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.
8 Human Resources BACCALAUREATE AND POST-BACCALAUREATE DEGREE PRODUCTION Compared to other academic disciplines, academic CS&E is new and is growing rapidly. The electronic stored-program computer is some 50 years old, and around this invention has grown a thriving and productive intellectual discipline. In this time, over 150 Ph.D.- granting CS&E departments have been established, along with per- haps 850 other CS&E programs nationally. These institutions have produced thousands of Ph.D.s and hundreds of thousands of gradu- ates with bachelor's degrees. In addition, many other institutions have developed programs in information sciences, library sciences, management information systems, and so on; in many cases, degrees awarded by these latter institutions include at least some of the CS&E material that other institutions might include as part of a CS&E un- dergraduate degree, although they tend not to cover such material as broadly or as deeply. This diversity in computer-related degree programs makes it dif- ficult to obtain detailed insight into degree production. In gathering data sources for this report, the committee considered whether or not to include in its definition of CS&E degree recipients those who had received degrees in "information sciences" or "information systems," since many sources group these categories together. Because it was 239
240 COMPUTING THE FUTURE most concerned with what might be considered `'core" activities in CS&E, the committee chose to exclude these categories, recognizing that in doing so it might also exclude, for example, those for whom CS&E database work was some part of their educational or research portfolios. Partly for definitional reasons such as these, data sources for Ph.D. production in CS&E conflict, as illustrated in Table 8.1.i However, despite these discrepancies, it is clear that growth in CS&E Ph.D. production has been large in percentage terms when measured over the last decade or so. In the short term, the future supply of Ph.D.s depends in part on the pipeline of people obtaining bachelor's and master's degrees. The major source of CS&E Ph.D. students is students graduating with bachelor's degrees in CS&E. As noted in Table 8.2, the number of bachelor's degrees awarded in CS&E climbed sharply in the early 1980s but began to drop after 1986. If this indicates an enduring TABLE 8.1 Discrepancies in Data Describing Ph.D. Production in Computer Science, 1980 to 1989 Published Number of Doctoral Degrees Awarded Source 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 NSFa 218 232 220 286 295 310 399 450 515 612 NCESb 240 252 251 262 251 248 344 374 428 538 OSEPC 231 248 231 276 269 264 365 398 454 531 Taulbeed N/A 230 235 244 256 326 412 466 577 625 aNational Science Foundation, Science and Engineering Indicators, NSF, Washington, D.C., 1991, p. 247 ("computer science" not otherwise qualified; category includes in- formation sciences). bNational Center for Education Statistics, Digest of Educational Statistics, 1991, U.S. Department of Education, Washington, D.C., NCES 91-697, Table 256 (category labeled "computer and information sciences"). CData from Survey of Earned Doctorates, Office of Scientific and Engineering Per- sonnel, National Research Council, Washington, D.C. ("computer science" excludes information science and computer engineering but includes "computing theory and practice," which is often listed as a subfield of mathematics). dTaulbee surveys; see David Gries and Dorothy Marsh, "The 1990-1991 Taulbee Survey," Computing Research News, Volume 4(1), January 1992, pp. 8 if. See also Orrin Taulbee, "Annual U.S. Summaries of Ph.D. Production and Employment in Computer Science, 1970-1985," SIGCSE Bulletin, Volume 18(3), September 1986, pp. 2-8,12. Data through 1984 above taken from the latter paper. ("Computer science" excludes infor- mation science and computer engineering but includes degrees awarded in both the United States and Canada.)
Go a' 5 - no · - o ·-l ~5 o a' 5 - a: ¢ EM au in ¢ to ~ a O 50 En ~0 00 00 00 00 Cal 00 00 ~ 1 to ~0 au ¢ au a 241 Cal ~ Cal =t ^ Cal ~ ED As' O o ~o CO to ~ LO [` Do ~ Us Us ~ o Do ^ Cal ~ Lr) `~ oo CN oo ~ .4 _ o ~t C~- ~o Lr o ~' C~' oo o ~ o ~ _ o ~CO C ~U~ CO ~ L~ O LO O ^ C ~ ~ C`i ~) oo _ CN ~ ~=N 00 CN 00 00 ~D ~ ~- ' C~ O LC) ~ ~ ~00 00 di ~ CN O' ~LC) CM ~ CO 00 ~C~ ^ 00 ~ CN ~ Lo Lr) ~ON _' CO ~ 00 00 ^ CN ~ ~O ~ .¢ ~ ¢ ~¢= ~ ,,, a 0 ~, . ~ d~ ~_ V) a - a 5 - ~n .= . ° cn C~ - _ i_ . cn C~> . - ° - ~ (t . ~ O ~ O -'S .= ~S U) a, CJ ~ ~ U Q ~ ~ (t U) ·- ~o CC ~ o ~ - 3 U] a~ bO a o . UO au Cd . U) . ~ cn O . ~ z ~ o u . . o C~
242 COMPUTING THE FUTURE trend, it could portend difficulties for the supply of quality graduate students in CS&E2 unless the attrition in supply is limited strictly to undergraduate students of lower quality. (Representatives from the computer industry who briefed the committee noted their concern about dropping degree production as well, since they are major em- ployers of persons with bachelor's degrees in CS&E.) The downturn in bachelor's degrees awarded has been a matter of some speculation in the academic community. Some believe the downturn is temporary, and indeed some institutions (such as Berke- ley and MIT) have reported an upturn in 1991 in undergraduate en- rollments. Others have reported no such turnaround. There is also no consensus concerning possible reasons for the downturn. Some note that the peak occurred roughly five years (i.e., about the average time it takes to obtain a bachelor's degree) after the introduction of the personal computer; perhaps personal comput- ers have demystified the field, reduced the need for students to ma- jor in CS&E to obtain access to computers, or otherwise changed its image and allure. Others have argued that an increase in the number of students taking programming in high school has led to the down- turn. Although Ph.D. production in CS&E has risen rapidly in the last decade, it is still small compared to that of other fields, as Table 8.3 indicates. Note in particular that the number of CS&E Ph.D.s pro- duced in 1989 is less than two-thirds that of its parent disciplines, electrical engineering and mathematics, and about one-half that of physics. Production of Ph.D.s in CS&E is also time consuming: the total time to degree (i.e., the interval between receipt of a bachelor's degree and receipt of the Ph.D. degree) is somewhat longer for CS&E than for other major science fields, and the change in total time to degree has been largest for CS&E and biological sciences (Table 8.4~. Given the employability of individuals with strong computer skills, it is likely that the reason for the greater total time to degree of CS&E Ph.D.s is that many with bachelor's and master's degrees in CS&E enter the work force prior to resuming Ph.D. study in CS&E. This possibility is consistent with the approximate comparability of "reg- istered" time to degree for CS&E Ph.D. recipients and those in other fields. The primary source of support for Ph.D. recipients in CS&E is research assistantships, although the percentage of recipients with this source of support has dropped slightly over the last decade (Ta- ble 8.5~. Of interest is the substantial fraction of recipients who are supported by "other" sources (which include industry, family, non- U.S. government support for foreign students, savings, and self).
243 oo o - ·_4 al o En En U ¢ o o PA o o z Cal ¢ Ed a o ~ .~ _ ~ ~ O O ~ Ed C' or or 00 00 00 00 00 Cal DO 00 00 O 00 Cat U O a ~ o CN CO et O Cal ~ ~CN CO 00 OF ~- ` cat ~Dot O CN Lo 00 Cal Lr) ~ O ~ ~Cal Lf) O 00 ~O '_ 00 ~ CM ~ ~- 0 r~ ~ ~ C~ O ~00 0 C~) ~ ~ ~Ct) O ~ ~ CO ~00 C`l ~ ~O ~ 43~\ 'S~ C~ U) O O \= LO ~ oo 00 00 C-1 ~) ~O. 00 C() ~ ~C{) 00 ~ o0 C~ ~ ~O C~ ~ ~1 ~ ~C~ C~ ~ d. O ~ C~ ~ ~0 00 00 ~ oo C~1 LO ~4 d~ ~ C~ ~O CN ~ ~0 00 ~ C~ 0O 1_ ~C~ C ~ ~ ~00 U~ 00 O d~ C~ C C V ~ ~ - = 5 - ~ ~_ t:L} ~ - Ct5 (t O {t a. ~ C ~ ~ C V O '= C o U) U: CD o - . ~9 cn tn v ~- cD o - ~u o ~ ~ u o ~ =. .Q c ~ ~ ~o , o o ~ . v ~ ~ bO O 0.= a oo C O ~ ' ~ 0 U) 0 O ·~ ~ (, ·= O o. · V ~ ~, O ~ 5- 0 - o . z - o CD . 5 - ·_I -0 C) ~ a Ct ~ a O .. b4 ~; cn 0 ~ UO _ v
244 COMPUTING THE FUTURE TABLE 8.4 Time (in Years) to Doctoral Degree for CS&E, 1980 to 1989 Field 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 TOTAL TIME,a B.A. to Ph.D. Computer science 7.3 7.7 7.7 8.3 8.5 8.5 8.7 8.6 9.0 8.5 Computer engr. 7.6 8.2 8.3 8.1 8.5 8.9 8.8 8.5 8.4 8.0 Electrical engr. 7.3 7.5 7.7 7.8 8.0 7.9 7.9 7.7 7.8 7.7 Mathematics 7.0 6.9 7.0 7.4 7.8 7.8 7.3 8.0 8.1 7.7 Physics and astronomy 7.2 7.0 7.4 7.2 7.2 7.4 7.3 7.3 7.3 7.2 Chemistry 6.0 6.0 6.0 6.2 6.3 6.4 6.5 6.5 6.5 6.5 Biological sciences 7.0 7.0 7.2 7.4 7.8 7.9 8.1 8.1 8.2 8.3 REGISTERED TIME,b B.A. to Ph.D. Computer science 6.0 6.0 6.3 6.4 6.2 6.1 6.4 6.5 6.6 6.4 Computer engr. 5.9 6.3 5.8 5.7 6.0 6.3 6.0 6.0 5.9 6.0 Electrical engr. 5.7 5.8 5.9 5.7 5.7 5.9 5.7 5.7 5.8 5.9 Mathematics 5.9 5.9 5.9 6.1 6.1 6.3 6.0 6.3 6.3 6.1 Physics and astronomy 6.3 6.2 6.4 6.4 6.5 6.5 6.2 6.3 6.3 6.4 Chemistry 5.2 5.2 5.2 5.4 5.4 5.5 5.5 5.5 5.5 5.5 Biological sciences 5.9 5.9 6.0 6.1 6.4 6.4 6.4 6.5 6.5 6.5 aTotal time refers to the elapsed calendar time between the award of the bachelor's degree and award of the doctorate. bRegistered time refers to the time actually spent in pursuit of the Ph.D. after award of the bachelor's degree. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineer- ing Personnel, National Research Council, Washington, D.C. As Table 8.6 indicates, a substantial fraction of new CS&E Ph.D. holders plan to go directly into faculty positions rather than the post- doctoral positions that characterize other fields. Industry absorbs a substantial portion of CS&E Ph.D.s as well. Increasing Ph.D. production in CS&E to 1000 per year is one stat- ed objective of the HPCC program. Given the lack of a systematic study of recent academic opportunities,3 the appropriate level of Ph.D. production for the CS&E field is a matter of some controversy in the community. On the one hand, many new CS&E Ph.D.s (and their faculty men- tors) report a recent tightness in the academic market, suggesting that even current levels of Ph.D. production are high given the de- mand for new faculty. On the other hand, other observers believe
HUMAN RESOURCES TABLE 8.5 Percentage of CS&E Ph.D. Recipients Receiving Primary Support from Various Sources Source of Support 1981 1983 1985 1987 1989 Teaching assistantship Research assistantship Fellowship Other 16 21 41 10 32 18 17 22 38 43 41 39 9 9 7 8 33 30 35 32 NOTE: Percentages of Ph.D.s with each type of support are based on the number with known sources of support. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineering Personnel, National Research Council, Washington, D.C. TABLE 8.6 Breakdown (by Percentage) of Work Plans of New Ph.D.s in Nlarious Disciplines, 1989 Percentage Choosing Category of Work Indicated 245 Discipline Total New Postdoca Academicb Industry Govt.d Othere Unknownf Ph.D.s or SelfC CS&E 648 11 43 29 3 3 10 Biological sciences 4115 68 12 6 4 3 7 Chemistry 1970 50 6 31 2 2 9 Electrical engr. 995 14 26 37 6 3 14 Mathematics 847 23 49 7 3 3 13 Physics and astronomy 1274 58 8 14 5 2 12 NOTE: Percentages include Ph.D.s with definite plans, negotiating, and seeking in each category at the time of the survey. aTemporary position in any sector. bPermanent position in academia (U.S. or foreign); may or may not be faculty. CPermanent position in industry (may or may not be computer industry), or self- employed. Permanent position in government (federal, state, local, or foreign). eIncludes nonprofit organizations, elementary and secondary schools, international organizations. fPlans unknown at time of response to survey. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineer- ing Personnel, National Research Council, Washington, D.C.
246 COMPUTING THE FUTURE that the reported tightness refers to faculty positions in the top tier of major research institutions, and that demand for new CS&E Ph.D.s is higher in other sectors, such as mathematics and computer science departments in four-year colleges. (The filling of such positions by CS&E Ph.D.s might well have a substantial and positive impact on the level and quality of CS&E instruction at such institutions, as sug- gested in Chapter 4, "Education in CS&E.") Since CS&E Ph.D.s have major roles to play in the computer industry and throughout society as well, some even suggest that 1000 Ph.D.s per year will ultimately prove inadequate (especially if their skill sets are broadened to ac- commodate responsibilities other than traditional CS&E research). Achieving this dispersion may entail a shift in job expectations among new CS&E Ph.D.s, as discussed in the Chapter 4 section "Employ- ment Expectations for Holders of CS&E Degrees." Even with an expansion in the number and size of Ph.D.-granting departments, positions in these departments will be only a portion of the total employment base for CS&E Ph.D.s. Information on the demand for holders of bachelor's and mas- ter's degrees is even less certain than that for holders of Ph.D.s. It is known that a very large fraction of bachelor's and master's degree holders go to industry and commerce upon graduation, and it makes sense to assume that a significant fraction of them take computer- related jobs (e.g., programming).4 Most current or proposed definitions of "computing profession- al" or "computer specialist" inevitably reflect a narrow characteriza- tion of the position as one in which a substantial portion of the job responsibilities require nonroutine interaction with a computer. Fed- eral statistics experts recognize that a finer degree of differentiation of computing professional is needed, and a proposed revision to the master list of occupations, the Dictionary of Occupational Titles, may add perhaps 30 computer-related occupations. A finer differentia- tion is made possible by both growth in the number of people in computer specialist jobs (supporting accurate statistics on subgroups) and recognition of the diversity of computer-related jobs. Moreover, narrow characterizations of the employment opportunities for CS&E graduates may become increasingly less appropriate (Chapter 4~. COMPOSITION OF ACADEMIC CS&E Representation of Women and Minorities Total numbers and trends tell only part of the story. Prospects for the CS&E talent pool depend also on its makeup. Women and
HUMAN RESOURCES 247 non-Asian minorities continue to be underrepresented in CS&E rela- tive to their numbers in the population at all levels in the CS&E educational pipeline. As shown in Figures 8.1 and 8.2, CS&E has shown no demonstrable improvement over time in the rates at which Ph.D.s have been awarded to women and non-Asian minorities. At present, CS&E attracts women and non-Asian minorities at approxi- mately the same rates as for the physical sciences at all levels, as noted in Table 8.7; however, for both fields, women and minorities are increasingly underrepresented at higher levels of educational at- tainment. The representation of women and non-Asian minorities in faculty ranks is somewhat lower than their representation as recipients of doctoral degrees in CS&E. According to the 1990-1991 Taulbee sur- vey,5 women and non-Asian minorities account for about 7.5 percent and 2.2 percent, respectively, of all tenure-track and tenured faculty in Ph.D.-grantir~g CS&E departments. About 4.4 percent of all full 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 1 1 1 1 1 1 1 1 1 1979 1980 1981 1982 Phys~,9~__~ / \ / ' , ~ \ ~\. / \ \~/ CS&E ~--~ an' A / / \% / \ / \'d' 1983 1984 1985 Fiscal Year 1986 1987 1988 1989 FIGURE 8.1 Percentage of doctorates awarded to women in CS&E and in physical sciences (physics, astronomy, and chemistry), 1979 to 1989. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineering Personnel, National Research Council, Washington, D.C.
248 5% COMPUTING THE FUTURE 4% r ~1 3% \. 2/ ~\\ / a too/ 1 1 1 1 1 1 1 1 1 o 1979 1980 1981 1982 1983 1984 Fiscal Year 1985 1986 1987 1988 1989 FIGURE 8.2 Percentage of doctorates awarded to non-Asian mirrorities in CS&E and in physical sciences (physics, astronomy, chemistry), 1979 to 1989. Percentage is calculated on basis of all Ph.D. recipients who are U.S. citizens or permanent residents, since data on race are not collected for those with temporary visas. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineering Personnel, National Research Council, Wash- ington, D.C. professors in these departments are women, and about 1.7 percent of all full professors in these departments are non-Asian minorities. Anecdotal evidence that enrollments of women and minority stu- dents are shrinking disproportionately has prompted individuals, departments, and professional organizations to examine opportuni- ties for women and minorities. Of special concern is evidence sug- gesting barriers to full participation by women and minorities. Al- though many of these barriers are typical in all science and engineering fields (Box 8.1), they are disturbing in light of the fact that CS&E is generally younger than other scientific and engineering disciplines. One might have presumed that the relative youth of CS&E relative to, say, physics, would have led to a more inheriting and welcoming
HUMAN RESOURCES 249 TABLE 8.7 Percentage of Degrees Awarded to Women and Non Asian Minorities in Computer and Information Sciences (CIS) and in Physical Sciences (PS), 1989 Percentage of Degrees Awarded Women Non-Asian Minoritiesa Degree CIS PS CIS P5 Bachelor's 30.S 30.9 13.5 8.0 Master's 27.9 26.8 6.3 4.4 Doctorate 17.6 19.0 1.8 4.7 NOTE: The slight discrepancy in percentage of doctorates awarded between this table and those of Figures 8.1 and 8.2 is due to the inclusion of information sciences and the exclusion of computer engineering in this table. aFigures for non-Asian minorities include only U.S. citizens and permanent resi- dents. SOURCE: National Science Foundation, Science and Engineering Indicators, 1991, Wash- ington, D.C., 1992, Tables 2-7, 2-8 for bachelor's data, Tables 2-14, 2-15 for master's data, and Tables 2-16, 2-17 for doctorate data. environment for women and minorities. That this is not so suggests inclusive view of the field. These trends have been recognized to a certain degree. For ex- ample, the Pearl et al. article cited in Box 8.1 summarizes a report prepared by the ACM Committee on the Status of Women in Com- puter Science. The Leveson report cited in Box 8.1 was prepared for the NSF Advisory Committee on Women in CS&E and described a variety of activities that the NSF could undertake to improve the status of women in CS&E. In addition, a variety of outreach pro- grams have been suggested; such programs include those aimed at encouraging women and minorities to take high-school courses in mathematics and science, computer camps specifically for girls, pro- duction of game software designed to appeal to the interests of girls, support groups for women and minorities to reduce their sense of isolation in graduate school, and introductory computer science courses that emphasize the use of computers as tools. Alternative CS&E degree programs can also be designed for adult students who wish to reenter the work force. For example, the Electrical Engineering and Computer Science Department at the University of California, Berke- ley has an outreach program, the Computer Science Reentry Pro a missed opportunity to support a more
250 COMPUTING THE FUTURE 8"~"~.""''~'x'1.2 'a."""""''''''''' ' ' ' ''''''' ' ''' '''' ''' ' ' ' ' Do' ............ ~ ... ...... .. ... .................. ............................................................................... .A,.,., .N,,,,,,.D,,,,, .M,,,.,.t,.N,,,,o,~T,l.,,65 IN Att FIEtDs oF sclE cE.- ND it.. .... ................. .... . ...... . . . . . . . .... . Mad 3~ - `n $~) ~ CItIS ~ .. matte cour$~s '0 bath school. than ~ ~ 'n ~ ~ th ~ fix o~ oy paren~ teachers end pee~ A m . ~.,~.~t,I,0. ~S the pu~O ~ ~ ~ ~ .... i, , ............. ... .... , i , , , .............................. .. ,, . , i ., . . . . ~ . .... ~ . ~ MY ~ . ~ . . ~.r,' ~. ~.~.~. ~.~.~ , in.} t.~.1 ~ ~ ~ ~ . . .... .. . . ..... .. ~ ... ........ . . .. . . .... ..... ............ ....... . . .......... . .. .. ..... ... · t~ ~ ~ _ Th ~{ ........................ .. .. .. . .. ................ .... . . ~. ... . ~.~. ~s mnt . ~,.~.~. ~anC . ~. ~.~.~.~. ~^~.~. ~o. ~.~ manors ~< . t,l~.~.~ ::,, ::':,,, :::2:2.. . :::.~..: .: .:-.::Y:::~.', .:,:',:',, ........................................ . . . . ~Ye: ~l~w~ env~ro~me ~1 ~: |~: ~ ~ ....... . . .. ..... A. mct31. dIscrImI.~.n hIt. ~. am dim . plays ~ ove~ dIsct mInatIOn - ~y than [n the .~. .. . .. . . ... .. . :::.~::: :~ : netwo vancement ... ^r lo: ~ =~:s ^r =~ n== . . . : : : : : : ::::: : . . :: :: : ~.~.~ The tt. e do i :: : : : : . :: mous and s&.E p~.~s .~. o~n spend. consIderable t. e. ln .. . .. .. . ~ . . ... ~ ~ ~ ~ .~.. . ~ ...~....~.~..~.~ wOme~ a~.~. : :: ::: ::: ::::: :: ::::: ::::::::: : ::: : : : : :: :: ::: : ::: :: ::::: ::: ..~.~ . : : : : ::: :: :::: :: ~ ~:~^ ~ K~: :~ +~:: :: : ..:~: `, =:" =l . ~ $:~: ~: $~.~. ~.'.. : :.: : :::. ~ . . . . . . ::::::::::::::~:::::::~::::~::::::~ :~:::::::::::::: :::::::: : : : : ::::::: : :: ::::: .. .. ~. . ... ~ . .. . ~ . ... t.~::~:~n :: ::~:~:~:n m(::~:~s ls: m~%s ~r ed %~t msources tower mChet ex~t 2' :: :'::::: . . . ...~.~.~.~:~s to adw.~::.: :: : ::: ~ : :::: : ~ :: : :: :: :: ::: ::: ::: : : : :::: :: :: ::: : :::: : .::: ::: ~ ::~: :: :::::::.: :: : :: ::::::: .2....'.'..'.'.'.'."''"'SVU~'E>"~""""""''"#'DC'l'u'des"'""''l'n~t'lOn""'""~# "'"'''N'a'n~''"'''~m'n""""'"'0~"""'' D"'""'"' :: ::::: ::~::: :: : :: :::: : :::: : :: ::: : :::: : ~.. .~. ~ .. . ~ ~ · '',,'" : . ~ K 0~ ~, ~ N~on ." '.' "",'.' ' ' ''"""D''C""""",""~' " 'b' ''."''.' "'9'89.'""'' ' 'd"."'.' '' ' ',,,'~' .' 't.'.'.'.'.'.'.' ' ,' .h . ~,.,., ,^. ' ' 2 2 K ' "" " . i~ " .,.2 :::' ' ' " 2 Ct6~tt~ " 0~ " / t " 6 " `~it~6 0~ "t " 0 5~5 ~ " ~6# # '' '" .,., ,, , , ~. .. .. . .~.~` '~.~.h ~ nn~ . : ~ ~ K" "''"'"' ..................... f ~.~.,y ~t,.~. ~`,.,.,~.,f,~ .,,.,,,.,~.~ .,,,,,=w,~.~.,f,
HUMAN RESOURCES 251 gram, designed for women and minorities who already have degrees. The academic core of the one-year program consists of a three-semes- ter introductory series arid three other courses in digital design, effi- cient algorithms, and discrete mathematics. The program also pro- vides tutoring. Results appear promising.6 These activities suggest that the CS&E field is awakening to the fact that involving all types of students more fully can broaden and enrich the pool of talent. Involvement of Foreign Students As in other scientific and technical fields, a significant fraction of CS&E graduate students consists of individuals who are not citizens or permanent residents. These foreign students account for a some- what higher fraction of Ph.D.s in CS&E than in the physical sciences, and the trend is uniformly upwards (Figure 8.3~. 50% 40% 30% 20% 10% 0% CS&E Phvsical Sciences -I ~ 1 1 1 1 1979 1980 1981 1982 1983 1984 1985 Fiscal Year 1986 1987 1988 1989 FIGURE 8.3 Percentage of doctorates awarded to foreign students in CS&E and in physical sciences (physics, astronomy, and chemistry), 1979 to 1989. Foreign students are defined as those with temporary visas. Percentage is calculated on basis of all Ph.D. recipients whose citizenship or visa status is known (always over 92 percent). SOURCE: Survey of Earned Doctorates, Office of Scientific and Engineering Personnel, National Research Council.
252 COMPUTING THE FUTURE The implications of this tiered are at present unclear. One issue is whether foreign recipients of U.S.-awarded Ph.D.s return to their r~a- tive lands (creating a "brain drain" from the United States to poten- tial foreign competitors) or whether they stay in the United States. One data point is that in 1989, the percentage of new Ph.D.s in CS&E that planned to work abroad in 1989 (7 percent) is much lower than the number who have temporary visas (about 35 percent). These data also suggest that rrew CS&E Ph.D.s tend to stay in the United States irt proportions about equal to those in other fields of science (Table 8.8~. These data do not account for visa-expiratior~ lag times, but a 1989 National Science Board (NSB) report noted that "overall, the U.S. research system shows a dependence on foreign scientists and engineers, and this dependence is expected to contin- ue."7 A second issue is whether foreign students displace U.S. students. The 1989 NSB report also noted that "the impact of foreign enroll- ment on the quality of programs was generally viewed as positive" and called special attention to "a shortage in the supply of high qual- ity U.S. applicants [and] a surplus of high quality applicants from abroad" and to a "substantial dependence upon the supply of foreign applicants . . . to maintain the quality of graduate programs [in com- puter science, physics, chemistry, and mathematics]."8 The NSB also concluded that "both industry and engmeer~ng schools would experience severe problems if engineering schools should se- verely restrict the traixling of foreign students or if the influx of for TABLE 8.8 Breakdown (by Percentage) of Planned Residency of New Ph.D.s in Various Disciplines, 1989 Percentage Choosing Residence Indicated Total New United Other Discipline Ph.D.s States Countries Unknown CS&E 648 60 7 33 Electrical engineering 995 52 10 37 Mathematics 847 54 11 36 Physics and astronomy 1274 59 8 33 Chemistry 1970 72 5 23 Biological sciences 4115 71 7 22 NOTE: Percentages include Ph.D.s awarded to both U.S. and foreign citizens. SOURCE: Data from Survey of Earned Doctorates, Office of Scientific and Engineer- ing Personnel, National Research Council, Washington, D.C.
HUMAN RESOURCES 253 eign students would diminish abruptly and significantly" industry because "a significant proportion of foreign graduate students ulti- mately obtain employment in the U.S." and engineering schools be- cause "U.S.-born students alone would be insufficient to keep engi- neering education and research programs at; their present le~rel."9 A third concern has been that high percentages of foreign stu- dents appear to correlate with low percentages of women across sev- eral scientific disciplines. Again, the reasons for this correlation are unclear. It is a matter of record that foreign students are predomi- nantly male, and so large numbers of foreign students would bias the overall gender balance towards men.~° But in addition, some have speculated that foreign cultures tend to be less accepting of women as scientific workers than is American culture, and that attitudes brought by foreign-born faculty and graduate students to American graduate education tend to discourage the full participation of women. Oth- ers have argued that the fields involved have simply found it less threatening or difficult to seek qualified students from abroad than to undertake the large-scale changes that would be necessary to at- tract larger numbers of women to these fields. Youth and Rapid Growth of Computer Science and Engineering The median age of faculty in a given field is one indicator of the maturity of that field. In 1989 the average doctoral faculty member in computer science was 2.8 years younger than counterparts in other scientific and engineering fields (Table 8.9~. These ace distributions v also suggest that faculty retirements in computer science are likely to lag somewhat behind those in other science and engineering disci- plines.~2 Note that the median age for faculty in computer science in 1989 was the median age for all science and engineering faculty in 1981 an 8-year lag. The youth and rapid growth of CS&E are also reflected in the distribution of its faculty ranks. As noted in Table 8.10, the fraction of CS&E faculty with the rank of full professor actually decreased between 1977 (35 percent) and 1989 (30 percent), probably as the result of a rapid influx of new assistant professors; the comparable fraction for other disciplines grew during the same period. Still another indicator of a field's youth is the fraction of faculty with degrees awarded in that field. Although the percentage of aca- demic doctoral faculty working in all institutions in computer sci- ence or computer engineering who also had Ph.D.s from computer science or computer engineering departments grew from 29 percent
254 COMPUTING THE FUTURE TABLE 8.9 Median Age of Faculty (Tenured arid not) Working in Various Fields, 1977 to 1989 Median Age (in Years) Field 1977 1979 1981 1983 1985 1987 1989 Computer sciencea 38.4 39.5 40.3 40.9 41.3 43.3 43.4 Electrical and electronic engineering 44.3 47.0 46.7 47.0 47.2 48.0 47.8 Mathematics 39.1 40.6 41.9 43.1 44.4 45.5 46.4 Physics and astronomy 40.9 43.2 44.7 46.3 47.7 47.4 48.5 Chemistry 41.7 42.4 43.9 45.3 46.0 47.1 48.0 Biological sciences 40.6 41.5 42.4 43.2 43.9 43.6 44.2 All science and . . engmeerlng fields 41.7 42.7 43.4 44.4 44.8 45.4 46.2 NOTE: Faculty without doctorates are not included in this tabulation. aExcludes information sciences and computer engineering. SOURCE: Data from Survey of Doctoral Recipients, Office of Scientific and Engi- neering Personnel, National Research Council, Washington, D.C. TABLE 8.10 Distribution (by Rank) of Faculty for Various Disciplines, 1979 and 1989 Percentage with Rank 1977 Discipline 1989 Full Prof. Asst. Prof. Full Prof. Computer science and engineering 26 35 28 30 Physics 17 43 13 52 Mathematics 25 36 19 51 Electrical engineering 18 48 19 56 Biology 26 34 18 39 All science and engineering disciplines 24 39 19 44 NOTE: Includes faculty with doctorates working at all academic institutions (except two-year colleges), both those that grant Ph.D.s and those that do not. SOURCE: Data from Survey of Doctorate Recipients, Office of Scientific and Engi- neering Personnel, National Research Council, Washington, D.C.
HUMAN RESOURCES TABLE 8.11 Degree Distribution of Doctoral Faculty Working in CS&E, 1977 and 1989 255 Percentage with Doctorate in Indicated Discipline Discipline 19771989 Computer science or computer engineering 2941 Electrical engineering 2525 Other engineering - 2620 Natural sciences or mathematics 127 Other 86 Total number of faculty with doctoral degrees from any discipline working in the fields of computer science or computer engineering 18735131 SOURCE: Date from Survey of Doctorate Recipients, Office of Scientific and Engi- neering Personnel, National Research Council, Washington, D.C. to 41 percent between 1977 and 1989 (Table 8.11), individuals with doctorates in other fields still make up more than half of all such CS&E faculty. With respect to the fraction of their CS&E faculty who have Ph.D. degrees ~ computer science or computer engineering, the Ph.D.-granUng departments differ sharply from the non-Ph.D.-granting departments (Table 8.12~. Between 1977 and 1989, the percentage of CS&E faculty TABLE 8.12 Percentage of Doctoral CS&E Faculty Whose Doctorate Is in CS&E, 1977 and 1989 Percentage with Doctorate in CS&E Type of School Worked at 1977 1989 Forsythe List schoolsa 33 55 (N=1091)b (N=2660) Non-Forsythe List schools 23 26 (N=782) (N=247l) aThe Forsythe List consists of institutions granting doctorates in computer science or computer engineering. The list from 1989 was used for this table. bN, number of faculty working at the indicated type of school in the year given. SOURCE: Data from Survey of Doctorate Recipients, Office of Scientific and Engi- neering Personnel, National Research Council, Washington, D.C.
256 COMPUTING THE FUTURE TABLE 8.13 Number of Doctoral Researchers Working in Various Fields, 1977 to 1989 Number of Researchers Field 1977 1979 1981 1983 1985 1987 1989 CS&E Electrical engineering Mathematics Physics and astronomy Chemistry Biological sciences All science and engineering fields 83,369 954 1,052 1,706 1,434 6,479 6,337 7,304 6,870 7,298 7,316 18,086 19,185 87,763 1,860 1,900 1,552 1,612 6,317 6,574 7,764 7,695 7,570 7,511 21,528 22,903 2,287 1,438 6,486 8,465 8,491 24,383 3,275 3,860 2,278 2,617 8,504 8,982 9,727 9,814 9,980 9,733 29,031 30,292 99,184 102,519 109,589 142,872 149,810 NOTE: Postdoctoral researchers are not included. All full-time and part-time doc- toral researchers, defined as individuals who indicate that their primary or secondary work is basic or applied research, are included. SOURCE: Data from Survey of Doctorate Recipients, Office of Scientific and Engi- neering Personnel, National Research Council, Washington, D.C. with Ph.D.s in CS&E grew substantially for Ph.D.-granting institu- tions, whereas that percentage barely changed in the non-Ph.D.-granting institutions. Recent Ph.D.s in CS&E who have entered academia in this time frame (and remained in the field) have gone predominantly to the Ph.D.-granting institutions. Both the number of researchers in CS&E and their output have increased substantially in the last decade. As Table 8.13 indicates, the number of academic researchers in CS&E increased by over a factor of four in the years from 1977 to 1989.~3 The output of CS&E researchers shows comparable growth. Between 1970 and 1990, the number of computer-related articles in the INSPEC database (a major research-oriented science and technology database) grew by 242 per- cent (from 19,278 to 65,863~; the number of physics-related entries in the same database grew by 99 percent (from 70,785 to 141,215~; the number of biology and life sciences articles in a different research- oriented database (Biological Abstracts) grew by 153 percent (from 211,759 to 534,911~. Finally, the number of Ph.D.s teaching in computer-related fields (including computer science, information science, and computer en- gineering) also increased, as did the number of degrees awarded in these fields at all levels (undergraduate and graduate) in each year of the period from 1977 to 1989 (Figure 8.4~.~4 (In 1977, the number of
HUMAN RESOURCES 5.5 5 4.5 a) 4 in c' a' .2 2.5 ct a' en' 1.5 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 257 / Total Degrees Awarded / ' ~_~_ / _~ / / PA - f iTotal Teaching Faculty I ~I I I I I 1 Year FIGURE 8.4 Relative growth in the number of degrees at all levels awarded and in the renumber of individuals with doctorates teaching in these areas, 1977 to 1989. 1977=1.0. SOURCE: Raw data on degrees awarded are pre- sented in Table 8.2, and otherwise taken from the same source as that for Table 8.2; in 1977, the number of degrees awarded at all levels was 9255. Degree data reflect "computer and information sciences." Comprehensive data on computer engineering degrees at the bachelor's and master's level are not available. However, it is known from the Taulbee surveys that at the Ph.D.-granting institutions, degree production in computer engineering is low compared to degree production in computer science. On the assumption that this trend holds for the non-Ph.D.-granting institutions as well, the ne- glect of computer engineering degrees in this figure does not appear unrea- sonable. Number of doctoral teachers includes those teaching in computer science, information science, and computer engineering and was obtained from the Survey of Doctorate Recipients, Office of Scientific and Engineering Personnel, National Research Council, Washington, D.C.; in 1977, the num- ber was 1495. degrees awarded at all levels in computer and information science was 9255,~5 and the number of faculty with doctorates teaching in computer science, information science, and computer engineering was 1495.~6 ~ However, the growth in the number of degrees awarded far exceeded the growth in the number of teaching faculty until 1986;
258 COMPUTING THE FUTURE such data do not account for the large amount of service teaching for non-majors that CS&E departments have provided. Figure 8.5 suggests that growth in the number of faculty posi- tions in CS&E did not keep pace with the growth in bachelor's de- grees awarded for several years, although if current enrollment trends continue, a better balance of degrees awarded to number of teaching faculty may be achieved. Note, however, that if the ratio of bache- lor's degrees awarded to number of teaching faculty in 1989 had matched the ratio for 1977 (i.e., 6.31 degrees awarded per teaching faculty member), a total of nearly 1200 additional filled teaching po- sitions would have been necessary in 1989. Clearly, teaching loads in CS&E are much heavier than those in other academic fields. More quantitatively, it would take over 11,000 additional faculty teaching in CS&E to achieve the degrees-to-faculty ratio (2.45 in 1989) that characterizes science and engineering fields across the board. 15 14 - 13 c~ 12 11 10 ~ 9 t 8 ~3 7 a' 6 5 In a' 4 ~0 3 a> 2 Ct o 1 ~i, "Computer" Fields _ - All Science/Engineering Fields ol 1 1 1 1 1 1 1 1 1 1 1 , 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 Year FIGURE 8.5 Number of degrees awarded divided by number of individuals teaching for computer science and for all science and engineering fields, 1977 to 1989. See source notes for Figure 8.4.
HUMAN RESOURCES 259 NOTES 1. The reasons for these discrepancies are unclear, but may include differences in who is asked to supply data (degree recipients or degree granters), different defini- tions of categories (e.g., "computer science" including or not including "information sciences"), different "binning" of the data, different institutions sampled (e.g., U.S. vs North American), and statistical sampling errors. Sources on patterns of employment reflect similar inconsistencies. For purposes of this report, all data issues that involve Ph.D. production levels or employment patterns draw on data provided by the Survey of Earned Doctorates (SED) and the Survey of Doctorate Recipients (SDR) of the Office of Scientific and Engineering Personnel (OSEP) at the National Research Council. The reasons for this choice are that OSEP collects data on Ph.D. production in a variety of fields (and thus cross-field comparisons can be presumed to have a measure of consistency in terms of category definition and the like) and that OSEP also collects a variety of statistics related to employment and graduation plans that are not collected by other surveys. The SED targets all those who received doctorates from U.S. universities in a given year but conducts a full census of this population. The Survey of Doctorate Recipients is conducted to obtain longitudinal data on employment of individuals with doctor- ates from U.S. universities over a 42-year time span. (Thus a figure reported in an SDR survey in 1989 samples from a universe of individuals who received their doctor- ates between 1947 and 1989.) Both surveys use self-reported classifications (so that, for example, the degree recipient is asked to categorize the field in which his or her doctorate is received). By contrast, the Taulbee survey best known within the CS&E field- makes inquiries of Ph.D.-granting departments to determine the number of Ph.D.s awarded, and it encompasses both U.S. and Canadian institutions. In recognition of a largely inadequate understanding of human resources in the computer field, the Computer Science and Telecommunications Board and the Office of Scientific and Engineering Personnel of the National Research Council held a work- shop in October 1991 to explore issues in the areas of data and taxonomy for computer specialists, demand for and mobility of people trained in CS&E, the CS&E pipeline and equality of opportunity, and implications for training. A report on this workshop will be released in the summer of 1992. 2. A major difficulty in tracking degree production at all levels in CS&E is the long lag time in the availability of data. Even as this report goes to press, 1989 is the most recent year for which comprehensive statistics on undergraduate degree production are available. Evaluating Ph.D. production is somewhat less problematic due to the relatively rapid publication of the annual Taulbee survey. 3. The Taulbee surveys report on departmental growth projected five years into the future. But the match between these projections and the actual number of oppor- tunities available is often poor. More to the point, the Taulbee surveys cover only the 150-odd Ph.D.-granting institutions and not the more than 850 other CS&E depart- ments in the rest of higher education. 4. No office or agency either tracks the employment of nondoctoral CS&E degree holders as systematically as the NRC's Office of Scientific and Engineering Personnel tracks employment plans of new Ph.D. recipients or has the data to correlate fields of employment with fields of degree. The Bureau of Labor Statistics does develop data in a couple of programs, both of which count "computer programmers" and "systems analysts and computer scien- tists," and forecasts demand in these categories, but such categories reflect the job responsibilities of those employed in those categories rather than their educational
260 COMPUTING THE FUTURE pedigree. At present, these forecasts predict growth in both occupational categories, and representatives front the computer industry who briefed the committee believe that industry will need large numbers of computer specialists for years to come. 5. David Gries and Dorothy Marsh, "The 1990-1991 Taulbee Survey," Computing Research News, Volume 4(1), January 1992, pp. 8 If. 6. Mary Grigolia, "Computer Science Reentry Program," Computing Research News, Volume 2(2), April 1990, p. 19. 7. National Science Board, Report of the NSB Committee on Foreign Involvement in U.S. Universities, NSB-89-80, National Science Foundation, Washington, 19.C., 1989, p. 19. 8. National Science Board, Report of the NSB Committee on Foreign Involvement in U.S. Universities, 1989, p. 8. 9. National Science Board, Report of the NSB Committee on Foreign Involvement in U.S. Universities, 1989, p. 7. 10. This point was made at the recent CSTB Workshop on Human Resources in CS&E. 11. "Most of these foreign teachers are men who come from cultures that do not view women as colleagues. The result can be what American women see as sexual harassment and as refusal to take them seriously as students." See Betty Vetter, "De- mographics of the Engineering Pipeline," Engineering Education, May 1988, pp. 735- 740. Cited in National Science Board, Report of the NSB Committee on Foreign Involve- ment in U.S. Universities, 1989, p. 8. 12. The Taulbee survey of 1990-1991 reports that in the 1990-1991 academic year, the 137 Ph.D.-granting computer science departments (with an average of 19.9 faculty members) had 35 deaths and retirements. In the steady state, a department with a faculty of 20 and an average faculty work life of 40 years (from age 30 to age 70) could expect to see, on average, about one retirement every two years, so that in a field of 137 departments, one could expect about 67 retirements each year. See David Cries and Dorothy Marsh, "The 1990-1991 Taulbee Survey," Computing Research News, Vol- ume 4(1), January 1992, p. 12. 13. "Academic researchers" are defined as doctorate holders working in CS&E as employees of an institution of higher education (but not two-year colleges) who indi- cate that their primary or secondary work is basic or applied research. Thus academic researchers include both faculty with research and teaching responsibilities and other academic scientists with only research responsibilities. 14. 1977 was chosen because it is the first year for which reasonably authoritative data on number of teaching faculty (grouping together all professorial ranks, instruc- tors, lecturers, adjuncts, and so on) are available. 15. National Science Foundation, Science and Engineering Indicators, 1989, NSF, Washing- ton, D.C., 1989, Table 2-12, p. 224. 16. Survey of Doctorate Recipients, Office of Scientific and Engineering Personnel, National Research Council, unpublished data.