A Data-Based Assessment of Research-Doctorate Programs in the United States (with CD) (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7 The Data and Principal Findings All doctoral programs have similarities. Typically, they admit students, maintain a curriculum of study, adhere to a “certification” requirement (preliminary or comprehensive examinations), and require completion of a piece of original work that demonstrates the ability of their students to conduct research that advances the state of knowledge. But that is where the similarities end. Depending on the discipline, a student may spend years perusing archival materials, or conducting fieldwork, or working closely on a problem as part of a research team based in a laboratory. Thus each discipline has its own ways of educating doctoral students. It makes sense, then, to examine doctoral education in each broad field or discipline separately. It does not make sense to compare time to degree in anthropology, with its years of fieldwork and close observation, with that for biochemistry, where a student typically works in a laboratory on a problem until it is solved. Within broad fields, however, many programs may exhibit similar characteristics and faculty preferences. Thus because this study covers large numbers of programs, the committee chose to summarize the data for broad fields rather than for individual disciplines. The committee also found similarities in particular characteristics among programs in a discipline. For example, larger programs in a discipline may have more resources for research, but provide a more impersonal environment for students than the smaller ones. Across disciplines, there may be fewer resources available for student support at public universities than at some private ones and, as a result, students in programs in these institutions may take longer to complete their studies, even in the same field. This chapter examines the program data in each broad field in an attempt to discover commonalities and differences. The focus here is on the variables that can be affected by administrative decisions or changes in program practices. To convey a sense of how the doctoral education enterprise has changed since the 1995 study (for which data were collected in 1993), this chapter presents data for programs in four broad fields1 that were included in both the 1995 study and the current study (designated here as “common programs”). It then moves to a discussion of the 2006 data on characteristics of all programs in the study grouped by broad field and classified by the following broad groupings of variables: research and faculty productivity, student support and outcomes, and diversity. The chapter concludes with a description of findings from the faculty and student questionnaires. 1 Agricultural sciences were not included in the 1995 study. The biological and health sciences field definitions changed too much to be strictly comparable. In the other broad fields, only the disciplines and programs included in both studies are included. 81 

82 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. CHANGES IN DOCTORAL PROGRAMS BETWEEN 1993 AND 20062 As described in Chapter 3 of this report, the changes in the eligibility criteria and definition of doctoral faculty render comparisons between the two studies inexact. It is possible, however, to match programs that were in both studies and see how the characteristics that were measured in both have changed. Table 7-1 displays these results for engineering, the physical and mathematical sciences, the social and behavioral sciences, and the humanities. TABLE 7-1 Weighted Measures for Faculty and Students, 1993 and 2006 Average Average Assistant Assistant 1993 and 2006 Faculty per Faculty per Professors, Professors, 1993 2006 Common Program, Program, 1993 (% of 2006 (% of Broad Field Programs Programs Programs 1993 2006 total facultya) total facultya) Engineering 588 747 457 21 40 20.3 20.0 Physical and mathematical sciences 786 876 552 26 37 17.8 20.1 Social and behavioral sciences 630 759 493 24 33 19.9 23.2 Humanities 656 716 498 23 31 17.2 19.6 a Average over all fields in the broad field. The number of programs grew in all fields and in all disciplines common to the two studies.3 Among the comparable broad fields, the greatest growth in number of programs was in engineering, with bioengineering a relatively new field in 1993, leading the way. The only field in which the number of programs declined was aerospace engineering. The social and behavioral sciences were next in growth of programs. Geography, a field revolutionized by the availability of satellite information, experienced the greatest percentage of growth in programs, and psychology experienced the greatest growth in number of programs. Although the humanities as a whole saw an increase of 60 programs, declines in the number of programs were most prevalent in the humanities, especially in English and the foreign language fields. The exception was Spanish language and literature. The number of programs in history grew. The physical and mathematical science expanded by 90 programs, and the greatest increase was in the earth sciences, which added 40 programs. As for the programs common to both studies, the average number of faculty in the common programs increased in all broad fields, but by far the highest growth over the period was in engineering (82 percent). This growth was experienced to some extent by all the fields, but by biomedical engineering and chemical engineering in particular. The growth in interest and funding in bioscience has fueled much of this change. Faculty per program in all other broad fields grew by more than one-third. The only field in which average faculty declined was music (–13 percent).4 Mathematics appears to have been a slow-growing field (9 percent), but this 2 A note on dates. The 1995 study collected data from the 1993 academic year. This study used a survey administered in 2007 to collect data from the 2006 academic year. Typically, the data in tables are identified by the academic year in which they were counted. 3 The committee does not know how much of this growth stems from the increased participation in the study between 1993 and 2006 and how much stems from the establishment of new programs. 4 This decline may be attributable to the greater emphasis on excluding performance faculty in the 2007 survey. 

THE DATA AND PRINCIPAL FINDINGS 83 finding may be stem from the committee’s decision in the 2007 survey to treat applied mathematics as a separate field. Table 7-2 summarizes these data for common programs. TABLE 7-2 Changes in Ph.D.’s, Enrollment, and Gender Composition, Common Programs, 1993 and 2006 Female Female Average Average Average Total Average Total Students, Students, Ph.D.’s, Ph.D.’s, Enrollment per Enrollment per 1993 2006 Broad Field 1987–1991 2002–2006 Program, 1993 Program, 2006 (average %) (average %) Engineering 7.4 8.6 64.2 68.2 14 23 Physical and mathematical sciences 7.9 8.2 67.4 74.9 24 30 Social and behavioral sciences 6.1 7.1 59.0 61.5 42 53 Humanities 4.8 5.8 59.0 53.0 50 50 The size of programs, as measured by Ph.D. production, generally displays a skewed distribution. A substantial fraction of Ph.D.’s is produced by a much smaller fraction of programs. Consequently, although the average sizes of programs as measured by either Ph.D.s or enrollments are interesting, their significance for changes in individual program size is not always clear. The changes in program size as measured by enrollment and by faculty are shown in Tables 7-3 and 7-4. The changes in average measures of size for the common programs—Ph.D.’s per program and enrollment—are much smaller in magnitude than changes in the average number of faculty. The growth in the average number of Ph.D.’s per program in all broad fields was generally modest, less than 15 percent over 13 years for all the broad fields except the humanities. Notable field exceptions were music (–8 percent), earth sciences (–34 percent), physics (–6 percent), and linguistics (–9 percent). In all other fields the average number of Ph.D.’s per program increased. Such increases can be achieved in two ways: programs grow in size, or programs shorten the time to degree, thereby producing more Ph.D.’s in the same time period. Although good data are not available from the earlier study, there is no evidence that the time to degree has diminished significantly. It seems likely, then, that the existing programs increased enrollments. A more complete analysis of the faculty-to-student ratios in the institutions that produce most of the Ph.D.’s would be required to ascertain whether there has been much of a change for most students. It is possible that most of the growth in faculty has occurred in institutions that produce relatively few Ph.D.’s.5 5 The difference in how the two studies defined doctoral faculty was discussed earlier. If anything, the definition in the 2007 survey was more restrictive than the definition used in 1993so we can be sure growth occurred but cannot be certain of its magnitude. 

84 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. TABLE 7-3 Average of Total Faculty, All Common Programs, 1993 and 2006, Number of Common Programs, 1995– Average Total Faculty Average Total Faculty per Broad Field 2006 per Program, 1993 Program, 2006 Engineering 457 21.88 39.75 Physical and mathematical sciences 552 26.40 37.22 Social and behavioral sciences 493 23.64 32.76 Humanities 498 23.35 30.97 The growth in the common programs is mirrored by the growth of doctoral recipients nationwide (see Table 7-4). The greatest percentage growth was in engineering. TABLE 7-4 Percentage Change in Number of Doctoral Recipients, Common Programs, 1993 and 2006 1993 2006 Difference Percentage change Engineering 5,061 6,707 1,646 33 Physical and mathematical sciences 6,425 7,283 858 13 Social and behavioral sciences 7,538 8,326 788 10 Humanities 3,174 3,821 647 20 Source: NSF Survey of Doctoral Recipients, 2006 and 1993. Growth in Postdoctoral Scholars One of the major changes in the academic research enterprise since the last study is the increase in the number of postdoctoral scholars, primarily in the sciences.6 Data on postdocs were not collected in the 1995 study; however, it is now clear that, especially in the biological sciences, these young scholars play a major role both in research and in the mentoring of Ph.D. students. In many respects they share some of the roles of both advanced graduate students and faculty. Thus, because some faculty time is spent on the education of postdoctoral scholars and part of postdoctoral scholars’ time is spent on the education of graduate students, any interpretation of the NRC data on student-to-faculty ratios, for example, should consider whether the increased number of faculty directly affects Ph.D. student mentoring. This study addresses doctoral education specifically, so it does not address the participation of postdoctoral scholars. However, the presence of substantial numbers of postdoctoral scholars changes the context of graduate education, especially the Ph.D. research experience. The number and therefore the impact of postdoctoral scholars differ significantly across disciplines and size of programs—being more prominent in the sciences and much less so in the humanities. Table 7-6 shows the number of postdocs by broad field in 2006. The greatest impact of postdocs is clearly in the biological and health sciences. 6 This growth, through 1997, was documented in National Research Council, Enhancing the Postdoctoral Experience for Scientists and Engineers: A Guide for Postdoctoral Scholars, Advisers, Institutions, Funding Organizations, and Disciplinary Societies, Committee on Science, Engineering, and Public Policy (COSEPUP). (Washington, D.C.: National Academies Press, 2000), 5. 

THE DATA AND PRINCIPAL FINDINGS 85 TABLE 7-5 Number of Postdocs by Broad Field, 2006 Broad Field Number of Postdocs Agricultural sciences 2,429 Biological and health sciences 24,084 Physical and mathematical sciences 9,952 Engineering 5,387 Social and behavioral sciences 1,430 Humanities 233 Changes in the Diversity of Programs For all the common programs, with the exception of classics (–5.2 percent) and linguistics (–2.5 percent), in all broad fields, the percentage of enrollment for women increased. For the broad fields the absolute numbers of women enrolled also grew. Increases of greater than 10 percentage points were found in several engineering fields (biomedical, chemical, civil, and mechanical), several fields in the physical sciences (astrophysics, earth sciences, oceanography, and statistics), and one field in the social sciences (economics). All these fields had relatively low levels of female enrollment in 1993. Data on racial and ethnic diversity were not collected in 1993, but the NSF data shown in Figure 7-1 reveal a considerable increase in minority Ph.D.’s across the board. However, in some fields, especially engineering, the social and behavioral sciences, and the physical and mathematical sciences, the numbers of non-underrepresented minorities (Asian Americans and whites) have been declining. 

86 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. FIGURE 7-1 Minority and Nonminority Ph.D.’s, 1993 and 2006 Note: Non-URM = non-underrepresented minorities; URM = underrepresented minorities. Source: National Science Foundation, Division of Science Resources Statistics. In summary, the 13 years from 1993 to 2006 saw an increase in the number of doctoral programs in the common broad fields and disciplines, growth in the numbers of faculty and students per program, expanded production of Ph.D.’s per program, and an increase in the gender and ethnic diversity of programs. These quantitative changes were accompanied by changes in the average faculty-to-student ratios, which increased significantly over this period. This chapter now turns to a more detailed description of doctoral education in 2005–2006. 

THE DATA AND PRINCIPAL FINDINGS 87 U.S. DOCTORAL PROGRAMS IN 2006: A DESCRIPTION Although this study does not include all doctoral programs or all fields, it does cover the vast majority of research doctorate programs in the United States.7 This section begins by describing the characteristics of the programs in the study. Of particular interest is the size of programs in these fields and the type of control (public or private). The section then moves to comparing and contrasting the fields along these dimensions. Coverage The programs with rankings in this study account for approximately 90 percent of the doctorates in the fields included in the study taxonomy in 2006. A comparison with NSF’s Doctorate Record File (DRF) by broad field is shown in Table 7-6. In view of these high rates of coverage, the generalizations drawn from the study sample can, in all likelihood, be applied to U. S. doctoral education as a whole. TABLE 7-6 Number of Ph.D.'s in 2006 NRC Study Compared with Ph.D.'s in NSF Doctorate Record File Broad Field DRF 2006 2006 NRC Study % in Study Agricultural sciences 1,470 1,147 78 Biological and health sciencesa 5,737 5,543 97 Physical and mathematical sciences 7,283 7,092 97 Engineering 6,707 6,716 100 Social and behavioral sciencesb 7,783 5,997 77 Humanities 4,624 3,754 81 Total 33,604 30,249 90 Total minus psychology 30,346 28,283 93 Note: The DRF responses are from a questionnaire administered to students at the time of graduation. The NRC data are from responses to the program questionnaire. The timing of the answers to the two questionnaires may account for discrepancies. The NRC questionnaire was answered in 2007, for 2005–2006 graduates, whereas the NSF questionnaire covered Ph.D.’s awarded in 2005–2006. a Integrated biological science, which appears in the NRC biological and health sciences taxonomy (Appendix B), is not a DRF field. The 867 Ph.D.’s from these programs were probably spread across the NSF biological sciences taxonomy. b Universities were asked to exclude programs in clinical and counseling psychology. The exclusion of these programs accounts for the low rate of coverage in the social and behavioral sciences. The coverage rate for all social science fields, excluding psychology, is 89 percent. Excluded from the NRC study are Ph.D.’s in professional fields as well as small fields and programs. The professional fields are excluded for historical reasons. The study originally included programs in the liberal arts and sciences. The committee then expanded this coverage to include fields outside the arts and sciences and Ph.D. programs in schools of medicine that award a research Ph.D., but professional fields with a substantial practice component were still 7 From this point on, to the term doctoral programs refers to the fields and programs included in the ranking study. This group does not include small doctoral programs that produced less than one Ph.D. per year during 2002–2006, nor does it include fields that were not ranked in the study. It also does not include the programs and universities that, for a variety of reasons, did not participate in the study. All told, the study covers the programs that produced about 90 percent of Ph.D.’s in the ranked fields during that period. 

88 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. excluded from the study on the grounds that publications in scholarly journals are not an adequate metric of the quality of these programs.8 Small fields and small programs in the humanities were excluded because they provide too few observations for reliable statistical analysis. Size and Control For all fields, most doctoral programs, most enrolled doctoral students, and most Ph.D.’s are found in public universities. Programs in public universities are typically larger than those in private universities, and there are far more of them. Seventy-one percent of the programs ranked in the NRC study are in public universities. The proportion of programs in the universities with the largest programs is similar (70 percent). Among the 37 universities that produced 50 percent of Ph.D.’s from 2002 to 2006, 70 percent were public (see Table 7-7). Although public universities rely increasingly on nonpublic sources of funding, cutbacks in public funding for universities has a powerful effect on doctoral education simply because of how many large Ph.D. programs exist in public universities. These cutbacks will, of course, affect public higher education in general. 8 A different view can be found in M. Goulden, et al. 

THE DATA AND PRINCIPAL FINDINGS 89 TABLE 7-7 Institutions with 50 Percent of Ph.D.’s in Ranked Programs, by Control (Public or Private), 2002–2006 (average number of Ph.D.’s) Institution Private Public Cumulative Total University of California, Berkeley 623 623 University of Texas at Austin 533 1,158 University of Michigan–Ann Arbor 523 1,681 University of California, Los Angeles 518 2,198 University of Wisconsin–Madison 513 2,712 Stanford University 505 3,217 University of Minnesota–Twin Cities 455 3,672 University of Illinois at Urbana-Champaign 455 4,128 Massachusetts Institute of Technology 448 4,576 Ohio State University Main Campus 439 5,015 Harvard University 436 5,450 Pennsylvania State University 431 5,882 University of Washington 429 6,311 University of Florida 412 6,723 Purdue University Main Campus 389 7,112 University of Maryland, College Park 378 7,490 Cornell University 373 7,863 Texas A&M University 347 8,210 Columbia University in the City of New York 344 8,554 University of California, Davis 341 8,895 Michigan State University 325 9,220 University of Pennsylvania 312 9,533 Johns Hopkins University 312 9,845 Georgia Institute of Technology 297 10,142 University of North Carolina at Chapel Hill 294 10,435 Yale University 293 10,729 University of Chicago 292 11,021 University of Southern California 287 11,308 University of California, San Diego 266 11,573 North Carolina State University 264 11,837 Princeton University 260 12,097 University of Arizona 259 12,356 University of Colorado at Boulder 257 12,613 City University of New York Graduate Center 255 12,868 Rutgers, the State University of New Jersey, New Brunswick Campus 254 13,122 University of Georgia 244 13,366 Northwestern University 244 13,610 Grand total of average number of Ph.D.’s—all universities 8,236 19,382 27,618 

90 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. Importance of Program Size After release of the 1995 study, some readers, and the report itself, commented on how important program size was to the ranking of a program.9 As mentioned in Chapters 5 and 6, the coefficient on size is especially large in most fields in the regression-based ranking (R ranking), and generally far less prominent in the survey-based ranking (S ranking). One example of an analytic use of the data is an investigation of the characteristics of programs as they relate to size as measured by number of Ph.D.’s. The committee divided programs in each field into size quartiles,10 grouped the fields into broad ones, and investigated whether the quartile with the largest programs also ranked high on the 20 characteristics. It found that this quartile has the highest levels of publications per faculty member. Citations also follow this pattern, although the dominance of the largest quartile programs is not significant for engineering. In line with the findings for the other research variables, the largest programs also have a significantly higher average number of awards per faculty member. The greater research productivity in the largest quartile occurs even though our measures of research activity are on a per capita basis. These values are shown in Table 7-8. TABLE 7-8 Research Measures and Program Size Row Labels Number of Average of Average Average of Average of Average of Average of Programs Number of Average Percent of Awards per Tenured Faculty Publications (2000- Citations per Faculty with Allocated as a Percent of 2006) per Allocated Publication Grants, Faculty Total Faculty, Faculty, 2006 Member, 2006 2006 Agricultural 312 1.32 1.66 81.8% 0.35 71.3% Sciences Largest Quartile 33 1.61a 2.10a 85.0%a 0.50a 68.7% by Program Size Biological and 1168 1.54 3.48 82.3% 1.10 66.0% Health Sciences Largest Quartile 101 1.95a 4.46a 87.8%a 2.43a 64.3% by Program Size Engineering 759 1.62 1.45 80.9% 0.61 72.2% a a a a Largest Quartile 39 2.42 1.54 86.4% 1.21 76.0%a by Program Size Physical and 911 2.05 2.05 75.2% 1.04 72.8% Mathematical Sciences Largest Quartile 54 3.98a 3.00a 86.6%a 4.54a 73.8% by Program Size Social and 924 0.52 1.40 44.3% 0.62 72.9% Behavioral Sciences Largest Quartile 83 0.81a 2.10a 54.7%a 1.20a 71.7% by Program Size Humanities 764 11.62 n.a 14.5% 2.00 76.2% a a a Largest Quartile 55 12.86 n.a 16.0% 3.42 77.8%a by Program Size Note: “n.a.” indicates not available a The fourth quartile is different from the average at the p= 0.05 level. b Publications in the humanities are measured by books and articles with books given a weight of 5 and articles a weight of 1. Humanities publications are not comparable to publications in other fields, which are measured by articles. 9 This discussion appears on pages 22–23 of the 1995 NRC study. 10 The programs were arrayed from highest to lowest by the number of Ph.D.s produced. The “Largest quartile” was defined as those programs that produced 25 percent of Ph.D.s at the top of this array. The number of programs in the Largest quartile is smaller than the number of programs divided by four because some of the programs in this Largest quartile are quite large. 

THE DATA AND PRINCIPAL FINDINGS 91 Student Variables and Program Size Findings for the student variables are shown in Table 7-9. Measures of interest to students include the annual average number of Ph.D.’s, GRE scores, completion rates, time to degree, employment destination upon graduation, and whether the program keeps track of its graduates after graduation. In the larger science and engineering programs students have higher GRE quantitative scores. The average humanities programs have lower GRE verbal scores and completion percentage within eight years than the programs in the largest quartile. Placement of Ph.D.’s in academic positions does not differ significantly by size quartile of programs, except in engineering where the placement in the largest quartile is lower.11 This percentage is highest for the biological and health sciences, and this is expected, since postdoctoral appointments are counted in this measure. The next highest percentage is in the humanities. Time to degree is significantly longer in the larger programs, except for the social and behavioral sciences, although completion rates do not seem to vary significantly with size. Finally, the largest physical science, biological and health sciences and engineering fields have a higher percentage of programs that collect placement data for their students. Among broad fields overall, GRE-Quantitative Reasoning scores are higher, as expected, in engineering and the physical and mathematical sciences than in other fields. The percentage of students with first-year support is greater than 80 percent in all fields and is over 90 percent in the physical and mathematical sciences. The data on completion rates and average time to degree raise important questions about the proportion of students entering doctoral programs who actually complete a degree. The completion rate in six years ranges from nearly 60 percent (agricultural sciences) to 37 percent (social and behavioral sciences and yet the median time to degree, only for those who complete their degrees, has a narrower range (4.8–6.2 years). In the humanities, where 43 percent of enrolled students complete their degree in eight years and the median time to degree is 7.1 years, it can be inferred that a very high proportion of humanities students who enter doctoral programs never complete a Ph.D. degree. The factors that influence attrition rates and student success in research doctorate programs are certainly worthy of ongoing attention. 11 Academic placement includes postdoctoral study in academic institutions. 

92 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. TABLE 7-9 Student Characteristics by Broad Field Average and for the Largest Quartile Row Labels Number of Average of Average Average of Average of Average Average of Programs Average GRE of Avg. Median of Collects Scores, Percent Completion Time to Percent Data About of First Percentage: Degree with Post- Year (Full- and Academic Graduation Students Part-Time Plans Employment with Full Graduates), Financial Support, Agricultural Sciences 312 656 88.0% 58.0% 4.81 56.0% 47.0% b b Largest Quartile by Program Size 33 680 83.0% 58.0% 5.12 54.0% 41.0% Biological and Health Sciences 1168 686 91.0% 49.0% 5.50 69.0% 55.0% b b Largest Quartile by Program Size 101 708 92.0% 50.0% 5.63 69.0% 62.0%b Engineering 759 760 83.0% 51.0% 4.88 35.0% 25.0% Largest Quartile by Program Size 39 789b 77.0% 55.0%b 5.13b 27.0% 44.0%b Physical and Mathematical 911 745 92.0% 43.0% 5.47 56.0% 56.0% Sciences Largest Quartile by Program Size 54 755b 98.0%b 49.0%b 5.51 54.0% 59.0% Social and Behavioral Sciences 924 662 81.0% 37.0% 6.16 57.0% 67.0% Largest Quartile by Program Size 83 700b 84.0%b 45.0%b 6.07 55.0% 61.0% Humanities 764 610 83.0% 43.0% 7.11 59.0% 67.0% Largest Quartile by Program Size 55 653b 77.0% 46.0%b 7.38b 59.0% 66.0% a For the humanities, the time to completion is eight years and the GRE score is for GRE-V. b Indicates that the largest quartile value differs from the average at the p = 0.05 level. 

THE DATA AND PRINCIPAL FINDINGS 93 Diversity Average measures of various kinds of diversity are shown in Table 7-10. It also shows the results of tests to determine whether the largest quartile is different from the average. TABLE 7-10 Diversity Measures Average of Non-Asian Average of Minority Non-Asian Average of Faculty as a Minority Female Percent of Students as a Faculty as a Average of Average of Total Core Percent of Percent of Female International and New Total Total Core Students as a Students as a Domestic Domestic and New Percent of Percent of Number of Faculty, Students, Faculty, Total Total Row Labels Programs 2006 Fall 2005 2006 Students, Students, Agricultural Sciences 312 4.3% 9.1% 9.3% 47.8% 42.2% Largest Quartile by Program Size 33 3.6% 8.5% 3.0% 49.5% 34.3% Biological and Health Sciences 1168 3.9% 11.5% 8.9% 55.0% 29.1% Largest Quartile by Program Size 101 3.9% 12.7%a 7.9% 54.7% 24.0% Engineering 759 5.2% 12.6% 0.5% 23.9% 61.0% Largest Quartile by Program Size 39 4.4% 11.5% 0.0% 18.4% 58.6% Physical and Mathematical Sciences 911 3.3% 8.4% 0.5% 32.2% 45.3% Largest Quartile by Program Size 54 2.7% 7.4% 0.0% 30.3% 35.8% Social and Behavioral Sciences 924 7.6% 13.3% 15.7% 55.0% 27.2% a Largest Quartile by Program Size 83 7.2% 13.5% 10.8% 58.3% 23.9% Humanities 764 10.8% 13.0% 23.3% 53.1% 17.5% Largest Quartile by Program Size 55 9.7% 12.6% 10.9% 53.3% 13.5% a The largest quartile value is different from the average at the p = 0.05 level. Increasing gender and racial and ethnic diversity has been a goal of the graduate community for many years. Although substantial progress has been made, that goal is far from achieved. The percentage of underrepresented minorities by broad field for students and faculty is shown in Figure 7-2. With the exception of the humanities, in no field is more than 10 percent of the faculty from underrepresented minorities, and the sciences are at or less than 5 percent. Because larger percentages of doctoral students are from underrepresented minorities, it is likely there will be larger pools of Ph.D.’s from which to draw from in the future. However, the underrepresented minority enrollments in the agricultural and physical and mathematical sciences and engineering are still less than 10 percent. The percentage of minority students and faculty in all broad fields is less than 15 percent. Nevertheless, in some individual fields, listed in Table 7-11, more than 10 percent of enrollments are from underrepresented minority groups. 

94 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. FIGURE 7-2 Percentage of Underrepresented Minority Faculty and Students by broad field, 2006. 

THE DATA AND PRINCIPAL FINDINGS 95 TABLE 7-11 Fields with More than 10 Percent of Enrolled Students from Underrepresented Minority (URM) Groups Percentage Broad Field Field URM Agricultural sciences Food science 13.2 Nutrition 15.0 Agricultural sciences total 9.1 Biological and health sciences Biochemistry, biophysics, and structural biology 10.4 Cell and developmental biology 11.0 Genetics and genomics 10.3 Immunology and infectious disease 13.3 Microbiology 12.8 Neuroscience and neurobiology 11.1 Nursing 12.4 Pharmacology, toxicology, and environmental health 13.9 Physiology 14.3 Public health 14.9 Biological and health sciences total 11.5 Physical and mathematical sciences Applied mathematics 11.5 Chemistry 11.0 Physical and mathematical sciences total 8.4 Engineering Biomedical engineering and bioengineering 11.3 Chemical engineering 11.0 Civil and environmental engineering 12.3 Electrical and computer engineering 12.6 Engineering science and materials (not elsewhere classified) 20.7 Materials science and engineering 12.1 Mechanical engineering 14.6 Operations research, systems engineering, and industrial engineering 17.0 Engineering total 12.6 Social and behavioral sciences Anthropology 14.7 Communication 13.3 Linguistics 10.6 Political science 12.8 Psychology 13.0 Public affairs, public policy, and public administration 20.0 Sociology 19.0 Social and behavioral sciences total 13.3 Humanities American studies 28.0 Comparative literature 14.0 English language and literature 10.0 History 12.0 Religion 10.0 Spanish and Portuguese language and literature 46.0 Theater and performance studies 13.0 Humanities total 13.0 Some of these fields are training students who will work with underrepresented minority populations or specialize in studies related to the history and culture of underrepresented minorities, but all have focused on increasing minority participation and have, to some extent, succeeded. 

96 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. The increased participation of women, as both faculty and students, is even more of a success story (see Figure 7-3). FIGURE 7-3 Percentage of Faculty and Students Female by broad field, 2006. Enrollments in a few broad fields (humanities, social and behavioral sciences, and biological and health sciences) are more than 50 percent female, but the representation of women in the faculty has yet to reach even 40 percent. In none of the broad fields in science or engineering is more than 30 percent of the faculty female. The disciplines in the science, technology, engineering, and mathematics (STEM) fields in which more than 15 percent of the doctoral faculty is female are shown in Table 7-12. TABLE 7-12 Science and Engineering Fields with More than 15 Percent of Doctoral Faculty Female Total Broad Field Field % Agricultural sciences Entomology 15.4 Food science 27.1 Nutrition 50.6 Plant sciences 20.4 Agricultural sciences total 23.2 Biological and health sciences All fields Biological and health sciences Total 29.2 Engineering Biomedical engineering and bioengineering 15.2 Operations research, systems engineering, and industrial engineering 15.8 Engineering total 11.4 Physical and mathematical sciences Earth sciences 15.8 Oceanography, atmospheric sciences, and meteorology 16.1 Statistics and probability 19.1 Physical and mathematical sciences total 13.3 Social and behavioral sciences All programs Social and behavioral sciences total 33.5 Humanities All programs Humanities total 37.7 One other aspect of diversity is the percentage of students who are from outside the United States. The variation in percentages of enrolled international students across the broad fields is considerable: engineering as a whole, 60 percent; the humanities, slightly more than 15 

THE DATA AND PRINCIPAL FINDINGS 97 percent; the physical and mathematical sciences, 45 percent; the biological and health sciences and the social and behavioral sciences, less than 30 percent. Within the broad fields, some disciplines differ noticeably from the average. In economics, for example, 63 percent of its students are international. FINDINGS The spreadsheet online at http://www.nap.edu/rdp contain a vast amount of data that could be characterized, mined, and modeled. With the previous discussion, the committee offers only a glimpse into the descriptive richness possible from analyzing the data available for many of the programs in 59 fields12. To illustrate one possible analysis, it looked at the characteristics associated with program size. Program size is positively associated with most measures of the research productivity of doctoral programs, even when productivity is measured on a per capita basis.13 As for student characteristics, the larger programs are also more likely to have higher average GRE scores, except in the humanities. There is a size difference for median time to degree; students in the larger programs take about half a year longer to complete their degrees. In the physical and social sciences a significantly greater percentage of large programs collect outcomes data for their students. Interestingly, size, analyzed within broad fields, does not appear to be associated systematically with the percentage of students with support in their first year, which is high across the board, or completion rates, or the percentage of students who plan on a position in academia (including postdoctoral study) after graduation. Readers should note that the committee has been careful to discuss association, not causation. The committee also looked at racial and ethnic and gender diversity. It found that much less diversity is found in the physical and mathematical sciences than in other broad fields. Although some fields have succeeded in becoming more diverse, most programs still have percentages of underrepresented minority students that are far under 10 percent. That said, although the percentages remain very low, they are in all cases significantly higher than were indicated by the NSF data from 1993 and discussed in the previous section. Thus doctoral programs have achieved far greater diversity with gender. This completes the discussion of examples of simple analyses that can be conducted using the program data. We now discuss a few findings from data obtained from the faculty questionnaire and the student questionnaire. Data from these questionnaires will be made available with the public use data set. CHARACTERISTICS OF THE FACULTY OF THE DOCTORAL PROGRAMS This report has focused primarily on information from the doctoral program questionnaire, but two other questionnaires—those directed at faculty and at students—also provide interesting insights into doctoral education. Although in the faculty questionnaire there were variations in the response rates by field, the overall response rate was 88 percent, and so it is likely that generalizations to doctoral faculty can be made from the survey responses. This section focuses on three areas: (1) the age profile of the faculty, (2) the length of time at their current institution, and (2) their demographic composition. Age is important because younger faculty are typically very active in research, although this activity may not yet translate into large numbers of 12 Actually, 62 fields, since data, but not ranges of rankings, are shown for computer engineering, engineering science and materials, and languages societies and cultures. 13 Of course, one would expect larger programs to have greater total levels of research production. 

98 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. publications or citations. Time at the current institution is a reflection of faculty turnover—the longer faculty stay in place, the less the turnover. The importance of demographic composition was discussed earlier. Selected data on faculty are shown in Table 7-13. TABLE 7-13 Faculty Data: Selected Measures, 2006 Physical and Biological and Mathema Social and Agricultural Health tical Behavioral Variable Sciences Sciences Engineering Sciences Sciences Humanities Number of responses 5,761 28,952 15,428 21,614 18,460 14,376 Time at current institution (% of total faculty) < 8 years 22 25 24 23 26 24 8–20 years 40 44 39 37 38 39 > 20 years 38 28 33 37 33 33 Nonresponse 0 3 4 3 3 4 Employment before current employment (%) Student 18 5 21 13 27 26 Postdoc 30 42 21 36 13 6 Faculty—associate or full 12 16 15 15 21 22 professor Faculty—assistant 11 12 10 12 20 23 professor Other 20 15 24 18 13 15 Nonresponse 9 9 7 7 6 8 Age in 2007 (%) < 40 years old 6 6 14 13 14 9 40–60 years old 58 59 53 51 50 49 > 60 years old 20 19 19 24 24 27 Nonresponse 16 17 14 12 11 14 Citizenship (%) U.S. citizen or permanent 84 83 83 84 87 85 resident Temporary visa holder 2 1 4 4 3 2 Nonresponse 14 15 13 11 10 12 Race and Ethnicity (%) White 76 72 63 72 78 77 Asian 8 10 21 14 6 4 Underrepresented Minority 5 4 5 4 8 10 Nonresponse 11 13 11 10 8 9 Gender (%) Male 66 59 77 76 59 53 Female 19 25 10 12 30 34 Nonresponse 16 16 13 12 11 13 The agricultural and the biological and health sciences appear to have fewer doctoral faculty under the age of 40 than the other broad fields. This is a result of different hiring patterns in the broad fields, as evidenced by the answers to the question about previous employment. About one-third of respondents in the agricultural, physical, and biological and health sciences have one or more postdocs before becoming doctoral faculty. By contrast, in the humanities, engineering, and social and behavioral sciences more than 20 percent of the faculty came to a faculty position directly from receiving their Ph.D. Engineering, which draws many of its faculty from industry brings in almost a quarter of its faculty from “other,” which includes nonacademic employment. Movement from within academia is about 25 percent in the sciences and engineering. The pattern is different for the humanities and social sciences, where more than 40 percent of respondents were employed in academia before moving to their current employer. The humanities distinguish themselves by the age of their doctoral faculty. More than 27 percent are over the age of 60, in contrast with 20 percent or less in the agricultural sciences, engineering, 

THE DATA AND PRINCIPAL FINDINGS 99 and the biological and health sciences. As for mobility, doctoral faculty tend to stay at one institution. About three-quarters of them, in all fields, have been at their current institution for 8 years or more, and more than one-third have been in one place for more than 20 years. The one exception is the biological and health sciences. The composition of the faculty by racial and ethnic diversity and gender was discussed earlier in this chapter, and is confirmed here. Only the humanities draw more than 10 percent of its faculty from underrepresented minorities. As for gender, in the social sciences, humanities, and biological and health sciences, one-quarter or more of the doctoral faculty is female. DATA FROM STUDENT QUESTIONNAIRES The 4,838 ranked programs in the study include 236,417 students. Despite the high cost of sending questionnaires to all these students, the committee believed that the voices of the students should be heard. It surveyed students in disciplines in five of the broad fields: engineering, the physical and mathematical sciences, the biological and health sciences, the social sciences, and the humanities. The specific fields chosen were chemical engineering, physics, neuroscience, economics, and English. Each of the programs in these fields was asked for the names and e-mail addresses of their students who had been advanced to candidacy but had not yet completed their degrees. This group was chosen because the committee believed these students would have experienced many of the program practices and would have formed views of their doctoral programs. Confidentiality concerns prevent the committee from reporting results by program for the smaller programs. However, more than 90 percent of the responding students were in programs in which more than 10 students responded and thus could be reported on a program-by-program basis. In all, the responses that were reportable at the program level represented about 64 percent of all the programs in the five fields, and for the participating programs the student response rates were high. The response rates and other details of the student survey are reported in Table 7-14. TABLE 7-14 Response Rates: Student Survey, Five Fields, 2006 Chemical Engineering Physics Neuroscience Economics English Overall Total students surveyed 2,411 5,250 1,997 2,903 3,878 16,439 Total students responding 1,820 3,596 1,562 2,067 2,544 11,589 Overall response rate (%) 75.5 68.5 78.2 71.2 65.6 70 Number of students in programs with more than 10 responses 1,538 3,322 1,373 1,829 2,354 10,416 Total programs 108 153 94 116 122 593 Programs with more than 10 responses 55 106 61 69 89 380 Percentage of responding students in programs with more than 10 responses 84.5 92.3 87.9 88.4 92.5 90 Percentage of programs with more than 10 responses 50.9 69.2 64.8 59.4 72.9 64 Student Satisfaction Table 7-15, which summarizes the results for each of the surveyed fields, reveals that most students are satisfied with their program. In all fields the percentages of programs whose students are not satisfied are less than 10 percent. On the whole, students value the intellectual 

100 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. environment provided by their programs. The main characteristic that receives a low rating from students is “quality of (program-sponsored) social interaction.” In the sciences and engineering, students report being highly satisfied with the quality of the research facilities available to them. Computing facilities are satisfactory in all fields, but students in programs in English and economics appear more critical of the research facilities and work space available to them. Investigators should look into the programs in these two fields in which students failed to say that research facilities were excellent or good. In fact, for the programs with more than 10 respondents, the questionnaire results may point to a follow-up agenda. The relatively low ratings in English and economics may indicate inadequate library facilities or inadequate support of other scholarly infrastructure. However, the survey did not collect data at this level of detail. TABLE 7-15 Student Satisfaction: Programs with More Than 10 Students Responding, 2006 Intellectual Computer Research Work Overall Quality Environment Social Interaction Resources Facilities Space Very satisfied with Rating Rating Rating Benefiting a lot the [program- computer research work from program's sponsored] social resources as facilities as space as Very Somewhat Not intellectual interaction in the excellent or excellent or excellent Field Satisfied Satisfied Satisfied environment program good good or good % % % % % % Chemical engineering 51 44 5 63 38 89 80 82 Economics 44 47 9 58 25 76 41 57 English 50 39 7 63 28 74 38 39 Neuroscience 54 41 5 67 37 89 91 86 Physics 46 47 7 60 26 85 70 76 

THE DATA AND PRINCIPAL FINDINGS 101 Student Productivity Generally, the programs in all fields seem to be performing well in encouraging students to become productive scholars. As shown in Table 7-16, well more than half the students in all fields have presented papers at conferences on campus, and a similarly high proportion has presented papers at national or regional meetings, even if only a smaller proportion found funds for their travel. A high proportion of students in the science and engineering programs also report that they have published articles in refereed journals, but less so in economics and English. In all fields the percentage of students who have published either articles or book chapters has risen since they enrolled in doctoral study. It is clear from Table 7-16 that students produce papers in refereed journals while studying in their doctoral programs. The data for individual programs (not shown) reveal that in more than 90 percent of the individual programs in all five fields at least one student had published in a refereed journal. TABLE 7-16 Student Productivity: Programs with More Than 10 Student Responses Publications Publications Before Graduate During Graduate Study Study Making Research Presentations at Securing Travel Average Making Research Meetings (Regional, Funds to Present Number Program Presentations on National, or Research at Refereed Book Refereed Book of Response Campus International) Meetings Articles Chapters Articles Chapters Programs Rate % % % % Chemical engineering 78 85 73 39 1 72 9 56 81.5 Economics 67 51 43 13 4 23 7 70 76.1 English 63 83 69 10 3 34 17 89 71.0 Neuroscience 88 86 74 48 2 69 10 63 83.7 Physics 61 76 69 37 1 69 4 106 72.2 

102 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. Advising and Academic Support Assessment of student academic progress appears to be the norm in neuroscience. In all but nine programs surveyed, more than 75 percent of the students responding indicated that their programs provided an assessment of students’ academic progress. Although not all fields reported this level of assessment, a high proportion of students in all fields indicated that they valued the assessments they did receive. A high proportion of students in virtually all programs also indicated that they received timely and helpful feedback on their dissertations. Doctoral education is characterized by the apprenticeship of students to mentors and advisers. For this reason a students’ evaluations of their relationship with the faculty is both interesting and important. Across the five fields surveyed, about 50 percent of the students in all fields reported that they had highly interactive and supportive mentors and advisers. This uniformity is striking considering that students in the sciences and engineering might be expected to have more sustained interaction with faculty in laboratory settings. Interaction with other faculty members appears very limited. This finding was also consistent across the fields surveyed (see Table 7-17 for the results). TABLE 7-17 Students: Advising and Academic Support (percent) Program Provides Assessment of Assessment of Timely Feedback Dissertation Highly Highly Mentor in Academic Academic Progress on Dissertation Feedback is Supportive Supportive Other Field Program Progress Is Helpful Research Helpful Adviser Faculty Chemical engineering 80 50 90 90 90 50 20 Economics 80 60 80 90 90 50 20 English 90 50 80 80 80 50 20 Neuroscience 92 87 90 87 86 51 24 Physics 83 52 81 81 79 50 19 Doctoral students enter programs with career goals in mind, but in most fields that were queried these goals undergo modification during the course of graduate study. Doctoral students learn what kind of scholarly work they enjoy, and they also learn how good they are at it. With the exception of chemical engineering students, who were most likely to select careers in the private sector, most students anticipated a career in the education sector as they began doctoral study. But this interest tended to wane during graduate school, as students appeared to explore options in government or the private sector. Advisers and mentors are students’ principal sources of career advice. Only students in chemical engineering reported making much use of university career centers. Students generally indicated that their advisers supported their career plans. 

THE DATA AND PRINCIPAL FINDINGS 103 Career Goals Yet another measure is what students want to do when they graduate (Table 7-18). TABLE 7-18 Student Career Objectives at Program Entry and at Time of Response (percent) At Program Entry At Time of Response Research and Management and Research and Management and Development Teaching Administration Development Teaching Administration Chemical engineering 76.5 11.6 6.2 74.9 8.0 9.1 Economics 68.1 20.4 3.4 64.6 16.6 5.4 English 34.2 58.2 1.0 34.3 52.0 3.9 Neuroscience 80.5 10.8 1.2 67.5 12.4 4.3 Physics 82.6 14.1 1.0 76.2 13.9 2.8 Note: Omitted choices are “professional services” and “other,” and so the percentage across a row for a particular point in time does not add to 100 percent Overall, only 38.2 percent of programs showed an increase in student interest in research and development. Eighteen percent of programs saw an increase in students wanting to go into teaching, and 47.1 percent of programs saw an increase in student interest in management and administration. These findings suggest that as students learn what is actually involved in research and teaching, they become more interested in other, untried undertakings. To summarize, the student questionnaire reveals that students are generally pleased with their doctoral programs and that the programs are successful at improving student research productivity, but that by the time students are working at an advanced level at least some of them have shifted their career objectives away from research. This effect is not large, but it may explain in part the lower completion rates observed by the committee. Although it is likely that the decline in interest in research careers is the result of students learning more about what such a career entails, programs may wish to look at their individual results to determine the steps that might be taken to address this falloff in student interest in research. SUMMARY This chapter has provided a glimpse into the large amount of data about doctoral programs available in the study’s online database. By matching programs, the committee was able to compare the 2006 and the 1993 data and see that the number of enrollments and the number of Ph.D.’s produced by the common doctoral programs have grown in most fields, with the exception of the humanities. Using NSF data, it also saw that the gender and racial and ethnic diversity of these programs has increased as well since the last study. The committee used the most current data to conduct an illustrative analysis in which it looked at program characteristics and size of program as measured by Ph.D. production. Although some smaller programs certainly have high research activity, generally the larger programs are associated with higher values for characteristics related to research. This association does not carry over to student support and outcome variables. The most consistent finding is that the larger programs have somewhat longer times to the completion of a degree. Size is not consistently related to differences in diversity. 

104 A DATA-BASED ASSESSMENT OF RESEARCH-DOCTORATE PROGRAMS IN THE U.S. The great deal of data to be made available from the faculty and student questionnaires can be used to explore relationships among program characteristics and the characteristics of faculty and of students in the five fields studied.