Engineering Graduate Education and Research (1985)

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.

Supply and Demand The question of "demand" for engineers, whether undergraduate or graduate, has always been a difficult one. Many projections of demand are extrapolations from past trends. Some of these are quite sophisti- cated and include the effects of projected economic conditions and the movement toward high technology. Nevertheless, they are still extrap- olations, and suffer from the basic problem of all extrapolations-they cannot anticipate surprises. Other projections are essentially opinion polls concerning the future and usually carmot retain their validity for more than a year or so. The most comprehensive recent study of the scientist/engineer labor force was published by the National Science Foundation in 1984.64 This study, which made projections based on various scenarios of economic growth, foresaw a general balance between supply and demand for engineers through 1987, with the exception of three fields: aeronautical/astronautical engineers, computer specialists, and elec- trical/electronic engineers. For these fields possible shortages were projected. Among the remaining engineering specialties, supply and demand for industrial and mechanical engineering were projected to be in rough balance, while all other engineering fields were projected to have personnel surpluses. All science, as distinguished from engineer- ing, fields were projected to have surpluses. A difficulty for the present report is that the projections were not differentiated by degree level, so one cannot draw any information from them expressly regarding sup- ply and demand for engineers with graduate degrees. 17

18 ENGINEERING GRADUATE EDUCATION AND RESEARCH The present study cannot claim to be superior in accuracy to any other, and it deals with "demand" only to the degree that certain conse- quences could arise from current trends, which could affect the demand for Ph.D. engineers. However, it should be noted that Ph.D.s from some science fields have shown occupational mobility for activi- ties in which engineering Ph.D.s are also sought, particularly in indus- try. This topic is developed below under the heading "Ph.D.s in Industry. " Supply of Ph.D.s The "supply" of engineers is easier to quantify than is "demand," at least for four or five years into the future, because students presently enrolled in school con be counted. The number of engineering gradu- ates with bachelor's degrees can be roughly predicted four years hence on the basis of freshman enrollments in the current year, although major changes in students' perceptions of future employment pros- pects can upset these predictions if the dropout rate changes signifi- cantly. In the case of graduate degrees, one can make rough predictions of future master's degrees by assuming that the percentage of master's to bachelor's degrees awarded 1 year earlier willremain at about its recent level of 30 percent {see Figure S). The ratio of doctor's to bachelor's degrees 5 years earlier* has ranged from about 6 percent to 10 percent in the last 15 years, and has remained in the range of 6 percent to 7 percent or so for the last 9 years. Thus, if students' perceptions of the attractive- ness of the Ph.D. remam constant, 65 and in the absence of major pro- grams to stimulate Ph.D. production, one could predict reasonably well the supply of doctorates 4 or 5 years into the future, based upon the percentage of bachelor's recipients in the recent past who subsequently earned Ph.D.s. However, some evidence in the statistics shows that Ph.D. study is actually becoming less attractive to recent bachelor's graduates than was formerly the case. Table 1 shows that the number of engineering bachelor's degrees increased by 81 percent from 1977 to 1983. Table 3, on the other hand, shows that full-time doctoral enrollment increased only 47 percent in the same period. * Reference 15 shows that the average registered time for engineering Ph.D.s in 1982 was 5.8 years from B.S. to Ph.D. Reference 16 shows that the median total elapsed time between the B.S. and Ph.D. ranged from 7.5 to 7.9 years during the period 1976-1981.

SUPPLY AND DEMAND 50 40 Cal CD C: 30 Z 20 Cal 10 o 19 [1 1 ~ I Master's Degrees (As a Percent of- B.S. Degrees One Year Earl ier ) _ _ /V - '/~J~, I Percent of B.S. Grads Going\ / Directly to Graduate Study - -Doctor's Degrees ( As a Percent of B.S Degrees Five Years Earlier) , . . ~ - 1955 1960 1965 1970 1975 1980 1985 FIGURE 5 Ratios of master's and doctor's degrees to B.S. degrees. SOURCE: Data from Engineering Manpower Commission. Table 4 provides a way to estimate Future Ph.D. production, at least until 1988, by examining the annual input of new doctoral students. The second and third columns of Table 4 j"total full-time doctoral enrollment" and "doctoral degrees granted"J are taken from Table 3 and extended backward in time to 1967. The fourth column is the estimated number of continuing students. For example, the figure of 13,419 continuing doctoral students in the fall of 1983 is estimated by taking the total enrollment in fall 1982 {16,442 students) and subtract- ing from that the total number of doctoral degrees awarded during the academic year 1982-1983 3,023 degrees). The difference between the total enrollment for fall 1983 {18,228 students) and the continuing  students {13,419) must be the number of new doctoral students {4,809) who entered in the fall term of 1983. The estimated number of new students each fall is computed by this method and entered in the fifth column of Table 4. However, it can be seen that the behavior of these numbers is quite volatile: the first three figures in the fifth column are 3,325; 1,875; and 4,124, for example. Therefore, running three-year

20 ENGINEERING GRADUATE EDUcATION AND RESEARCH TABLE 3 Full-time Doctoral Enrollment in Engineering, Number of Doctor's Degrees in Engineering, and Percent of Estimated Ph.D.s Available for Academic Employment, 1970-1988 . Estimated Ph.D.s Available for Full-time Doctoral Academic Enrollment Degrees Employment _ . _ Foreign Total ~ ~ Foreign Nationals Percent Total Nationals Domestic Percenta Number . . _ 1970 14,802 1971 14,100 1972 13,460 1973 11,904 1974 10,628 1975 11,281 1976 10,963 1977 12,359 1978 12,321 1979 13,461 1980 14,465 1981 1S,472 1982 16,442 1983 18,228 Estimated: 1984 - 1985 - - 1986 - 1987 - 1988 - - - 3,620 - - 31.3 1,133 - - 3,640 741 3,169 - - - - 3,774 773 3,001 34.6 1,306 - - 3,587 708 2,879 - - - - 3,3621,014 2,348 29.5 992 - - 3,138891 2,247 - - - - 2,9771,060 1,917 36.2 1,078 4,383 35.3 2,813995 1,818 - - 4,273 34.6 2,573874 1,699 36.0 926 5,256 39.0 2,815929 1,886 - - 5,995 41.4 2,751982 1,769 34.9 960 6,876 44.4 2,8451,052 1,793 - - 6,756 41.1 2,8871,167 1,720 32.8 947 7,687 42.2 3,0231,179 1,844 33.3 1,007 3,250 3,400 - 3,600 3,750 3,900 1,300 1,360 1,440 1,500 1,560 1,950 2,040 2,200 2,250 2,340  33.3 33.3 33.3 33.3 33.3 1,080 1,130 1,200 1,250 1,300 aSee Table 6 of this report. SOURCES: Enrollment data: Engineenng and Technology Enrollments 1New York: Engineer- ing Manpower Commission, various yearsl. Degree data: Engineenng and Technology Degrees (New York: Engineering Manpower Commission, various years). averages of the figures in column five have been computed and entered in the sixth column in order to smooth the data. The figures from the sixth column {"estimated new students"J of Table 4 have been plotted in Figure 6, together with the figures for annual engineering doctoral production. The shapes of the two curves are similar, both showing pronounced "troughs," but with the troughs displaced by five years. In Figure 7, the two curves of Figure 6 have been superimposed. A dashed line has been drawn through the "estimated new doctoral students" curve from 1980 on as an estimate of the man- ner in which the new student input is trending. A dashed line has been drawn parallel to that trend curve as an estimated five-year extrapola- tion of doctoral degree production. By this method, the engineering

SUPPLY AND DEMAND TABLE 4 Growth in Full-time Doctoral Enrollment, 1967-1983 21 Continuing Students Estimated Estimated Total (enrollment in New Students new students Full-time Doctoral year x, less (total enrollment {running Doctoral Degrees degrees granted less continuing 3-year Enrollment Granted in year x + 1) students) average) 1967 15,376 2,614 - - - 1968 15,768 2,933 12,433 3,325 - 1969 14,298 3,345 12,423 1,875 3,108 1970 14,802 3,620 10,678 4,124 2,979 1971 14,100 3,640 11,162 2,938 3,399 1972 13,460 3,774 10,326 3,134 2,701 1973 11,904 3,587 9,873 2,031 2,417 1974 10,628 3,362 8,542 2,086 2,636 1975 11,281 3,138 7,490 3,791 2,899 1976 10,963 2,977 8,143 2,820 3,606 1977 12,359 2,813 8,150 4,209 3,188 1978 12,321 2,573 9,786 2,535 3,566 1979 13,461 2,815 9,506 3,955 3,415 1980 14,465 2,751 10,710 3,750 3,854 1981 15,472 2,845 11,620 3,852 3,815 1982 16,442 - 2,887 12,585 3,837 4,166 1983 18,228 3,023 13,419 4,809 - doctoral output of U.S. universities in 1988 is estimated to lie between 3,800 and 4,000. Figure 6 shows an apparent anomaly in that the Ph.D. production in the years 1977, 1978, and 1979 appears to be larger than the number of new students who entered 5 years earlier. {This effect is most visible in Figure 7, where the curves have been superimposed.) There are two explanations for this apparent anomaly. One is that not all students take the same time to complete their degrees. The figure of a 5-year delay is probably a reasonable average, but some students will finish in 3 or 4 years after initial enrollment as doctoral students, while others may take 7 or 8 years. {The average of S.8 years of registered time mentioned in the footnote on page 18 is from B.S. to Ph.D. and not from initial doctoral enrollment to Ph.D., which is the basis being used here. ~ The second explanation of the apparent anomaly is more powerful than the first: the calculations ~ Table 4 only allow for the new stu- dents who enter in the fall term of each year. New doctoral students also enter during the winter and spring terms, and there is no way to enumerate these from the available data. It is assumed in this analysis that the effect of the midyear enrollees is a constant one, serving princi

22 ENGINEERING GRADUATE EDUCATION AND RESEARCH foot Lit to o - 2 o g of A: o ~ 2900 o ~ 1000 o ESTIMATED NE. DOCTORAL STUDENTS EACH YEAR \ I RUNNING 3- YEAR AVERAGE I ENGINEERING . . , . 1970 1975 1~0 YEAR 1985 1990 FIGURE 6 Engineering doctoral degrees per year and estimated new doctoral students per year (running 3-year average!. use Hi> ~ 3000 O ,_ a. a: O o 02000 lo. ~7 to to  1000 a W ENGINEERING / DOCTORAL DEGREES ESTI - TED NE' DOCTORAL STUDENTS EACH YEAR ( CURVE FROH FIGURE 6 DISPLACED S YEARS T0 THE - RIGHT ) 0 1970 l97S 1 - YEAR leas 19~ FIGURE 7 Engineering doctoral degrees per year, with curve for estimated new doc- toral students per year from Figure 6 displaced five years to the right.

SUPPLY AND DEMAND 23 pally to shove the "estimated new doctoral students" curve in Figure 7 upward, but IlOt materially changing the slope of the dashed lines upon which the estimates are based. {Note that the foregoing considerations do not apply to the "engineering doctoral degrees" figures, which include all the Ph.D. degrees granted during the year including those granted during the midyear period. ~ The estimated figures for anneal Ph.D. production from Figure 7 have been transferred to Table 3 (years 1984 through 1988~. The propor- tion of foreign rational and domestic Ph.D.s is also shown there using constant 40 percent as the estimate for foreign nationals. Also, the numbers of Ph.D.s presumed to be available for academic employment are shown in the last column of Table 3, using a constant percentage of 33.3. The average number per year available for academic employment for the years 1984-1988 is calculated at 1,190 per year. The average number per year for the 1970s is calculated at 1,087 per year, from the entries in the last column of Stable 3. Thus, even though the Ph.D. production of the country is rising markedly, the average annual num- ber available for academic employment in the near future is only about 100 more than it was during the 1970s, when the engineering educa- tional establishment was smaller than it is today. It was during the late 1970s and early 1980s that the shortage of engineering faculty devel- oped, with the result that engineering schools reported 1,400 vacant faculty positions in 1982 nationwide.23 The question is whether the future supply will be enough to meet the needs of educational institu- tions and industry simultaneously. * Quality of Entering Graduate Students The question of the trends in quality of engineering graduate stu- dents has frequently been raised. In particular, In view of the decline in the popularity of graduate study in the 1970s, some have wondered whether there was a corresponding decline in the ability levels of those who were admitted. One cannot form an absolute judgment on this matter, because the available national data only show what has hap- pened to the ability level of the applicants as manifested by Graduate * A sunrey taken in 1983 showed 1,570 vacant engineering faculty positions nation- wide. See P. Doigan, "ASEE Survey of Engineering Faculty and Graduate Students, Fall 1983," Engineenng Education, October 1984. The percentages of unfilled positions, by field, were as follows: Computer Science/Engineering-15.8 percent; Electrical Engi- neering-9.7 percent; Mechanical Engineering-7.7 percent; Chemical Engineering- 7.1 percent; Civil Engineering-5.2 percent:

24 ENGINEERING GRADUATE EDUCATION AND RESEARCH Record Exam (ORE) scores Rid not what happened to the selectivity of those admitted from the applicant pool. Also, the GRE scores apply to all those seeking engineering graduate study and do not permit us to separate those who terminate with a master's degree from those who go on to doctoral study. GRE quantitative aptitude mean scores for prospective graduate stu- dents in engineering have ranged, during recent years, from a low of 649 {1974-1975) to a high of 665 {1972-1973~.~8 The most recent available data show a quantitative mean score of 657 {1977-1978~. The mean scores of prospective engineering graduate students have consistently been second only to students in the mathematical sciences, ranking just ahead of those in the physical sciences. In 1977-1978 the relative scores were as follows: GRE (Quan~tativeJ Mathematical Sciences ~ · . engmeer~ng Physical Sciences Life Sciences Health Professions Basic Social Sciences Arts and Humanities Applied Social Sciences Education 669 657 636 559 jl7 514 497 472 449 In verbal aptitude GRE scores, engineers consistently rank near the bottom. In 1977-1978, the relative scores were as follows: GRE (VerbalJ Arts and Humanities Physical Sciences Basic Social Sciences Mathematical Sciences Life Sciences Health Professions Applied Social Sciences ~ . . rng~neenng Education 532 517 516 504 503 498 483 459 446 Engineering students, then, compete very well in the quantitative GRE, and there was little variation in their scores during the 1970s. They are much less competitive on the verbal GRE, although the data show a slightly improving trend in the 1970s. i9 fit is worth noting that the range of mean scores for the quantitative GRE is 220 points,  whereas the mean verbal scores are more clustered, with a range of only

SUPPLY AND DEMAND i: 25 86 points.) The most that one can conclude from the data is that the quantitative ability of engineering graduate school applicants is high among the highest-and is holding steady. The low scores on the GRE verbal scale may be a measure of the degree of difficulty and frustration many engineering students experience later with respect to career advancement. This situation cannot be expected to change unless steps are taken to improve students' communication skills, or unless engi- neering succeeds In drawing a larger share of high-scoring verbal applicants. One small further insight regarding recent trends in the quality of engineering graduate students can be gained from an examination of the fractions of first- and second-decile bachelor's graduates at the Uni- versity of Illinois {Urbana-Champaign) who go directly on to engineer- ing graduate school. iThe data include those who go to graduate school anywhere, not just those who continue at Illinois. ~ Figures for the most recent 12-year period are shown In Table 5. The data in Table 5 for the first decile show that the peak was reached .n 1973-1974, when 65 percent of the first-decile students went directly on to graduate school. Subsequently, this fraction declined each year until 1978-1979 and 1979-1980, when it reached ~ low of about 40 percent. Since then, the fraction has increased each year until TABLE 5 B.S. Engineering Graduates at University of Illinois, Urbana-Champaign, Going Directly on to Any Engineering Graduate School, 1971-1972 to 1982-1983 Number From: Fraction From: 1st Decile 2nd Decile 1st Decile 2nd Decile 1971-1972 29 16 0.45 0.25 1972-1973 39 21 0.54 0.29 1973-1974 44 32 0.65 0.47 1974-1975 35 25 0.57 0.41 1975-1976 32 32 0.51 0.51 1976-1977 36 28 0.50 0.37 1977-1978 40 25 0.48 0.30 1978-1979a 26 19 0.39 0.29 1979-1980 42 40 0.40 0.38 1980-1981 57 39 0.48 0.33 1981-1982 67 55 0.54 0.44 1982-1983 79 46 0.64 0.37 al978-1979 includes data on spring graduates only; fall data not  available. SOURCE: Dean's Office, College of Engineering, University of Illi- nois at Urbana-Champaign.

26 ENGINEERING GRADUATE ED UCATION AND RESEARCH by1982-1983 it was almost back to its former high, at 64 percent. The fraction ~ 1982-1983 from the second decile going directly to graduate schooll37 percent) was the same as the overall average for the 12-year period. While no broad conclusions can be drawn from this information for the nation as a whole, Illinois is one of the country's major engineer- ing schools, and the information provides some encouragement regard- ing the quality of students entering graduate school. However, even though the Illinois figures are encouraging, they do not reveal how many of the brightest students are continuing on to doctoral study after completing a master's degree. Some engineering deans feel that not enough of the brightest students who are also U.S. citizens are pursuing doctoral study and that ways are needed to improve the attractiveness of doctoral work. Ph.D.s in Academic Employment The proportion of the nation's Ph.D.s that is available for academic employment can be estimated from Tables 6 and 7, which are taken from data collected by the National Research Council. is The Research Council periodically asks doctoral graduates about their postgradua- tion plans; the replies, by percentages of the total engineering doctor- ates for each year, are shown in Table 6, for selected years from 1960 to 1982. A 1981 follow-up study of doctoral graduates one year after gradu- ation four d that 100 percent of engineering doctorates seeking employ- ment had been successful and that their actual type of employment closely matched their plans. Table 7, providing selected data from that follow-up survey, shows that 95.5 percent of those who had been seek- ing postdoctoral employment immediately after graduation were in academic employment one year later. Of those who were "seeking employment," 14 percent had gone into academic employment, 83.6 percent were employed in industry or government, and 2.4 percent were in "other" employment. In Table 6, then, it is assumed that 100 percent of those who were planning to go into postdoctoral study would eventually enter aca- demic employment and that 14 percent of those who were "seeking employment" would also enter academic employment. These propor- tions were used throughout the entire 1960-1982 span in Table 6 although they are actually known for only 1981, and then only for a sample of the total population. Based on this kind of estimation, the proportion entering academia in recent years is believed to be about one-third. The fraction of one-third has been employed in the "esti

27 C~ o~ - C~ oo 1 o G~ ._ - CO - P~ o ._ C~ C~ ~o C~ o - CC C) 4 - e~ o 4 - o ~o ._ ._ ~4 o oo C~ oo Cx r_ C~ - ~_ C~ - C~ ~_  o C~ _ _ o C~ _ o - ~ Cx . . . . ~ o ~ _ _ e~ o _ . . . . ~ ~ Cx _ _ ~ . . . . o o _ _ ~ _ C~ . . . . o - _ ~ _ ~ ~ ~o ~ ~ Cx . . . . . . . o ~ ~ ~ U~ C~ C~ _ I~) c~ _ ~ C~ 1 . . . . . . . ~ ~ ~ C~ ~ ~ _. _ oo o ~ ~ ~ ~ . . . . . . . ~ oo ~ o ~ ~ _ _ _ ~ _ ,_ . . . ~ e~ U~ _a C~ ~ ~ ~_ _. ~o t _. ~ ~ o C ~o ~4 ~ o =, oo ~ ~ O I · · · 4 · _ ~ _ oo - C ~ ~: ~ - . o C~ U~ _ o - o ~- ~ - ~ .~O ~ ~ ~o ~  ~ ~ v a ~ ~v - ~ 5 ~ eat C) ~4 - C) ~ ~ G ~ ~ O .5 ~8- ~ ~ o _ _ ~ 8° °- ~ o ~ CO ~4 .a ~ o ~ ~ O o ~ o ~ ~ C) ~ 04 oc .~ o . o oo ~ ~ o . . . . o~ ~ ~ C~ oo . . . . o ~ ~ ~ ~ ~ d - . . . . C~ ~ C~ ~ ~ oo . . . . oo ~ oo :^ o - ~i O ;> 5 8 4 c~ ~ ~ ~ ~ ° et,V ~ 4- ~ ~ ~ =° ~ -C, ~ ~ ~ ° a - I C~ o U~ -, C~ ~oo U~ . . . . . C~ Cx C~ . . o ~ ~4 - I oo . . . . . o U~ ~ ~ - ¢, _ 5 E ° ~ °  o ~ ~ en e> _ I~ ~oo ~ . . .. ~ ~ o ~ . . ~ o I. . - o ~, oo ;` 6 o ~ C~ Z ·: g ._ ~S C~ .~ .~ CO _l ~: C~ CO 4 - .~ .= _] C~ C`i C~ o .= ~ ~C C~ . . L~ - O ~ V, ~

28 ~ ~ EDITS ~ RESEARCH TABLE 7 Engineering Doctorates, Post~adustion Employment Plans in 1980' and Actual Status in 1981 Actual Status, 1981 . _ _ _ _ _ _ _ _ _ Post~aduation Academic Covemment Postdoctoral Plans, 1980 Employment Employment Study Otbe . _ _ Postdoctorals 95 5 ~0 4~5~ - 0 Seeking employment 142 ~8 ~0 2.4~ SOURCE: 3~ Moors JP82: Docco=~a bangs ~ ^~d awes ~jve~- b~s {Washington, D.C.: National Academy Press, 1983). mated,, portion of gable 3 to project an estimaled supply of 1~080 to 1~300 doctoral engineers available snugly for academic employment the period 1984-1988. Percentages of Pb.D.s estimated to be in academic employment for the yews going back to 1970 brave been ~ansfe~ed Mom gable 6 to the next-to-last column affable 3 ~ order to comp~e the numbers pIe- sumed to be available in past years HI academic employment with the numbers estimated for the Lathe. As Was mentioned previously the aveIage~nnualnumbeIpIoTected for 1984-1988 is only about 100 Neater tab the Revere for the I970s. The data ~ Table 3 me those suppled by the Eng~eeI~g Manpower Commission (ETCH. noveve~ the Nadona1 Center for Education Sta- dstics {NCESJ and the National ResesIcb Council also collect cats on doctor degrees. Comparisons of these thee sources me abort gable 8. As can be seen' the fineries bow the thee sources We mosHy Within S percent to ~ percent of each hotbed aRbough ~ recent yews they brave deviated by 8 percent to 10 percent' With the EPIC Ivies on the high ~de. The variations should not be surprising' Since the thee agencies collect the data using deferent methods. The EPIC cats come Tom deans' offices' the NCES files We Tom registrars/ and the Rese~cb CouncDfiguIes Venom reportsfUedbytbeindividuaide~ee recipients.Someoftbe dtTerencesintbe Ogulesprobably arise hom norlepoIdng' buttbere me also likely 10 be differences among the agencies ~ inte~pretationsconcemingtbecategoriesin hick figures ~etobecounted.Ibefi~doicomputersc~nceisaprominentexam- ple.SomellniveIsideslumptboseincomputerscience ~ithengineers; somedo not.Portbepu~posesoftbisrepoI~ boweve~ the Eagles me sufficiently closets oneanotberso that the meow conclusions me not ~ected.Ihedatahom theEngineertogAi~npo~erCom mission brave

SUPPLY AND DEMAND TABLE 8 Engineering Doctorates, 1970-1983 : Comparison of Numbers Prom Various Sources 29 Ratio- Ratic EMC NCES NCES:EMC NRC NBC 1970 3,620 3,691 1.02 3,434 0.95 1971 3,640 3,638 1.00 3,498 0.96 1972 3,774 3,671 0.97 3,503 0.93 1973 3,587 3,492 0.97 3,364 0.94 1974 3,362 3,312 0.99 3,147 0.94 1975 3,138 3,108 0.99 3,002 0.96 1976 2,977 2,821 0.95 2,834 0.95 1977 2,813 2,~86 0.92 2,643 0.94 1978 2,573 2,440 0.95 2,423 0.94 1979 2,815 2,506 0.89 2,490 0.88 1980 2,751 2,507 0.91 2,479 0.90 1981 2,845 2,561 0.90 2,528 0.89 1982 2,887 2,636 0.91 2,644 0.92 1983 3,023 - - 2,780 0.92 - NOTE: EMC = Engineering Manpower Commission; NCES = National Center for Education Statistics; NRC = National Research Council. SOURCES: Engineering and Technology Degrees iNew York: Engineering Manpower Commission). Projections of Education Statistics to 1988-1989 (Washington, D.C.: National Center for Education Statistics, 1980); plus computer run updates, 1979- 1982. Science and Engineenng Doctorates: 1960-1981 (Washington, D.C.: National Science Foundation, NSF 83-309, 1983). Summary Report 1982: Doctorate Recipi- ents From United States Universities Washington, D.C.: National Academy Press, 1983). been chosen for use here because the EMC reports degree data at all levels and also includes enrollment data. An exception to the use of EMC data occurs in the case of foreign degree recipients. EMC reports these as "Foreign Nationals," but the Research Council figures, as reported by the National Science Founda- tion, place foreign students in two separate categories: {1) Non-U.S. Citizens, Permanent Residents; and {2) Non-U. S. Citizens, Temporary Residents. Table 9 gives the Research Council figures, with the last column showing the ratio between the temporary residents and the total. This ratio had reached 42.1 percent by 1983. According to Research Council data, approximately half of the "temporary resident" engineering Ph.D.s plan to stay in this country after graduation. iThe distribution of all engineering Ph.D.s in 1982, by citizenship, ethnic background, and sex, is shown in Table 10. ~ The "needs" of educational institutions can be roughly estimated as

30 ENGINEERING GRADUATE EDUCATION AND RESEARCH TABLE 9 Foreign Engineering Doctorates, 1970-1983 {National Research Council data) u.s. Total Citizens Non-U.S., Non-U.S., Permanent Temporary 471 518 519 622 704 815 813 773 768 815 851 943 1,028 1,169 Unknown Percent- Citizenship Temporary :Total 13.7 14.8 14.8 18.5 22.4 27.1 28.7 29.2 31.7 32.7 34.3 37.3 38.9 42.1 1970 3,434 1971 3,498 1972 3,503 1973 3,364 1974 3,147 1975 3,002 1976 2,834 1977 2,643 1978 2,423 1979 2,490 1980 2,479 1981 2,528 1982 2,644 1983 2,780 2,514 2,418 2,330 2,142 1,752 1,716 1,557 1,472 1,261 1,293 1,255 1,169 1,165 1,162  430 530 622 557 515 418 390 326 325 322 299 298 296 319 19 32 32 43 176 53 74 72 69 60 74 118 1~5 130 SOURCES: Science and Engineering Doctorates: 1960- ~ 981 Washington, D. C.: National Science Foundation, NSF 83-309, 1983) . Summary Report 1982: Doctorate Recipients From United States Universities (Washington, D.C.: National Academy Press, 1983). Summary Report 1983: Doctorate Recipients From United States Uni- versi~es "Washington, D.C.: National Academy Press, 19841. TABLE 10 Engineering Doctorates, 1982 Non-U.S. Citizens Permanent Temporary U.S. Citizens Residents Residents Total (incl. those who did not report citizenship) Male Female Male Female Male Female Male Female American Indian 3 - -- - -3 Asian 69 3 1686 545 20806 29 Black 9 - 11- 33 -So Hispanic 21 1 112 50 285 5 White 945 69 805 318 81,350 82 Other and unknown 44 1 121 of 1221 8 1091 74 282 14 997 31 2.52n 124 SOURCE: National Research Council, Survey of Earned Doctorates, Office of Scientific and Engineering Personnel, Doctorate Records File.

SUPPLY AND DEMAND 31 follows. Table 11 shows what happened between 1976 and 1982 to 51 large U.S. engineering schools* with respect to growth in enrollment and numbers of full-time faculty in the professorial ranks. During this period the number of students increased by 50 percent and the number of faculty by only 10 percent. Table 12 summarizes the data, showing that the "raw" student:faculty ratio for the aggregated 51 schools increased from 17.5 ire 1975-1976 to 24.0 in 1981-1982, an increase of about 37 percent. This ratio is called "raw" because it results simply from adding the total full-time undergraduate and graduate enroll- ments and dividing the sum by the total number of faculty in the professorial ranks. A more useful measure of true workload would come from converting such numbers to so-called full-t~me equivalents, both for students and faculty. Since the formulas for such conversion vary from institution to institution and are not consistently available in a national data base, "raw" ratios are used here. Such raw ratios are useful, nevertheless, to show the major change that took place between 1976 and 1982. The full-time enrollments in the 51 schools accounted for 50.4 per- cent of the total U.S. enrollment in 1975-1976, and 46.7 percent in 1981-1982. Using these percentages, the numbers of full-time faculty for all schools in the country were estimated by increasing the number 7,724 {1975-1976) by 1/0.504, and increasing 8,466 {1981-1982) by 1/0.467. This gave an estimated number of 15,320 faculty for 1975- 1976and 18,130for 1981-1982. Thelatter number corresponds closely with the number of 18,000 faculty reported for all U.S. engineering schools in the fall 1981 survey of the Engineering College Faculty Shortage Project.20 If the student:faculty ratio of 1975-1976 were to have been restored, approximately 6,750 additional faculty members would have been needed in 1981-1982. This has been labeled "esti- mated 1981-1982 faculty shortfall" in Table 12. An analysis by W. Edward Lear, using 1968-1969 and 1980-1981 as the comparison years, produced almost exactly the same number as the "shortfall": 6,700 faculty.2i Some have armed that there is no need to restore the 1975-1976 ratios and that engineering schools should find ways to increase their * The S1 schools were selected as follows: first, all those schools with a full-time undergraduate enrollment greater than 1,300 in fall 1975 were placed on the list; next, any schools that were not already on the list but were in the top lO Ph.D. producers in the years 1980-1983 (see Table 231 were added to the list; finally, since the sizes of faculty were taken from the publication Engineering College Research and Graduate Study for the years 1977 and 1983, any schools that were not listed in both years were removed from the list.

32 ENGINEERING GRADUATE EDUCATION AND RESEARCH TABLE 11 Enrollments arid Faculty for 51 Large U.S. Eng~neenng Schools, 1975-1976 and 1981-1982 . _ . School Arizona Auburn CCNY Calif., Berkeley Calif., Davis Calif., UCLA Calif. Polytechnic, San Luis Obispo Cincinnati Clarkson Colorado Colorado (Mines) Connecticut Comell Drexel Florida Georgia Tech. Illinois, Chicago Illinois, Urbana- Champaign Iowa State Louisiana State Lowell MIT Maryland Michigan Michigan State Michigan Tech. Minnesota Missouri, Columbia Missouri, Rolla N.C. State N.J. Inst. Tech. Nebraska Northeastem Northwestem Ohio State Oregon State Penn State Pittsburgh Polytechnic Inst. of N.Y. 1975-1976 . Full-time Full Students time G Faculty 93 107 71 201 82 133 UG 1981-1982 Full-time Full Students time UG G Faculty  . 3,308 3,641 2,893 2,150 1,535 1,818 1,762334 1,68258 2,21879 2,2411,538 1,360269 1,385913 1,65423 2,182359 1,718101 2,015272 1,675307 1,320192 2,269721 1,713117 1,508504 4,051470 2,080200 4,137 3,387 1,807 1,809 1,475 2,018 3,099 2,382 2,471 2,342 1,420 2,780 3,087 2,087 1,324 2,155 884 2,818 1,783 4,866 1,590 1,368 94 108 56 92 154 79 218 71 190 263 76 1,616 369 407 273 173 101 67 76 1,724 331 195 111  747 273 313 112 84 125 500 188 284 106 258 182 227 145 1,021 106 81 101 121 86 411 130 675 249 255 105 426 307 176 100 485 3,088 2,277 2,192 2,742 2,222 1,943 2,383 2,798 2,063 6,667 2,912 5,885 4,824 3,652 2,782 2,328 3,424 3,918 4,152 4,683 4,072 2,342 4,930 4,659 2,299 2,184 4,145 1,281 6,453 2,921 6,613 2,041 2,104 434 69 258 1,577 613 1.655 105 140 77 227 104 135 13 100 4'3 98  145 70 399 176 302 194 201 88 872 202 176 608 698 49S 79 230 253 80 1,579 429 309 147 2,131 228 871 350 124 737 167 153 311 130 139 217 492 859 327 520 291 265 422 289 127 90 366 121 264 120 127 200 129 184 174 123 95 110 130 302 108 253 100 141

SUPPLY AND DEMAND TABLE 11 {ContinuedJ 33 1975-1976 1981-1982 Full-time Full- Full-time Full Students time Students time SchoolUGG Faculty FIGG Faculty Purdue4,999960 297 6,262875 289 Rensselaer2,225405 122 2,751777 161 SUNY, Buffalo1,409268 79 3,394293 96 Stanford7831,313 141 1,2791,686 158 Texas3,041808 177 5,678874 169 Texas A&M4,215503 227 8,290686 327 Texas Tech.1,367186 81 2,531211 80 Univ. of Washington2,287496 180 2,610688 183 Virginia Tech.2,827469 206 4,892617 263 Washington State1,362191 137 2,935295 117 Wisconsin2,572615 189 4,818790 207 Worcester Polytechnic1,50377 71 1,968141 83 - Tota1112,51222,994 7,724 175,73227,677 8,466 NOTE: UG = undergraduate; G = graduate. SOURCES: Student enrollment: Engineenng and Technology Enrollments, Fall 1975 (New York: Engineering Manpower Commission, April 1976); Engineenng and Tech- nology Enrollments, Fall 1981 (New York: Engineering Manpower Commission, 1982~. Faculty: Engineenng College Research and Graduate Study, Engineenng Edu- cation, March 1977; Engineenng College Research and Graduate Study, Engineenng Education, March 1983. productivity and to handle the increased student loads without requir- ing new faculty. The increased use of computers and of television is often cited as offering promising possibilities in this respect. (The sec- tion below on "Educational Technology and Productivity" discusses this matter in detail. ~ The reply of engineering educators generally is that computers and television are already being rapidly incorporated and utilized, but they point to the loss of quality that Is occurring in much of the nation's engineering education delivery system primarily because of reduced student-faculty interaction.22 The increase in workload represented by the figures in Table 12 has had a number of consequences for engineering schools in addition to that of diminished student-faculty interaction. For some, the image of excessive overload has acted as a disincentive with regard to entering academic careers. In addition, there is a perception in some fields of engineering that opportunities for research are better in industry than in academia. This perception relates partly to the student overload and

34 ENGINEERING GRADUATE EDUCATION AND RESEARCH TABLE 12 Comparison of Engineering Student: Faculty Ratios, 1975-1976 and 1981-1982 1975-1976 1981-1982 Student enrollment, 51 large schools Full-time undergraduate 112,512 175,732 Full-time graduate 22,994 27,677 Total 135,506 203,409 Full-time faculty (professorial ranks) 7, 724 8,466 Raw student:faculty ratio 17.5 24.0 291 schools 286 schools Total enrollment Pull-time undergraduate 231,379 387,577 Full-time graduate 37,285 47,772 Total 268, 664 435,349 Percent of total, for 51 large schools 50.4 46.7 Estimated total faculty, all schools 1~,320 18,130 Estimated faculty needed to restore 1975-1976 ratio - 24,880 Estimated 1981-1982 faculty shortfall - 6, 750 partly to the inability of universities to provide sufficient funds to keep up with facilities needs, both for space and for equipment. These fac- tors tend to aggravate the problems that keep universities from obtain- ing their "fair share" of Ph.D. production. It has been argued that the current large enrollments in engineering and computer science constitute a "bubble" that will subside in the future. Past cycles in enrollments are cited in support of this view. And some have said that even if more faculty are needed, it is not necessary for them all to have Ph.D.s; this argument would tend to reduce the apparent need. Although these arguments have some validity, they will not solve the shortfall problem for the following reasons: · National needs are such that engineering enrollments of the future should be maintained at a substantially higher level than was charac- teristic of the 1970s, even though enrollments may subside somewhat from the current high levels. · Educators are strongly of the opinion that desirable student-fac- ulty interaction has been interfered with as a result of high student:

SUPPLY AND DEMAND 35 faculty ratios. Maintenance of student-faculty interaction is vital to quality education, and more faculty will be needed even if student numbers decline from present levels.22 · A portion of the new faculty members needed by the country can function appropriately without Ph.D.s, but the large majority of fac- ulty should be educated at the doctoral level. (This point is examined greater detail in Chapter 8, "The Engineering Faculty." ~ An additional factor affecting "need" is the anticipated near-term increase in the rate of faculty retirements. Engineering faculties expanded in the 1950s and 1960s, and many faculty members will be ready for retirement in the next 10 or 15 years. Table 13 shows the distribution of faculty in the professorial ranks by age, as provided by an "AAES/ASEE Survey."20 Since the survey included 168 schools that responded, the distribution has been extrapolated on a pro rata basis to include the entire estimated population of 18,130 faculty for 286 schools, derived from Table 12. An estimated number of about 3,600 faculty are in the 56 to 65 age group, and most of them can be expected to retire In the next 10 years. There are approximately 5,400 in the 46 to 55 age group. These faculty members will begin retiring in substantial numbers about 10 years from now, and probably more than half will have retired in 15 years. Thus, it can be expected that about 7,000 faculty will retire in the next 15 years, an average of 450 or so per year, probably rising from 300 per year in the near future to 600 per year toward the end of the period. Furthermore, there is the question of whether there is a net flow in, or out, between academia and industry. Data maintained by the Sunrey of Earned Doctorates {National Research Council) show the following: * 1979 Employer Business/ University or 1981 Employer Industry 4-year College . . . Busmess/Industry 21,997 927 University or 4-year College 259 13,565 Thus, of those who were employed in industry in 1981, 21,997 had been ~ industry two years earlier, and 927 had been in academia. Of * The numbers are based on a sampling survey, and the numbers given here are the weighted numbers, based on the distribution of the sample.

36 ENGINEERING GRADUATE EDUCATION AND RESEARCH TABLE 13 Age Distnbution of Engineering Faculty in Professorial Ranks ~. . Estimated Number of Number of Faculty in Faculty in Professorial Ranks Distribution Professorial Ranks Age jl68 schoolsJ {%) (286 schools) 70 31 0.3 54 66-70 239 2. 1 38 1 56-65 2,196 20.1 3,644 46-55 3,287 30.0 5,439 36-45 3,315 30.4 5,512 25-35 1,861 17.0 3,082 25 12 0.1 _ 18 Total 10,941 100.0 18,130 . _ SOURCE: J. Geils, "The Faculty Shortage: A Review of the 1981 AAES/ASEE Survey," Engineenng Education, November 1982, pp. 147-158. those who were in academia in 1981, 13,565 had been in academia two years earlier, and 259 had been In industry. From this, we would con- clude that the net flow had been outward from academia between 1979 and 1981, with the outflow rate 3.5 times as great as the inflow. More recent data suggest that the flows have become approximately balanced. In 1981 the Engineering College Faculty Shortage Project reported that 266 faculty went from academia to industry and 251 went the other way, based upon responses from engineering deans.20 In 1982, 227 went from academia to industry, and 252 went in the opposite direction.23 In the present study, the flows are assumed to be in balance. * There has been a great deal of discussion in recent years regarding the percentage of Ph.D. engmeenng students who are foreign nationals in the United States on temporary visas. In Table 9, it can be seen that this * A survey taken in 1983 showed 300 going from academia to industry and 497 going from industry to academia. See P. Doigan, "ASEE Survey of Engineering Faculty and Graduate Students, Fall 1983," EIIgineenng Education, October 1984. It would appear that the balance of flow has shifted in favor of academia since 1981, with a net flow to academia of 197. Nevertheless, the same study reported that the total unfilled faculty positions had increased to 1,570, from the 1982 level of 1,400 unfilled positions.J Whether a net flow to academia can be regarded as permanent or whether it should be viewed as a transient response to crisis is open to question. Thus, for the purpose of the present analysis, the flows are assumed tO be in balance.

SUPPLY AND DEMAND 37 percentage rose from about 14 percent in 1970 to 42 percent in 1983. During this same period, the absolute number of U.S. citizens in the doctoral pool dropped by more then half, frown 2,500 per year to about 1,170. iThe number of non-U.S. citizens on permanent visas has remained at about the level of 300 per year since 1977. ~ The Research Council surveys show that about half of all foreign engineering Ph.D.s expect to remain in this country, so they have constituted an increas- ingly significant portion of the engineering Ph.D. employment pool. Clearly, if these non-U.S. citizens had not been available, the engineer- ing schools of this country would have faced an even more aggravated recruitment problem than was the actual case. A vital policy matter, however, is whether this country should be forced to rely on the impor- tation of non-U.S. citizens in order to staff its engineering schools. It would appear that ways are needed to stimulate more U.S. citizens of high ability to enter the Ph.D. pool. Ph.D.s in Industry It is not possible to define the need for Ph.D.s in industry as clearly as it is to identify the need in academia. Nevertheless, in May 1984 an attempt was made to determine the general attitudes toward the Ph.D. by writing to top executives of 10 organizations known to be significant employers of engineers. All were asked if they perceived a trend toward increased recruitment of Ph.D. engineers and whether they had diffi- culty obtaining Ph.D.s of the quality needed; they were asked to pro- vide typical starting salaries for Ph.D. and B.S. engineers. Five of the 10 respondents stated that they did perceive an increasing trend in recruitment of Ph.D. engineers in their own organizations and in industry generally. The others responded that they saw no changes occurring in the demand for Ph.D.s, or that any trends they perceived were confined to certain special fields for which they had current need. Most respondents reported that they had no overall difficulty in recruit- ing Ph.D.s, although one did report having some difficulty and another said there is a shortage of U.S. citizens among Ph.D. engineers. Several remarked that they were anxious to recruit people of the very highest quality, and two mentioned that they seek graduates from the top dec- ile of the graduating cohort, either at the bachelor level or doctoral level. The 1984 average monthly starting salaries reported by these respon- dents were as follows:

38 ENGINEERING GRADUATE EDUCATION AND RESEARCH Type of Company Ph.D. B.S. Chemical and petroleum $3,284 $2,364 Computers, electrical manu facturing, aerospace, and others 3,271 2,240 Construction 2, 750 2,175 These figures, in general, tend to be slightly higher than those reported nationally by the College Placement Council {CPC) in March 1984 {see Table 14). It has been suggested that trends ~ staring salaries for new Ph.D.s, particularly as compared to starting salaries for new B.S. graduates, might give some insight into the demand for Ph.D.s by industrial TABLE 14 Average Monthly Salary Offers act New F.n~rint".erin~ Degree Candidates, 1965-1984 , ~_ c~ B.S. Candidates Year Chem. Civil Elec. Comp. Mech. Chem. Civil Doctoral Canadidates Elec. _~ Consumer Price Index Comp. Mech. 11967 = 100) $1'l66 $1,116 94.5 1,203 1,136 97.2 1,261 1,212 100.0 1,361 1,282 104.2 1,379 1,309 109.8 1965$ 642$ 618$ 641$ 635 $1,058 1966682658679670 1,102 1967733706728720 1,175 1968790750774768 1,247 1969849797826820 1,319 $1,128 1,197 1,234 1970902837869867 1,375 1,236 1,429 1,370 116.3 1971920850877881 1,395 1,102 1,388 1,278 121.3 1972928869888894 1,405 1,227 1,439 1,381 125.3 1973962908931927 1,438 1,298 1,508 1,418 133.0 19741,4029679861,001 1,550 1,426 1,551 1,479 147.7 19751,1961,0641,0811,122 1,645 1,382 1,550 1,624 161.2 19761,2791,1081,1551,197 1,743 1,597 1,693 1,687 170.5 19771,3891,1851,2451,286 1,882 1,625 1,811 1,777 181.5 19781,5131,2881,3671,404 2,074 1,830 1,974 2,030 195.3 19791,6421,4021,5201,536 2,231 2,260 2,221 2,160 217.7 19801,8011,5541,6901,703 2,451 2,089 2,534 2,436 247.0 19812,0301,7751,8821,908 2,745 2,523 2,793 2,659 272.3 19822,2561,9252,0642,098 3,019 2,851 3,099 2,959 288.6 19832,2281,8692,1282,096 3,130 2,936 3,200 3,150 299.3 19842,2731,8832,1752,173 3,201 3,195 3,247 3,162 303.5 {March) SOURCES: College Placement Council, Bethlehem, Pa. For Consumer Price Index: Monthly Labor Review, U.S. Dept. of Labor, March 1984, p. 82. Last value shown is for December 1983.

SUPPLY AND DEMAND 39 employers. Table 14 and Figures 8 through 16 provide ~fo~ation on this point. Average monthly salary offers to new engineering degree candidates at the bachelor's and doctor's levels for 1965 through March 1984 are given in Table 14. The table only shows offers for chemical engineering, civil engineering, electrical and computer engineering, and mech~ni- cal engineering because these are the only engineering categories reported by CPC at the doctoral level. The average dollar offers for these four fields are plotted in Fi~res 8 through 11, where it can be seen that the dollar differentials between bachelor's- and doctor's-level offers in each field remained approximately constant until about 1977. At that time the differentials began to increase, with bachelor' e-level offers increasing at about $140 to $150 per month per year {except for civil engineering, which increased about $115 per month per years, and doctoral offers increasing at about $210 to $220 per month per year. Figure 12 shows the differentials for all four fields from 1965 through 1984. In the years since 1980, civil engineering has increased its differ- ential the most. The effects of inflation tend to obscure the picture. Table 14 shows the behavior of the Consumer Price Index {CPU for the years finder S3000 o ~ >A o ~ i= S2000 :C ~ I&J he o ~ Z Cat ' o at: Slooo a Hi . _ DOCTORAL CANDIDATES ~ CHEH. ENGR. ) i/ SLOPE-S210/80/ OR // ~ r SLOPE _ SlS0/HO/YR  / ~ AS. CANDIDATES ( CHEH. ENGR. ) _ / i i - - lets 19?0 19?5 YEAR 1980 1985 FIGURE 8 Average monthly salary offers to new chemical engineering degree candi- dates. SOURCE: Data from College Placement Council.

40 ENGINEERING GRADUATE EDUCATION AND RESEARCH S3000 V, to o ~ >- o t... ~ fir Z ~ to to C:' ' o ~S1000 o DOCTORAL CANDIDATES / ( CIVIL ENGR. ) r JO SLOPE- ~' S220/1910/ YR ~  ,1 ~ Y ~ /-SLOPE-SllS /~0 ~ YR B.S. CANDIDATES ( CIVIL ENGR. ) 1965 1970 1975 YEAR 1980 1985 FIGURE 9 Average monthly salary offers to new civil engineering degree candidates. SOURCE: Data from College Placement Council. / S3000 DOCTORAL CANDIDATES ( ELEC. ~ COUP. ENGR. ) ~ ~- | SLOPE -Sllo/Ho/YR o ~SLOPE _ S220/~0/YR~ ~ z S2000 .= ~ ~ Oo / ~ / B.S. CANDIDATES ~( ELEC. ~ COUP. ENGR. ) a 1965 1970 1975 YEAR 1980 1985 FIGURE 10 Average monthly salary offers lo new electrical and computer engineering decree candidates. SOURCE: Data from College Placement Council.

SUPPLY AND DEMAND S3000 _ z S2000 ~ ' To 41 DOCTORAL CANDIDATES ( HECH. ENGR- ) 7 SLOPE _ Sat S ~ / YR ~/ - ~B.S. CANDIDATES ~ ( HECH. ENGR. ) 1 1 1 1965 1970 1975 YEAR / SLOPE - S140 ~ HO Ye it' 1980 1985 FIGURE 11 Average monthly salary offers to new mechanical engineering degree candidates. SOURCE: Data from College Placement Council. consideration. All of the salary offers reported by CPC have been con- verted to constant 1967 dollars by use of the CPI, and replotted in Figures 13 through 16. These figures can be interpreted as the real value of engineering degrees to the market, particularly with regard to the relative market values of B.S. and Ph.D. degrees. For all four engineering fields, there was an increase in the value of both bachelor's and doctor's degrees during the late 1960s, which was a period of intense recruiting for the space program. Subsequent years show a decline in value of the B.S. degree, followed by relative stability. A marked exception to this observation is civil engineering, which shows a gradual decline throughout almost the entire period. The most pronounced effect to be observed ~ Figures 13 through 16 is the large decline in the relative value of the doctor's degree from the late 1960s until about 1975. The same figures show a gradual increase in the value of the doctor's degree since 1975, these trends being most apparent for civil engineering and for electrical and computer engineer- ing. In the years since 1980 especially, an increase in value is apparent  for the doctor's degree relative to the bachelor's for all four fields. Not much can be concluded from these figures that is not already well known. The early 1970s, which was the period of marked decline

42 ENGINEERING GRADUATE EDUCATION AND RESEARCH S1400 S1200 S800 S200 To CIVIL EHGR. _,1 ELEC. ~ COUP ~ ENGR ///; CHEM. ENGR. 1 l 1 1965 1970 1975 YEAR 1980 1985 FIGURE 12 Comparison of average monthly salary offers to new B.S. and doctoral engineering degree candidates, for four engineering disciplines. in the relative value of the doctorate, coincides with what was identi- fied earlier as the "retrenchment" period; it was a period of relative unpopularity for engineering education generally. Many engineering schools struggled with the problems of low enrollment {see Figure 4), and a few engineering schools went out of business. It is not surprising that the '`real" value of engineering starting salaries declined. The years since 1975 have seen surging enrollments and popularity for engi- neering. Perhaps the most surprising thing about Figures 13 through 16 is that they do not show a greater increase during recent years in the value of the B.S. degree in constant 1967 dollars. Given the publicity and recmitment activity for engineers, one would have expected a rising level of constant-dollar starting salaries. The fact that the doc- toral offers do exhibit just such a rising trend in the last four years implies that recruiters are increasing their emphasis on doctoral degrees. However, the trends are too tenuous to draw any marked con- clusions about scarcity. Faculty salaries at many,,~iversities have been rising in recent years, but overall, salaries for assistant professors still lag significantly behind

SUPPLY AND DEMAND S1500 V, to z Cat I_ S1000 Z. o o ~ o V, _ fir ~ ~ - o Z >A ~ TIC Z SSOO A 43 DOCTORAL CANDIDATES CHEN. ENGR. ) \ B.S. CANDIDATES ( CHEN. ENGR. ) 1965 1970 1975 1980 1985 YEAR FIGURE 13 Average monthly salary offers to new chemical engineering degree candi- dates (data adjusted by Consumer Price Index to reflect constant 1967 dollars!. SlSOO ~ SSBO He to :x _ ~_ C:} _ _ 1 .. 1 1 1 1 1 o '' DOCTORAL CAND IDATES -\ ( CIVIL ENGR ) _ L B.S. CANDIDATES - ( CIVIL ENGR. ) / 1965 1970 1975 YEAR 1980 1985 FIGURE 14 Average monthly salary offers to new civil engineering degree candidates (data adjusted by Consumer Price Index to reflect constant 1967 dollars I.

44 ENGINEERING GRADUATE EDUCATION AND RESEARCH SlSOO _ C) z Cal i" A: o ~ ,^ SI0OO ~ J lo lo ~ o .~ _ o Z Cal ~ - : Z - ~ lo - : ~ V} _ SSOO I _ ~ DOCTORAL CANDIDATES ( ELEC. 8 COUP. ENGR. ) - B.S. CANDIDATES ( ELEC. ~ COUP. EHGR. ) 1965 1970 1975 YEAR 1980 1985 FIGURE 15 Average monthly salary offers to new electrical and computer engineering degree candidates [data adjusted by Consumer Price Index to reflect constant 1967 dollars). SlSOO '~ 2  S o SSOO to _ Am_ ,w DOCTORAL CANDIDATES ~ ( HECH. ENGR. ) - \~N B.S. CANDIDATES ( HECH. ENGR. ) - I I I 1 1 less 1970 1975 YEAR 1980 1981  FIGURE 16 Average monthly salary offers to new mechanical engineering degree candidates (data adjusted by Consumer Price Index to reflect constant 1967 dollars).

SUPPLY AND DEMAND 64,000 60,000 56,000 s2 000 48,000 6 44,000 6 at 6 40,000 36,000 32,000 28,000 24,000 20,000 16,000 45 Industry.Supervisor _~ _ _ ~;~;= - - _ ~Associate Prof. Assistant Prof. _ ~Indus~ry-Nonsupervisor 0 3 6 9 12 15 18 YEARS SINCE BACCALAUREATE DEGREE 21 24 27 30 33 and Over FIGURE 17 Comparison of academe-industry engineering Ph.D. salaries fall profes- sorial salaries adjusted to 11-month basis). SOURCE: Data from Engineering Manpower Commission.  salaries for Ph.D. engineers in industry, even when adjusted to a full- year {[l-month) basis. {See Figure 17.) In fact, salaries lag at all aca- demic levels, since it may be argued that full professors should be compared with industry supervisory Ph. D.-holders and since some full professors are recruited into these positions. Figure 17 shows that the salaries of full professors are intermediate between those of nonsuper- visors and supervisors in industry. Even these comparisons may be deceptive, however, because they involve median salaries. A crucial point is that for tenure-track positions schools typically attempt to hire the best doctoral engineers available. These same recruits can some- times command significantly higher-th~n-median salaries in industry, so that the real disparity may be greater then the chart indicates.

46 ENGINEERING GRADUATE EDUCATION AND RESEARCH It is worth noting that engineers by no means have a monopoly on the industrial need for doctoral-level personnel. The Research Council's Survey of Eamed Doctorates shows that the number of Ph.D. s from the physical sciences that enter Industry is about the same as it is for engineering. is For the 1982 graduates, 854 Ph.D.s from engineering had definite postgraduation plans for industrial employment, while 869 from the fields of chemistry, physics, and mathematics had similar plans {chemistry: 624; physics: 184; mathematics: 61~. The same reporter compares the postgraduation plans of the 1980 graduates with their actual status one year later: Of those who had dlefinite postgraduation employment plans in 1980 for going into indus- try, 97 percent were actually in industrial employment one year later, whether they were engineering graduates or physical sciences gradu- ates. However, of those who were "seeking employment" in 1980, their actual status one year later was as follows: Physical EngineenngSciences Actual Status, 1981Ph.D.sPla.D.s Academic employment14.0%7.8% Industrial employment82.944.3 Government employment0.76.4 Other employment2.44.6 Postdoctoral study0.022.4 Not employed0.014.5 100.0100.0 The conclusion to be drawn is that industry hires about equal propor- tions of Ph.D.s ~ engineering and in the physical sciences but that engineers appear to have a slight edge in attractiveness to industry; as shown above, O percent of the engineers seeking employment in 1980 were, one year later, in either the "riot employed" or "postdoctoral study" category. Further, although one could not declare that there is a shortage of engineering Ph.D.s for industry, there does not appear to be a surplus either. Increasing the Supply of Ph.D.s It seems imperative for the nation to increase its supply of Ph.D. engineers. Table 3 and Figure 7 imply that this is indeed happening, with an increase from about 3,000 in 1983 to nearly 4,000 expected in 1988. However, as was mentioned, the renumber of U.S. citizens in this

SUPPLY AND DEMAND pool will remain substantially below the levels they represented 47 in the 1970s unless actions are taken to encourage more of them to enter doctoral study. Certain federal agencies have attempted to make projections of the future need for engineers. As mentioned above, the National Science Foundation published a report in 1983 that projected possible shortages in computer specialties, aeronautical/astronautical engineering, and electrical/electronic engineering, but the study did not assess the situ- ation for persons with advanced degrees separately.2464 In a separate study, also in 1983, the Department of Energy {DOE) assessed the adequacy of the supply of engineers and scientists for energy-related employment during the 1983-1988 period. The report concluded, relative to energy fields: "The potential for labor shortages during 1983-1988 is expected to be greatest at the Ph.D. level."25 An increase of 20 percent in the supply of engineering Ph.D.s was projected by DOE, relative to 1982. Under the assumption that approximately half of the foreign nationals on temporary visas would remain in the United States, significant scarcities were projected for petroleum engi- neering and the earth/environmental sciences. However, under the alternate assumption that none of the students on temporary visas would remain in the United States, scarcities of 10 percent or more were projected for mathematics/computer science, chemistry, earth/ environmental science, chemical engineering, nuclear engineering, petroleum engineering, mining engineering, and materials science.25 The matters of "shortage" and "supply and demand" are controver- sial. Many have pointed out that supply and demand necessarily become balanced at a price the market is willing to pay, so that one cannot properly speak of "shortages." On the other hand, engineering deans declare that the distress they have experienced in their inability to hire enough faculty for their needs is sufficient evidence of a "short- age." In reply, it has been said that academic employers need only improve salaries and working conditions sufficiently so that they can get their fair share of the existing Ph.D. production; if this were to occur, then an increase in total output would not be needed. If we were concerned only with total numbers, the foregoing consid- erations would have considerable force. However, as has been seen, not enough of the brightest of U.S. citizens are motivated to enter the doctoral pool, and there are other ways in which the Ph.D. "market" deviates from an ideal one. The delay in the response from market stimulus to market response {five years or more for a Ph.D. ~ is enough in itself to interfere with an ideal response. An additional factor is that even if universities raise their salaries for engineering Ph.D.s to com

48 ENGINEERING GRADUATE EDUCATION AND RESEARCH petitive levels-an outcome strongly recommended herein the prob- lem is not wholly solved, because the four or five years that must be spent during the period of graduate study with a marginal or submargi- nal income is enough to deter many new B.S. graduates from going on to graduate school. Improvement in graduate stipends is needed, at least to a level of 50 percent of what a new graduate could cam by going to industry instead of to granulate school. The level of 50 percent has become recognized by custom as a reasonable balance between the giving up of a salary of $26,000 or so { 1984 levels, and the opportunity to be paid to attend school full-time. home of the fellowships offered by federal agencies in 1984 were at the $13,000 to $14,000 level. ~ There is a further disincentive for continuing on to graduate school that produces some concern. It is caused by the fact that many young people finish undergraduate school with large loan obligations and may wish to enter employment as soon as possible to begin reducing their debts. A solution to this disincentive might be to "forgive" such a loan if the individual goes on to complete a Ph.D. While considering the disincentive of a financial burden carried by a student from undergraduate to graduate status, one should also con- sider the desirability of doctoral loans that are forgivable if the recipient enters academic employment for a specified number of years. There has been success in the past with forgivable loans of this nature, and such programs have been broadly supported because they focus financial aid on the location of greatest concern academic employment. However, these arrangements might not be as attractive to students as might be supposed. From students' point of view, boding themselves to an obli- gation of academic employment several years in the future may not necessarily appear to be in their own best interest. No doubt some mix of forgivable loans and outright fellowships will prove to be most advantageous. There is a question, too, in considering reliance upon natural market forces, about whether the Coventry can afford to wait while the market works itself out to a condition of balance, especially in view of the present disincentives for attending graduate school. Mary actions on many fronts are needed: universities must improve faculty salaries as well as their base of facilities, equipment, and support; industry needs to become involved in many ways, some of them financial; and a major fellowship program is needed to draw more of the top decile of B.S. graduates into doctoral study. One of the principal advantages of a fellowship program is that it shortens the time required to earn a doc- torate, because students can attend school full time without needing to be employed in part-time jobs. Thus, the supply con be increased more

SUPPLY AND DEMAND 49 quickly then by relying only on natural market forces. Fellowships can also be used to stimulate the entry of a greater portion of the top decile of U.S. citizens and permanent residents and thus lessen the county's dependence on importation of Ph.D. talent. If, for example, the nation wished 100 percent of its Ph.D. engineers to be U.S. citizens {or per~ua- nent residents), then an increase of 1,560 citizen Ph.D.s would be needed by 1988, according to the figures in Table 3. If, on the other hand, we were to return to the situation in 1972 when only 20 percent of the Ph.D.s were foreign nationals, then an increase of 700 to 800 citizen Ph.D.s would be needed in the projected output of 1988. {Since the current proportion of foreign Ph.D.s is 40 percent, and, since it is believed about half of the foreign Ph.D.s plan to remain in this con n try, a reduction of the foreign fraction to 20 percent would produce ~ condi- tion of approximate balance with the fraction of foreign Ph.D.s who presently leave the Coventry. J In order to stimulate an increase of 700 to 800 more citizen Ph.D.s by 1988, new fellowships numbering substantially more than 700 or 800 perhaps 1,000 will be needed. This would allow for the attri- tion of those who do not complete their Ph.D. programs and also would make some allowance for those who would qualify for fellowships but who might have gone to doctoral study anyway, with or without a fellowship. A fellowship program of this type undoubtedly will have a combina- tion of federal and industrial support Within the last two years, for example, the American Electronics Association {AEA) has established approximately 100 new Ph.D. fellowships, with a goal of 200. In approximately five years' time this program alone will increase Ph.D. production by 100 or so, but 100 new fellowships would have to be added each year for the next four years, lentil a total of 500 students is in the "pipeline," for the incremental output to be sustained in subse- quent years at the level of 100 per year. In this program, a stipend of $10,000 per year plus tuition is offered to the fellow, and the student is expected to supplement this with employment as a research assistant or teaching assistant, plus at least one summer in industry. The ALA fellowships are of the "forgivable loan" type. The ALA program is focused exclusively on electrical engineers and computer scientists. However, a recent informal survey by Hewlett- Packard produced the estimate that industry presently is offering approximately 200 to 300 new fellowship positions per year, mostly for engineering Ph.D.s, covering all disciplines. Most of these are of long standing and thus are part of the "constant" base. Perhaps only 100 or , . . so are ot recent origin.

50 ENGINEERING GRADUATE EDUCATION AND RESEARCH The National Science Foundation provided the information that, of its 450 doctoral fellowship awards in 1982-1983, approximately 100 went to engineers. The NSF estimated that perhaps 150 new fellow- ships each year are offered by all other federal agencies combined, most of these going to engineers. The present stipend offered by NSF is $6,900 to the student, plus tuition and fees. The stipend was raised to $8,100 in the fall of 1984. Stipends being paid by some of the other federal agencies range as high as $12,500 to $14,500, with the higher stipends going to students in the third or fourth year of graduate study. Typically, in the federal fellowships, students are expected to engage in full-time study throughout the 12-month year. The ALA fellowships are provided for four years. NSF offers three years of support, while the programs of the other federal agencies gener- ally are based on support for four years. A few of the industrially spon- sored fellowships cover up to five years of support. If the fellowship stipend is 50 percent of starting salaries for B.S. engineers, then in 1984, for example, the stipend would have been $13,000 in the first year, and should be increased by a modest amount each year the student is ~ the program. With 1,000 new starts each year, and assuming a four-year program with some attrition each year, there might be 3,500 students actually in the program by the fourth year, when the program is fully under way. If the average stipend for all 3,500 students is $14,000 {1984 dollarsJ, and if there is an accompany- ing grant to the institutions of up to $6,000 for tuition and fees, the cost per year would be in the range of $60 million to $70 million, divided between the federal government and industry. A fellowship program provides only a part of the answer, giving an initial stimulus to prospective students and a "bridging" over the prob- lem of financial support in graduate school. A permanent solution requires that universities take steps to make academic life more attrac- tive then it has been recently by increasing salaries to competitive levels and reducing current overloads. Also, they need to provide mod- em laboratory space end equipment for both instructional and research purposes, so that students and faculty can have an opportunity to work with facilities that reasonably represent the state of the art. The National Commission on Student Financial Assistance, in a 1983 report to the President and the Congress, recommended a sub- stantial increase in the number of federally supported science and engi- neering fellowships. The commission also recommended substantial increases in funds for improving and modernizing university laborato- ries, equipment, and instrumentation.61 For its part, industry can help by continuing to increase its support of

SUPPLY AND DEMAND 51 doctoral fellowships, giving state-of-the-art equipment, providing fan cling for laboratory renovations, increasing its support for university research, and providing grants to help support departmental expense such as that for computers, travel, and student projects. Federal and state governments can help industry in this regard by allowing gener- ous tax deductions for contributions that help stimulate U.S. students to study graduate engineering. Educational Technology and Productivity It was mentioned earlier that developments in computers are signifi- cantly changing the manner in which engineering is practiced. Com- puter developments are also changing the way in which engineering education is practiced. Simultaneously, the instructional use of televi- sion has affected education with regard to both on-campus and off- campus use. It is not the purpose of this report to review these and related develop- ments in depth. That task has already been well carried out in the study Educational Technology in Engineenng, prepared for the National Academy of Engineering by Lionel V. Baldwin and Kenneth S. Down.26 However, our perception of need for additional faculty is heavily influ- enced by our vision of the prospects for improved educational produc- tivity through the use of educational technology. If new technologies can somehow permit handling larger numbers of students with the current number of faculty and, of course, with no loss of quality, then an increase in engineering doctoral output would not be needed. The simplest and most obvious measure of productivity is the stu- dent:faculty ratio, and the simplest way to increase this ratio is by grouping students in large classes, with the majority of student-faculty contact provided by teaching assistants iTAs) . The technique is widely used by universities in handling large numbers of students in classes like physics, chemistry, and biology, but it is bitterly criticized by students {and their parents) because it deprives them of personal con- tact with the regular faculty. One of the earliest uses of instn~ctional TV [ITV) lectures by videotape, with the professors presumably thus made more available for personal contact came essentially under the same criticism: the hundreds of students involved could not gain satis- factory contact with the professors in charge and so were shunted off to TAs anyway. Even though it has been shown that reaming is not impaired in such courses, students exhibited strong objections to this kind of TV use and sometimes demonstrated their objections through declining attendance at the TV lectures.26 However, in subsequent

52 ENGINEERING GRAD RATE ED UCATION AND RESEARCH years more imaginative ways have been found to use ITV than merely producing the equivalent of large lecture classes. Probably the most effective of these has been in the form of supplementary instructional modules. Videotaped supplementary instructional modules have been used in a variety of ways: for lecture review, making up missed lectures, class- room demonstrations, simulations, presentation of laboratory proce- dures, and self-paced instruction. The tapes are usually made available in individual study carrels at times that are convenient to the students. But even in this case students seemed to have demonstrated a prefer- ence for live tutors: in the NAB study cited above, one university found that only about 25 percent of its freshmen reported using supplemental lecture videotapes, although the system was wired into every dormi- tory room. Live tutors at the dormitories in the evening, on the other hand, drew an 80 percent response.26 Nevertheless, ITV has been widely used for the purposes described above, in spite of heavy initial costs, principally for beginning-level courses in which the content is not subject to rapid obsolescence. However, in more advanced courses the need for regular revision of the material makes about as much demand on a professor's time as do the more conventional methods, thus offering no productivity gain. The use of ITV for off-campus instruction brought something truly new to the educational scene, beginning at the University of Rhode Island in 1961. By 1980, 37 U.S. universities had adopted ITV, either "live" or by videotape, for engineering graduate study, for both credit and noncredit. In the typical "live" TV mode, students at remote locations {usually at industrial sites) participate via TV in a class as it is being given simultaneously on campus. The cost-effectiveness comes about by saving time for off-campus students, who need not leave their places of employment to participate. However, there are extra administrative costs associated with the TV system: a camera operator must be hired; the TV system has maintenance costs; an operator must be in the system control room {as required by the Federal Communications Commission); provision usually is made for a "talk-back" system, probably through leased telephone lines; there must be a "courier" to carry homework and examinations back and forth; and additional office personnel are required to coordinate the system. Instruction by videotape resembles that by simultaneous "live" tele- cast, with the exception, of course, that the students cannot ask ques- tions of the professor during class. To compensate for this, proctors are usually provided by the industrial employer who receives the tapes.

SUPPLY AND DEMAND 53 Frequently the proctors are employees who took the course previously. If the courses change rapidly, which is the nature of graduate courses, such proctors can quickly lose touch with the course material. Never- theless, some universities have approved videotaped courses for aca- demic credit, prominent among them, Stanford University.26 Although the cost of maintenance of a TV transmitter is obviously avoided, there are special costs associated with videotape systems. Administrative costs are incurred in supervising and coordinating the systems, because tapes are constantly being sent to and received from many locations, generally by mail. Provision must be made, also, for sending and receiving homework and examinations. The delays associ- ated with receiving such student work, correcting it, and sending it back have been the most troublesome aspect of using videotape sys- tems. It is hoped that future reliance on transmission by satellite might alleviate such problems, but it has been estimated that satellite rental fees may range from $100 to $1,000 per hour, and satellite time must be provided for transmission not only of lecture material but also for stu- dent work if the present delays are to be overcome. The recently organized National Technological University iNTU) began offering televised M.S. programs to a rational audience in the fall of 1984. During the first year of operation, videotapes are being used, but it is expected that televised courses via satellite will be available in 1985. Long-term plans call for 80 graduate-level engineering courses to be offered per term, with approximately 9,000 students enrolled. The courses will originate from 18 member universities and will be distrib- uted nationwide, but the degrees will be conferred by NTU. Courses can tee received anywhere in the country once suitable "downlinks" to receive satellite signals have been installed at the receiving locations. Electronic mail and facsimile transmission will also be provided via satellite. A three-unit course will cost the student $1,000 to $1,400, with $600 to $1,000 of this going to the originating university and $75 to the instructor teaching the course.27 Computers are becoming ubiquitous in engineering primarily because they permit us to do things which were not possible before. It has become virtually impossible to design very large scale integrated circuits without computers, and structural analysis has been com- pletely revolutionized by finite element analysis. New process plants are controlled by units that have computers at their hearts, and all of manufacturing is being revolutionized by robotics ~d computer-con- trolled methods. It is difficult to find any phase of engineering that is not being overturned by computer technology. The advent of interactive computer graphics has had an especially

54 ENGINEERING GRADUATE EDUCATION AND RESEARCH strong and beneficial impact on engineering education. It has been pointed out, for example, that computer graphics is effective in engineering education because it emphasizes intuition rather then exact calculations. For years, engineering students used computers simply to get answers expressed to 10 decimal places. In order to understand the underlying relationships, the student generally had to print a number of solutions during orate computer run and then try to interpret the tables of numbers on the crude plots from the line printer.... The pedagogical significance of changing with a light pen the location of a single charged particle in an electrical field with other charged particles, and watching all the field lines move as if they were rubber bards cannot be overstated! [Ref. 26] Even as it becomes apparent that computers have deeply enriched engineering education, it also becomes clear that engineering educa- tion cannot rely entirely, or even extensively, upon prepackaged com- puter programs for educational purposes. Although it may be true that much of engineering in industry will utilize such programs, an educa- tional curriculum relying excessively upon packaged programs will inculcate a "button-pushing" mentality on the part of the students and ill equip them to face new situations. Fundamental theory and mathe- matics must still be taught and learned, with computers interlaced to provide pedagogical improvements where appropriate. Unfortunately, instead of lowering costs, computers have tended to increase them. The NAE study mentioned earlier concluded: Today, few people seriously consider lowering costs an argument for computing in instruction. The early literature abounds with cost-effectiveness discus- sions, but any honest comparison of computerized teaching costs with con- ventional teaching costs per hour are disappointing.... University-based advo- cates generally employ "anyhow" accounting-"we are going to do it any- how" when discussing costs. [Ref. 26] One aspect of the cost of computers that has surprised and dismayed many engineering schools is that associated with technical support personnel and software maintenance. In the days when centralized computer centers represented the way business was done, the technical support personnel resided principally in the centers. But as minicom- puters have proliferated, increased in power, and decreased in cost, computers are found everywhere, along with a bewildering variety of software systems. Individual academic departments are now finding that they need pellllanent support staff to manage these systems, par- ticularly as networking enters the picture. Otherwise, the job falls on

SUPPLY AND DEMAND 55 the shoulders of faculty members who are already overworked, com- pounding an already difficult situation. Nevertheless, there are cases in which computers have aided instruc- tion in a cost-effective sense, in a fashion analogous to that of video- tape. Many beginning-level courses are taught by self-paced instruc- tion; the student "contracts" to master certain modules of subject matter in a certain period of time. The student, going at his or her own pace, interacts with a computerized instructional module that provides pedagogical material selected in accordance with the student's rate of progress. Proctors are available to answer questions. When the material is supposedly mastered, the student takes a test from the proctor, which validates command of the material. Courses in calculus, statis- tics, elementary accounting, computer programming, and journalism have all been taught by such methods, or very similar ones.26 Findings and Recommendations 1. The nation can probably look forward to approximately 3,800 to 4,000 engineering Ph.D.s per year by 1988. Approximately 40 percent of these Ph.D.s are expected to be foreign nationals on temporary visas. 2. There has been little variation in the GRE scores of engineering graduate school applicants during the past decade. Engineering appli- cants consistently rank near the top in scores on the "quantitative" GRE, and consistently near the bottom in scores on the "verbal" GRE, when compared with applicants in other disciplines. 3. About one-third of new engineering Ph.D.s have entered aca- demic employment in recent years. To maintain that fraction in future years, universities should take steps to make academic life more attrac- tive than it has been recently for engineering faculty, in all ranks. The number of Ph.D. s available each year for academic employment during the next five years is expected to average only 100 or so more per year than was the case during the 1970s. 4. The percentage of non-U.S. citizens on temporary visas among engineering doctoral graduates has increased from 18.5 percent to 42.1 percent between 1973 and 1983. It is believed that about half of these graduates plan to stay in this country after graduation. In recent years, if there had been fewer foreign students in the employment pool, the difficulty for U.S. universities in obtaining engineering faculty would have been much more severe than it was. As a matter of national policy, it is questionable whether the United States should rely to such a degree upon the importation of Ph.D. talent.

56 ENGINEERING GRADUATE EDUCATION AND RESEARCH S. The workload for U.S. engineering schools, as measured by stu- dent:faculty ratios, increased by about 37 percent between 1976 and 1982. To keep even with this growth would have required about 6,700 more faculty in 1981-1982 than actually existed (24,800 instead of the actual 18,1001. 6. The image of excessive student overload has acted as a disincen- tive to some for entering academic careers. There is also a perception that opportunities for participating at the research frontier are dimin- ishing in academic institutions, partly because of the student overload and partly because of the inability of universities to provide sufficient funds to keep up with facilities needs, both for space and equipment. These factors tend to aggravate the problems of universities in obtain- ing their "fair share" of Ph.D. production. 7. The view has sometimes been expressed that high engineering enrollments are a passing phenomenon and, in any event, that engi- neering schools could handle high enrollments by increasing their productivity and by hiring more non-Ph.D. faculty. The co,~nterargu- ments are these: a. Enrollments of the future may subside somewhat from the cur rent high levels but will be at a substantially higher level than was characteristic of the 1970s. b. Present student:faculty ratios are too high and interfere with stu- dent-faculty interaction; maintenance of a high level of such interac- tion is vital to a quality education. c. A portion of the new faculty members needed by the country can function appropriately without Ph.D.s, but the large majority of fac- ulty should be educated at the doctoral level. 8. An estimated 3,600 engineering faculty are in the 56 to 65 age group, and an estimated 5,400 are in the 46 to 55 age group. Of these 9,000 faculty, perhaps 7,000 will retire in the next 15 years. 9. The flow of engineering faculty to industry is assumed to be approximately in balance with the flow in the opposite direction. 10. Industry employs Ph.D.s from many physical science fields as well as from engineering. However, engineering Ph.D.s seem to have a better chance for industrial employment. The data do not demonstrate that there is a shortage of engineering Ph.D.s for industry, but they do suggest that there is no surplus readily available for academia. 11. The supply of engineering Ph.D.s for academic employment is short enough that universities experience distress in faculty recruiting,

SUPPLY AND DEMAND 57 resulting in approximately 1,400 unfilled faculty positions in 1982 nationwide, and 1,570 unfilled positions for 1983. 12. In order to improve the faculty situation for engineering schools. several actions are necessary: a. The perception of academic life must be improved: universities must reduce the current high workloads, improve salary levels to competitive levels, and provide state-of-the-art facilities for instruc- tion and research; b. The number of doctoral fellowships should be increased in order to increase the proportion of U.S. citizens from the top decile of their graduating classes who enter doctoral study. About 1,000 never "starts" should be available per year, with stipends at least equal to 50 percent of the average starting salaries of graduates going directly to industry. Industry and government should work together in pro- viding this program. The total cost per year would be in the range of $60 million to $70 million for the nation. c. Industry, in addition to providing fellowships, should increase its financial support for engineering education, giving state-of-the-art equipment, providing funding for laboratory renovations, increasing its support for university research, and providing grants to help sup- port departmental expense such as for computers, travel, and student projects. Federal and state governments can help by allowing gener- ous tax deductions. 13. New developments in educational technology, principally involving computers and television, can be of major assistance in improving the quality and versatility of engineering education. Cost savings from such developments are not likely, however, and produc- tivity improvements in the conventional sense of large student: faculty ratios have not so far materialized except at a cost to program quality.