyle, Editor: Research Corporation, Tucson, AZ, 2000.

Academic Excellence: A Study of the Role of Research in the Natural Sciences at Undergraduate Institutions

Executive Summary

This Study was undertaken to provide a basic understanding of the environment for research in the natural sciences at predominantly undergraduate colleges and universities. These institutions have served as a national resource for a significant proportion of students who undertake professional careers in the sciences, and a primary reason cited for their output has been the research experiences of undergraduate students with faculty mentors. However, prior to the Study there was a growing perception that resources and productivity were declining. Concern over these perceived trends by five private foundations with interests in the natural sciences prompted the intensive data collection and analyses that are contained in the SourceBook.

Information was obtained from both public and private sources. Substantial data and statistics from private foundations and government agencies, not previously compiled, and from comprehensive surveys completed by 136 institutions and their faculty are compiled and summarized in the SourceBook, along with abstracts of statements and concerns received from a vast majority of these institutions. The institutions surveyed were private (104) and public (32), those offering only bachelors degrees in the natural sciences (98) and those offering advanced degrees (38), and the period covered was the decade of the 1990s. In all, more than 100,000 items of data were included on topics ranging from budgets, grants, and funding to faculty demographics and student involvement. An advanced statistical analysis of the data according to two models, the Enrollment Model and the Research Model, has identified unique subgroups of institutions (clusters) and their characteristic profiles. Some of the more striking preliminary observations made by this Survey are:

Publication Productivity and Grant Activity at Surveyed Institutions

Publication Productivity

Data from nearly 3,000 faculty in the natural sciences at 133 predominantly undergraduate colleges and universities show significant discipline- and gender-based differences in peer-reviewed publications. This information was submitted as one part of a larger survey entitled Academic Excellence: A Study of the Role of Research in the Natural Sciences at Undergraduate Institutions. Faculty surveyed were tenured or tenure-track. Sixty-seven percent of all faculty in the natural sciences at the surveyed institutions completed the surveys. The publication data was reviewed to confirm peer review and the probable origin of the publication at the surveyed institution, rather than from the faculty member's doctoral or postdoctoral work.

Tables 1 and 2 were constructed from the sum of institutional averages. In other words, each faculty member is counted within their respective institutions, and the summary data for each institution (publications per faculty member per year) is averaged according to numbers of institutions reporting at least one faculty member of that gender in the specified rank or discipline. Not all institutions had faculty in all of the listed disciplines, so that of the 133 surveyed institutions, only 19 accounted astronomy, 28 had geosciences faculty, 15 reported environmental science, and 21 provided information on neuroscience faculty. However, all institutions provided faculty reports for biology, chemistry, and physics.

Since the categorization of faculty is dependent on individual institutional administrative assignments, faculty in the categories of astronomy, geosciences, environmental science, or neuroscience are not necessarily representative of the whole. Faculty with the same work functions, research interests, and professional affiliations may be found in the traditional natural science disciplines of biology, chemistry, and physics. If this data has a fault it is that only publications of faculty at the surveyed institutions are counted. If a faculty member came from an institution that was not surveyed, only publications from the surveyed institution were counted; and if a faculty member left a surveyed institution, that person was most often not included.

Table 1. Publications per faculty member per year, adjusted for years of service (1990-2000), averaged by institution.

There are significant discipline-related differences in publications per faculty member per year. Faculty in the less-populated divisions - astronomy, environmental science, and neuroscience - have a significantly higher publication rate than do faculty in the traditional natural science disciplines. In biology, chemistry, and physics there is an overall decrease in the publication rate with increasing faculty size in those departments, but the factor or factors responsible for this (for example, expectations, workload, number of journals, financial support) are not known. Nor is the quality of publications (especially based on citations) known from this study. Since those reviewing the reports of peer-reviewed publications had more experience with chemistry and physics disciplines, it is more likely that more non-peer reviewed publications leaked into the final count in biology.

Women publish less frequently than men in all disciplines except neuroscience. Why this is so is unknown, but the data obtained is statistically relevant. Underreporting by faculty units is unlikely in the natural sciences, and spot checks on specific faculty via their web site did not reveal discrepancies in reporting. There is a possibility that more women did not respond to our survey than men, but this would suggest an even higher proportion of women in the disciplines that ours (see Figures 5.2 through 5.8 of The SourceBook) and other data (for example, Women in the Chemical Workforce, a Workshop Report to the Chemical Sciences Roundtable, National Research Council, Washington, D.C., 2000, and demographic data from the American Institute of Physics at http://www.aip.org/statistics/trends/undtrends.htm) suggest.

Table 2. Publications per faculty member per year, adjusted for years of service (1990-2000), averaged by institution.

As might be expected, publication productivity increases with rank, indicating that tenure and promotion decisions are based, at least to some extent, on faculty publications. Women publish less frequently than men in all ranks with the gender difference being greatest at the associate professor level. Here again, the factor or factors responsible for this are unknown.

A separate assessment of publication productivity by discipline and gender is reported in Table 3. Here all faculty are accounted by discipline, independent of institution. Total years of employment for the period 1990-2000 is given along with the total number of publications resulting from faculty activity. The ratio presented is that from the number of publications divided by the total years of employment. Note that dividing the total years of employment for women by those for men does not provide a ratio of persons but does of person-years.

Table 3. Publications per faculty member, adjusted for years of service (1990-2000), based on total composite of faculty in the category

Although there are some differences in ratios representing the number of publications per faculty member per year between Tables 1 and 3, the largest discrepancies are in disciplines that are not highly populated and can be associated with "the tyranny of small numbers." The ratios for the traditional disciplines are remarkably alike. Thus two ways of measuring publication productivity lead to the same outcome and conclusion.

Grant Activity

The same population of faculty who reported their publications for the Academic Excellence study also reported their external grant activities. Self-reported grant activities were reviewed did not include grants designed for larger organizational units, such as those to departments, institutions, or consortia (for example, from NSF-ILI, NSF-REU, HHMI, Research Corporation Department Development). Internal "grants" such as start-up provisions and student or faculty summer stipends were not included. Both education/outreach and research grants were counted. Research grants accounted for 89% of the total dollars reported. Additional details can be found in Section 5 of The SourceBook.

The total external grant dollars was $90,724,791 (1,662 grants) reported for women and $279,193,853 (5,472 grants) reported for men. The total external funding that was reported by the institutions was $559,442,530 for the same institutions plus three others, so the amount reported by faculty is sixty-six percent of the total reported by institutions. Since the total number of faculty reporting is sixty-seven percent of the total number of faculty at these institutions in 2000, we believe that the numbers reported are representative. The percent response of faculty for each surveyed institution is given in Table 5.1 of The SourceBook.

Table 4. External grant dollars and number of grant awards per faculty member per year, adjusted for years of service (1990-2000), averaged by institution.

In all but categories but Environmental Science Departments women have more grant dollars per faculty member per year, and they received more grants. There is consistency in these differentials through disciplines and by ranks. The difference is not due to the availability of gender-specific programs, since such awards constitute a very small fraction of the total.

An important revelation comes from the per capita grant dollars available to chemistry faculty, which is the lowest of all of the disciplines. There is a common perception that there are more foundations and agencies that support research in chemistry than in any other discipline. In other words, it's thought to be easier to obtain funding in chemistry than in other disciplines. It may be easier, although certainly not greatly so, but the per capita outcome is smaller in chemistry than in any other discipline in the natural sciences.

Zipf's Law Revealed at Predominantly Undergraduate Institutions: Million-Dollar and Thirty-Publication Faculty

The amplification of positive deviants, Harvard Professor and Nobel Laureate Dudley Herschbach reported ("Understanding the Outstanding:Zipf's Law and Positive Deviants," The SourceBook, pg. 70-74), is a desirable strategy. "It emphasizes," he says, "the key role of 'outliers,' whose example can empower a community or institution to achieve performance well beyond presumed limitations." But who are these "outliers" whose activities allow them to stand out as exceptional? Having grant and publication data from nearly 3,000 faculty in the natural sciences at 133 predominantly undergraduate colleges and universities, we set out to identify the characteristics of faculty who received $1,000,000 or more in external grant support during the 1990s. This selection required that the faculty member received $100,000 annually through ten years, which is, however, only about four times the average amount of research support per faculty member per year reported in The SourceBook (Table 5.2). Who are they? From where do they come? How are they supported? What is their involvement with undergraduate students?

Consistent with Zipf's Law, the number of faculty receiving $1,000,000 or more in total external grant support is only 51 out of the 2,980 natural sciences faculty responding (1.7 %). Twenty-three are at seventeen, mainly (15/17) bachelors-degree granting, private institutions (out of 104 total), and twenty-eight are at eleven, mainly (9/11) advanced-degree granting, public institutions (out of 32 total). Thirty four of these million dollar faculty (67 %) are at only eleven of the 133 responding institutions (8 %), and nine are in departments that offer the Ph.D. degree. One institution reported seven faculty who had $1,000,000 or more over ten years; another reported five. Seventeen institutions reported one each of the million-dollar faculty members. Relevant characteristics of rank, gender, and discipline are in Table 1.

Table 1. Rank, gender, and discipline characteristics of the 51 million-dollar faculty.

A word of explanation is in order here. What is meant by external grants? In The SourceBook we define this accounting as "traditional faculty-driven external proposals." Examples include NSF-RUI grants, but not NSF-REU grants, NSF-CAREER Awards, but not NSF-MRI grants, NIH-R01 and NIH-AREA grants, but not group NIH-MBRS grants. Grants awarded to departmental units or to larger entities, such as HHMI awards and NSF-ILI grants, are not included in the compilation of individual faculty awards. The distinction made is not related to who wrote the grant proposal, but how it is considered by reviewers and the funding agencies in making an award. For example, a NSF-ILI proposal is judged primarily for the educational impact in academic departments rather than on the professional development of the "principal investigator," and often the designated contact is not the person who wrote the proposal. On the other hand, review of a NSF-RIU or NIH-AREA proposal is based primarily on the proposed activity and the ability of the "principal investigator" to accomplish what is proposed.

Individual total external funding ranged from $4.26 million to $1.03 million with the median at $1.4 million. The average number of grants per faculty member was 18 with a range of 2 to 126. Twenty-five faculty had ten or more grants during the 1990s. By far the greatest source of funding for the million-dollar faculty was the National Science Foundation, which was the major contributor to twelve of them. Few faculty had a single source of funding, but for thirty-three of them one national source contributed more than two-thirds of their total funding:

Those funded mainly by the NSF were five physicists, five chemists, and two geoscientists, ten of whom ranked at the full professor level with two as associate professors. NASA funded three physicists, two environmental scientists, one chemist, and one geoscientist, all at the full professor level. The NIH provided major funding for three biologists, two chemists, and one neuroscientist, four of whom were full professors, but one each at the associate and assistant professor levels. Thirteen faculty received their million dollar support from a combination of sources, most of which included significant grants from the National Science Foundation: NIH/NSF (3), NSF/private foundations (2), ONR/NSF (2), NASA/NSF/DOE (2), ONR/AFOSR (1), WHO/NIH/-private foundations (1).

For the fifty-one faculty with at least a million dollars in external grant support, their high grant totals was not matched on average by their publication productivity. These faculty averaged 1.5 publications per faculty member per year, the vast majority (97 %) of which were research publications rather than textbooks, encyclopedia entries, or publications in education. For all 2980 faculty in the survey, whose average grant support was only a quarter of those to whom they are compared, the average number of peer-reviewed publications was 0.6 of which 90 % were research publications. For the million-dollar faculty only 21 % of their publications included undergraduate coauthors compared to 26 % for all faculty. Only 60 % of the fifty-one faculty had ten or more publications during the decade, and only half had any publication with an undergraduate coauthor. However, six of the fifty-one faculty had ten or more publications with undergraduate coauthors, and this small faculty subset accounted for more than sixty percent of the student coauthored papers.

For faculty whose principal source of funding was the NSF, excluding those funded primarily for education-related activities, publication productivity for the decade was nineteen per faculty member, ranging from 7 to 38, or nearly two per person per year. Faculty funded from NASA averaged 1.1 publication per faculty member per year, and the range was 0.4 to 2.3. From the NIH there was a bimodal distribution with two faculty reporting 23 and 39 publications, respectively, while four others had between 0 and 2 peer-reviewed publications during the decade.

Five faculty, three from biology and two from the geosciences, received their funding from state or local sources, mainly for surveys and environmental studies that were evaluated in the study as research. Several of these were a consequence of the existence of an institute or center at the institution with significant infrastructure devoted to such activities. Publications from two of the five, virtually all identified as research publications, numbered 40 and 60, respectively for the period 1990-2000, and one had twelve undergraduate student co-authors. The remaining three faculty had a total of fourteen publications during the decade, and only one reported an undergraduate coauthor.

Five faculty, three as full professors and two as associate professors, received the majority of their funding for education and outreach, mainly from the NSF. A total of nearly $9m was reported by these faculty for the decade of the 1990s. Three faculty are in physics departments, with one each in chemistry and geoscience. These faculty reported eleven peer-reviewed publications for the decade, and none of them included undergraduate co-authors.

Well, if million-dollar faculty do not provide a clear and consistent pattern of activity, perhaps thirty-publication faculty could. We have surveyed the publication records of the nearly 3000 faculty respondents for this information. Not included in these totals are "proceedings of a conference" or "transactions of a state academy of science" or newsletters, papers in press, or those outside of the time window of the survey. The criteria for inclusion would be those commonly agreed to be professional contributions that received external oversight. To have thirty peer-reviewed publications is five times above average, and there are only 38 such individuals among the 2980 faculty who reported (1.2 %). They represent 26 institutions, 17 that are private and 9 public. Ten of the 38 faculty are in Ph.D.-granting departments. Relevant characteristics of rank, gender, and discipline are given in Table 2.

Table 2. Rank, gender, and discipline characteristics of the 38 thirty-publication faculty.

The average number of publications per faculty member for the 1990s was 42, ranging from 30 to 78, and 98 % of them were research publications; there was no difference here between faculty in Ph.D.-granting and non-Ph.D.-granting departments. However, only 19 % of the reported publications included undergraduate co-authors. Fourteen of the thirty-eight faculty with thirty or more publications had at least ten of their publications with undergraduate students, and an equal number of these faculty had zero publications with undergraduates, only four of whom were in Ph.D.-granting departments. There is an obvious disparity here with equal numbers of faculty involved and uninvolved in research with undergraduate students.

External funding for the thirty-publication faculty averaged $830,811, with a range from no external funding to $2.8 million. The average of external funding for only those faculty in Ph.D.-granting departments was $888,750, not significantly different from the composite group.

Moving now to those faculty who had $500,000 in external grant support over the ten-year period and ten or more peer-reviewed publications, there were 98 identified, mainly in biology departments (29), followed by chemists (27), physicists (24), then geoscientists (6), astronomers (5), neuroscientists (4), and environmental scientists (3). Of the 98, nineteen were women, and twenty were at the assistant professor or associate professor levels. These faculty published, on average, 2.3 peer-reviewed papers per faculty member per year - significantly greater than the corresponding average (1.5) for the "million-dollar faculty." The 98 faculty included undergraduate students on 29% of their publications, which is a percentage that is above the average for all faculty surveyed. Still, among the 98 faculty, the publications of 33 of them did not include any undergraduate co-author, suggesting once again the difficulties in involving undergraduate students in research at predominantly undergraduate institutions. Those who had half or more of their publications with undergraduate coauthors numbered 23.

The conclusions that might be drawn from this analysis are not what might have been first envisioned - that those with the most external grant dollars or those with the greatest numbers of publications had a distinct advantage in drawing students into science at predominantly undergraduate institutions. In fact, few among the "million-dollar faculty" do so. In fact, it's a rare individual who chooses conduct a significant research program that involves undergraduate students in meaningful ways.