Women: Their Underrepresentation and Career Differentials in Science and Engineering: Proceedings of a Conference (1987)

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.

WOMEN GRADUATE STUDENTS: A LITERATURE REVIEW AND SYNTHESIS Ulli S. Hornig Introduction The purpose of this paper is to review the literature on women graduate students in science and engineering (S/E), focusing on the period since l970, in an attempt to identify major trends in women's participation and the major factors that are responsible both for the observed changes and for certain differences from the male pattern of participation. At the beginning of this period, the concern of most studies of women in graduate programs focused on equity issues, with secondary attention to the social utility of advanced education for women. By the mid-l980s, however, the declining interest of white men in S/E training and careers has added more specific public-policy concerns about the future adequacy of the S/E human resource pool to many investigations of women's participation. A similar shift of focus, from the earlier emphasis on women's attitudes and personal judgments toward a more dispassionate examination of data on actual outcomes, appears to be still in progress. The dominant theme remains concern with what is variously described as " underrepresentation," "low participation," or "lack of interest" by women in S/E—despite the fact that certainly by the end of this period the number of new women S/E Ph.D.s produced each year is well over three times that of humanities degrees—5,ll9 total doctorates in the sciences in l985, compared to a total of l,489 in the humanities (Coyle, l986:5l). Additionally, marital and parental status are often invoked to explain some of the gender differences in graduate study, and in later career outcomes, without much foundaton in empirical findings. Sex differences vary markedly among fields, and it is therefore essential to analyze most of the issues separately for the various disciplines to gain an understanding of the actual situation. Unfor- tunately, many empirical studies account inadequately (if at all) for field differences and thus tend to generalize their findings beyond the level sustainable by the underlying data. In particular, very signifi- cant differences exist in such institutional structures as financial aid mechanisms for different disciplines, and these account for differ- ent patterns and timing of graduate study in, say, physics, engineer- ing, life sciences, and psychology. These underlying field differences affect both men and women; since the distribution of students by sex l03

varies among these fields, a study that compares, for example, the timing of graduate study in "the natural sciences" by sex would produce an apparent sex difference that disappears when field-specific data are used. The problems broadly referred to as "career choice" are tradition- ally of great interest in women's education. Properly speaking, they are no longer significant at the graduate level if the meaning is taken to be "field choice"; that choice is typically made several years earlier in the case of S/E careers. What is relevant for graduate education is the level of career aspirations in the field already chosen. Traditional interpretations of women's participation in S/E suggested, well into the l970s, that the length and rigor of graduate study in these fields were poorly suited to women's life course (even though more traditionally feminine fields characteristically require much longer graduate study) and that women's aspirations were more cultural than professional, in contrast, the underlying assumption of the present discussion is only that women, like men, undertake graduate S/E study (and probably all graduate and professional study) as career preparation rather than simply as a form of intellectual enrichment. Minority representation in S/E is generally small except for high Asian participation in engineering, physical, and biological sciences, and women of the various subgroups tend to follow a field distribution pattern similar to that of majority women. As a group, however, and again excepting Asians, minority women are more likely than white women to obtain Ph.D.s compared to similar men. Although these statistical patterns have been evident for some years, empirical studies that ad- dress combined race and sex differences have not been carried out. It should be noted that trends in minority participation are more diffi- cult to trace than those for women, since separate data for minorities were not collected before the mid-l970s; in many instances it was illegal to collect such data. The year l970 serves as an important reference point, being the last in which it was legal for institutions to maintain different pol- icies by sex with regard to admissions criteria, financial aid awards, residential requirements, and so on. Revised Order Number 4 in December l97l inaugurated the era of legally mandated equal opportunity for women in higher education, and the Higher Education Affirmative Action Guidelines issued in October l972 produced many of the changes documented here. The Statistical Outlines Graduate Enrollments and Degrees Since 1970 The basic statistical framework that outlines the enormous changes in women's S/E participation in the past l5 years consists of two de- tailed surveys, one of graduate enrollments in full-time and part-time programs at doctorate-granting and other institutions carried out by the National Science Foundation (NSF) and a second that is an essen- tially complete annual census of doctorate recipients carried out for l04

NSF and other federal agencies by the National Research Council. Information in both surveys is disaggregated by sex, race/ethnicity, citizenship, and discipline; and special tapes and tabulations are available. Data from these surveys continue to show some progressive underrepresentation of women among S/E graduates in most fields at progressively higher degree levels in the mid-l980s, although the disparities relative to men have greatly diminished since l970. Thus, women's share of baccalaureate S/E degrees—the necessary precursors of graduate study—rose from 26 percent in l970 to almost 38 percent in l983, representing a 68 percent growth rate for women compared to a 2 percent decline for men during this period. The number of women earning S/E master's degrees rose by 99 percent in that time, and the number of those earning doctorates had increased l8l percent by l984 (NSF, l986:26). Although women's participation is most often expressed as a percentage of total degrees, probably partly because that is a relevant consideration in determining share of employment, the use of such proportions for analytical purposes is frequently misleading because men's participation has also changed rapidly during this period. The rate at which S/E baccalaureate degree-holders enter and complete graduate training has declined for both sexes since l970, but much faster for men than for women [National Research Council, Committee on the Education and Employment of Women in Science and Engineering (CEEWISE) l979:2l; NSF, l986:l73-l74]. Both sexes reached their lowest Ph.D. attainment rates in l980 and l98l (6.3 percent for men and 4.4 percent for women) and have shown a slow increase since then (NSF, l986:l73-l74). Since the numbers of women S/E baccalaure- ates were growing rapidly while those of men stayed level or decreased, the net result was a convergence of graduate participation patterns. Overall, the number of S/E doctorates granted to women annually has grown from l,626 in l970 to 5,ll9 in l985 while the corresponding numbers for men are l6,7l7 and l4,045 (Coyle, l986:48-5l; NSF, l986:l7l-l72), yielding an increase in women's share of these degrees from 9.2 percent to 26.7 percent. Summary figures for male-female comparisons for S/E as a whole distort the actual situation because about one-quarter of all baccalaureate S/E degrees in engineering are awarded to men and less than l0 percent to women, yielding comparable disproportions at graduate degree levels (NSF, l986:l69-l70). Since engineering is such a large field, however, these low proportions in fact represent quite large numbers of women. In l985 about ll,000 women earned bachelor's degrees in engineering—a figure that compares with about l5,000 in English, the academic field considered to be quintessentially femi- nine. The increases in numbers of women engineers have been very rapid since l970, when fewer than 350 graduated. Consequently, the propor- tion of women doctorates, which is based on the number of baccalaure- ates eight years earlier, appears small—5.2 percent in l984 (Coyle and Syverson, l986:38)—compared to the proportion of baccalaureates in the same year. In fact, however, it should be compared with the proportion of women baccalaureate engineers who graduated in l976, which was 3.7 percent (Vetter and Babco, l986:l68). These figures l05

20000 | 15000- o JT 10000 - I 5000- Physical Sci. Life Sci. Social Sci. Arts&Hum. 1920s 1930s 1940s 1950s 1960s 1970s 1980-85 Decade Figure l Number of doctorates awarded to women, by decade and field group, l920-l985. yield a higher rate of persistence to the doctorate for women engi- neers than for men, an issue which is discussed in more detail below. Within the broad area of the sciences, there are also considerable sex differences in the distribution among fields. Male Ph.D.s are relatively concentrated in the engineering, mathematical, and physical (EMP) sciences, which together account for almost half of all science doctorates granted to men. In contrast, this field group currently accounts for less than one-fifth of female science Ph.D.s, while almost half are in social and behavioral sciences. The differences in these proportions, however, are less marked now than they were in the early l970s; as more women students became better prepared in quantitative fields, a slow rise occurred in the proportion of EMP doctorates. These trends are depicted in Figure l, which is based on data from the National Research Council's annual Survey of Earned Doctorates. Shifts in graduate field distributions are foreshadowed by under- graduate majors. Since about l970, the single most striking change in women's undergraduate major fields is a massive shift out of education and into business; in the remaining fields, a gradual redistribution appears to be in progress with the more traditional areas like human- ities and social sciences losing students to the natural sciences and engineering. These trends are shown in Figure 2. Based on these num- bers, continued moderate growth can be expected for women S/E doctor- ates in all fields except social sciences for the remainder of the l980s. l06

Male Female Bios Bus Educ Engi Lit Math Phys Soc Field 1 Male Female Bios Bus Educ Engi Lit Math Phys Soc Field Figure 2 Bachelor's degrees in selected fields, by sex, l972 and l982. l07

Attrition and Persistence An undergraduate major in a given field of science (as distinct from engineering) does not necessarily signify a commitment to a high- level professional career in that field for either women or men. Many students of both sexes decide to pursue jobs rather than further study on completing a baccalaureate. Such a decision is not necessarily correlated with ability and undoubtedly has some relationship to eco- nomic status and goals. Overall, higher proportions of women than men do not attend graduate school in most science fields. Students who do continue can be categorized according to type of institution (Ph.D.-granting or master's only), type of control (public or private), and part-time or full-time attendance. Such categories are useful in assessing sex differences in persistence because the proportions of men and women vary among them; women are generally less likely to be full-time students, more likely to enroll in master's-only institutions, and more likely to be in public universities. These distributions are surveyed periodically by NSF and reported in detail in its series, Academic Science/Engineering: Graduate Enrollment and Support. Disproportionate enrollment of women in lower-prestige and/or lower-cost institutions is noted by various authors in various contexts (Berryman, l985; Carnegie Commission on Higher Education, l973:85; Feldman, l974:l5-l6; and Roby, l973:4l). Although Feldman points out that women graduate students are relatively well-represented in medium- quality and low-quality colleges, this finding is not especially per- tinent to S/E students, since virtually all graduate programs in such institutions are in education or other human service areas. Numerous other reports (CEEWISE, l979; l983:2.2-2.3; Cole, l979:2l3; Feldman, l974:l5; Folger, et al., l970:285) note the remarkably even distribu- tion of women S/E doctoral graduates among different quality groupings of institutions, emphasizing that, field for field, women are as likely as men to receive their doctorates from distinguished or strong departments and that this has been true over a long period. Among these various studies, only the reports of CEEWISE disag- gregate the data on quality of doctoral department by field. Others, notably Cole's, use proportions averaged over all fields, which can be misleading because the field distributions differ by sex. However, CEEWISE found significant differences in only four fields: women were overrepresented in highly rated departments in microbiology and psy- chology, and men were overrepresented in similar departments of physics and math during the period l970-l980 (l983:2.7). These recent data represent a significant departure from the tra- ditional situation in which women had much higher dropout rates than men. A much-cited study by Mooney (l969) investigated the graduate school and degree histories of nearly 7,500 men and women who had held Woodrow Wilson Fellowships between l958 and l963. By l966 the attri- tion rate for the women science students in this group was twice that for men (54 percent versus 26 percent). Roughly similar results were obtained in a study by Joseph (l97l), which followed up the first five l08

classes of National Defense Education Acts Title IV fellows. Mooney's analysis was reviewed at length by Patterson and Sells (l973) in a chapter of Rossi and Calderwood's Academic Women on the Move, a volume that has served as one of the basic texts of modern studies on academic women. A study similar to Joseph's was carried out by Harmon (l977), spanning a longer period (l959-l973) and also finding sub- stantially higher attrition for women than for men, as well as much poorer subsequent career outcomes. A problem alluded to by some authors but not resolved is how to distinguish between sex-differentiated rates of entering any graduate program (or, specifically, a doctorate program) and rates of leaving such a program without completing any degree. information available on graduate enrollments in general (from NSF and the Council of Gradu- ate Schools) does not discriminate among those planning master's or doctorates or even those not registered for any degree (Caret and Butler-Nailin, l982:33ff). It is possible to distinguish, by sex, en- rollments in master's versus doctorate-granting departments, but many students in the latter may only be enrolled for a master's and many students from master's-only departments may subsequently pursue a doc- torate elsewhere, so the distinction has limited value. Overall, as a report to the U.S. Congress notes, the proportions of women among S/E baccalaureates and among graduate students are almost identical, so that no significant attrition takes place in entering advanced study [Office of Technology Assessment (OTA), l985:ll7-ll8]. Only two major departures from this rule occur: in mathematical sciences women were 37 percent of baccalaureates and 27 percent of graduate enrollments in l982, and in life sciences they were 40 percent of B.A.s and 52 percent of graduate students. The latter figure probably reflects the very high interest by women in the various health-related profes- sions rather than in doctoral programs in biosciences. The OTA report, citing Malcom (l983), states that "the problem appears to be with persistence in graduate school" (l985:ll8). Com- parison of their figures with degree attainment rates (calculated from NSF l982, l984, and l986) corrected for field-specific Ph.D. completion times (first derived by CEEWISE, l983:l.ll-l.l2, and termed "parity indices"; see also Russell Sage Task Force) allows a somewhat more de- tailed description to emerge. A sample calculation using NSF degree data (NSF, l986:l67-l72) shows that 22 percent of the l977 women bac- calaureates in physical sciences earned master's degrees in l979 and that ll percent of the original group earned doctorates in l984; for men the corresponding figures were 30 percent and l6 percent. Alterna- tively, women obtained 20 percent of physical sciences bachelor's de- grees in l977, l8 percent of master's in l979, and l5 percent of Ph.D.s in l984. Mathematics shows a much greater loss, from 35 percent of baccalaureates to l5 percent of doctorates. In engineering, however, a marked increase in representation occurs, going from 3.7 percent of bachelor's degrees in l976 to 5.2 percent of doctorates in l984; women engineers in that period were therefore much more likely than men to persist to the doctorate. l09

Factors Affecting Continuation and Persistence Broadly speaking, only three factors determine, in some combina- tion, a college student's decision to continue to an advanced degree: ability, access, and motivation. Each of these factors may in turn be affected by a rather large number of demographic, socioeconomic, psy- chological, and possibly genetic variables. Direct causal connections between any of these variables and success in graduate study have not been established, although some highly suggestive correlations are well recognized. Among these is the fact that being female lowers a stu- dent's chances of completing a Ph.D. in nearly all fields, as shown above. Similarly, lack of financial support is likely to lead to abandonment of study, or to prevent its being undertaken in the first place. Mooney's data on Woodrow Wilson Fellows as recomputed by Patterson and Sells (l973:89) showed dramatic decreases in dropout rates when any second-year support was offered. The tacit assumption of most studies on women in graduate programs is that the observed sex differences must be related either to ability or to motivational factors; access issues are not even mentioned by most authors, with rather few exceptions (CEEWISE, l983:l.l3; Hornig, l984:34; National Commission on Student Financial Assistance l985:32, 80; and Russell Sage Task Force). In particular, many studies focus on marriage and parenthood as the putative cause of women's lesser success without considering the role of mundane factors such as money or job prospects in a woman's decision making. Ability As a general rule and across the spectrum of academic fields, women compile significantly better performance records than men at all levels of schooling, including graduate study (CEEWISE, l979:23-26; Feldman, l974:l8; Harmon, l965:28-32; Patterson and Sells, l973:83). Among entering college students who plan on S/E majors, far larger proportions of women than men earned "A" averages in high school, with the greatest advantage (59 percent of women versus 38 percent of men) occurring in engineering and the smallest (29 percent of women versus 24 percent of men) in social sciences (NSF, l986:l63). However, these trends are not uniformly reflected in scores on standardized tests, at any level. On the Graduate Record Examination (GRE), women S/E stu- dents do somewhat better on the verbal and analytical tests but sig- nificantly less well than men on the quantitative portion. These discrepancies, which occur to roughly the same extent on other stan- dardized tests, have so far eluded complete clarification. Possible sex differences in true mathematical ability, as distinct from level of achievement, have been the subject of hundreds of studies over the last few decades, but no conclusive or unambiguous results have been reported. Many of these findings are reviewed in Women and the Mathematical Mystique (Fox, et al., l980). Sex differences in performance on certain tests of mathematical reasoning are quite large among precocious children, but social and cultural causes for this effect, rather than biological ones, can by no means be ruled out. An ll0

analogous problem exists with respect to racial differences, where, for example, the lower scores of black and Hispanic children are customarily ascribed to socioeconomic differences, although such an explanation takes no account of the uniformly higher scores of Asian children. Sex differences in mean quantitative scores on the Scholastic Aptitude Test (SAT) are largely explained by differences in high school course-taking (Fennema, l980) and also correlate with demographic and socioeconomic differences among the pools of male and female test- takers (Hornig, l984:34). Similar data for the GREs are not available, and it is therefore not possible to sort out these issues completely. However, the large, rapid recent increases in women's level of mathe- matical preparation and in the number of women pursuing quantitatively- based fields suggest the absence of decisive genetic differences. Although facility in mathematics is undoubtedly necessary for most science fields, the direct relationship between mathematical ability and ability in various science and engineering fields remains largely unexplored. In particular, it is not known whether the kind of mathe- matical precocity tapped by Stanley and coworkers (Fox and Cohn, l980) and demonstrated by so many more boys than girls is a factor in later scientific ability and performance, although it seems intuitively obvious that it should be so. Less obvious is whether the score dif- ferences in the GRE represent differences in ability or in college course-taking, an issue which remains to be studied for this set of tests. On the SATs, however, it is known that most of the difference disappears when course-taking is held constant (Fennema, l980:8l-82). It is also not known for certain that the residual discrepancies be- tween women's performance in coursework and on standardized tests are not due to some bias inherent in test construction. Mean score differences by sex are not constant over time and have been diminishing quite rapidly in recent years. For physical science majors, the score differences on the quantitative GRE decreased from 40 to 33 points (out of a possible spread of 600) between l979 and l984, for mathematical scientists from 46 to 37, for engineering majors from 58 to l0, and for biologists from 49 to 29. For behavioral and social science majors, however, the difference increased during this period, from 43 to 48 and 55 to 62 points, respectively (calculated from NSF, l986:l65-l66). These changes strongly support the interpre- tation that women's traditional deficit in math is primarily the result of differential preparation. In l984 the highest mean quantitative scores for both sexes were in engineering, followed by mathematical, physical, and biological sciences and then by behavioral and social sciences. The total range for mean scores is nearly 200 points, from 667 to 476 (NSF, ibid.). Due to the lack of correlation between women's grade-point averages (GPAs) and test scores, the latter con- sistently underpredict women's actual performance. No evidence exists, however, to suggest that this effect is given consideration in admis- sions decisions. Certain proxy indicators of ability, in addition to measures such as GPAs and test scores, are widely used in subjective appraisals of graduate students or new Ph.D.s—for example, in consideration for lll

academic appointments or other jobs. One such indicator, receiving a doctorate from a highly rated department, has already been discussed. Another is being relatively young at receipt of the degree, as are re- lated ones such as short degree-completion times. These assessments have been subject to much misinformation concerning women. Older studies typically compared male and female completion times averaged across all fields, a spurious comparison that yields long times for women because there are relatively higher proportions of women in the humanities and in education. These fields are characterized by long completion times for both sexes, a factor generally related to the amount of financial support available (compare Coyle, l986:20-28; Hornig and Ekstrom, in preparation; Syverson, l982:l3-l4, 32-33). In most physical, natural, and mathematical sciences as well as in engi- neering, women complete their Ph.D.s as fast as men or faster (CEEWISE, l979:38; l983:2.3, 2.8-2.9; compare the National Research Council's Summary Report series, especially l985). Notable exceptions occur in two fields, computer sciences and health sciences, which have been shown to have especially unfavorable graduate financial support pat- terns for women but not for men (National Commission for Student Financial Assistance, l985:32, 80; Russell Sage Task Force; Syverson, l982:l6-l7). A factor pertinent to success in a chosen field and related in some way to ability is self-confidence. in psychological attribution theory, self-confidence is generated and reinforced by achieving pre- dictably successful and rewarding outcomes through factors under one's own control, such as effort, rather than through fortuitous events or "luck" (Kaufman and Richardson, l982:5l-53). A number of studies show that young men's and women's self-confidence in both the undergraduate and graduate years, especially in the sciences, is affected in opposite ways. Men's self-confidence is enhanced and women's is diminished even though women's grades, the formal tokens of achievement, remain as high as men's or higher (CEEWISE, l979:l2-l3; Feldman, l974:95-l0l; Maccoby and Jacklin, l974:l54, l57-l58; Zappert and Stansbury, undated:9, l7). These studies, while they demonstrate the effect convincingly, have not traced its causes. Nonetheless, some of their findings suggest strongly that women's loss of self-confidence may be related to dif- ferences in the nature and quality of male and female students' con- tacts with faculty. If so, the effect is likely to depend strongly on the sex of faculty members, and this indeed appears to be the case for both undergraduates (Tidball, l973:l30-l35; l976) and graduate students (Perrucci, l975:l09-ll0). Some possible relationships with a less satisfactory reward structure for women (less recognition of success, restricted career opportunities) will be explored in the concluding discussion. Access The assumption that all students with the requisite ability and interest should have ready access to as much education as they wish is basic to the nation's system of education, although its implementation ll2

may be limited by financial constraints and/or by policy considerations that favor some fields over others. This assumption did not fully in- clude women until passage of the Women's Educational Equity Act, also known as Title IX of the l972 Higher Education Amendments. Until then, women's access to both college and graduate education could legally be restricted, and often was, through various means including outright exclusion, overt and/or covert quotas, and substantial sex differen- tials in financial aid. Total exclusion of women from graduate S/E programs as a matter of institutional policy was no longer widespread by the l960s, although it did persist into that era—e.g., at Princeton. Exclusion of women by departmental fiat or practice was quite common; Harris (l970) pro- vided a comprehensive catalog of methods by which women's participation was restricted or actively discouraged. However, the fact that the major research departments have for some years now educated similar proportions of male and female Ph.D.s argues against the continued ex- istence of restrictive quotas and policies in general, although it in no way disproves their existence in specific cases. A detailed statis- tical study of possible sex bias in graduate admissions at Berkeley (Bickel, et al., l975) found that no such generalized bias had existed in l968-l973 but that some bias favoring women developed at that time. Statistically significant bias against women was shown to have existed in an unidentified huma-nities department. The fact that attrition of women between college and graduate de- grees varies so widely by field, from essentially none in biological and behavioral sciences to about 25 percent in physical sciences and 57 percent in mathematics, suggests that structural biases may occur in some fields but not in others. Since attrition is in general about evenly divided between graduate school entry and degree completion, some as-yet-underaonstrated factor in graduate admissions to certain departments may play a critical role. No precise studies exist concerning the effects of women under- graduates' traditionally skewed distributions among institutional categories—i.e., their overrepresentation in public versus private and lower-ranking versus higher-ranking colleges and universities. Although such sex differences have diminished somewhat since l970 and despite legal mandates to the contrary, most major universities en- rolled and graduated smaller numbers of women than men into the l980s, beginning at a level of about 40 percent in the early l970s and cur- rently standing at about 47 percent, compared to over 50 percent women baccalaureates nationally. An analysis of the undergraduate origins of science doctorates by Coyle and Syverson (l986:l9-2l) shows that private institutions, including many with relatively small female en- rollments, tend to be much more productive of future Ph.D.s than those public institutions that typically graduate large numbers of women. Some of the female underrepresentation in S/E graduate programs, there- fore, is probably the result of skewed undergraduate enrollments. Short of actual exclusion or quotas, the major factor that affects access to graduate study is financial aid. Providing support for the training of scientists and engineers has long been the cornerstone of ll3

all public policy concerned with the continued health of science. The assumption that availability of initial aid and assurance of continuing support will guarantee an adequate supply of highly trained people has been basic to U.S. manpower (sic) planning at least since World War II. The effectiveness of financial support in attracting women to graduate training has not been studied explicitly, but Mooney's (l968) and Patterson and Sells' (l973) findings suggest a strong correlation. Financial aid for women graduate students in S/E is not well studied; indeed, the topic often has not even been mentioned among the many possible factors that influence graduate participation. In his book-length study of women graduate students, Feldman (l974) devoted one small table and two brief paragraphs to a superficial discussion dealing with adequacy of finances for men and women students in rela- tion to marital status. His lack of attention to the topic is all the more surprising because women students were well known to receive sub- stantially less aid than men (Haven and Horch, l972) and to come from wealthier families (Harris, l972:Ch. l). (The latter observation was traditionally interpreted to mean that poorer families were not much interested in educating daughters, although an obvious alternative interpretation is that only daughters of wealthier parents could afford to study if they received less aid than men. ) Feldman stated (l974:l36): Marital status has very little relationship with the receipt of financial aid. The general pattern is that men are more likely to receive teaching or research assistantships and women are more likely to receive fellowships. Decisions con- cerning financial aid may take sex into account but probably not marital status. (Underlining added.) With the possible exception of the final phrase, these statements are borne out by the relatively sparse data available. Roby (l973:47-50) noted that sex differences in financing graduate study are more complex than for college students. She cited a study by Creager (l97l:l9), which showed that women Ph.D. candidates received substantially less support than men from teaching assistantships (TAs) and research assistantships (RAs) but more aid from families. Roby noted that these patterns discriminate especially against women from lower-income backgrounds. Astin (l973:l44-l45) found, using National Research Council data, that for l950-l960 doctorate cohorts from all fields, only l2 percent of women but 22 percent of men received support from governmental sources while 50 percent of women but only 42 percent of men relied on self-help. The National Research Council's annual Survey of Doctorate Recipi- ents regularly collects data by sex and field on all sources of finan- cial aid reported by students, but not on dollar amounts. The annual tabulations show a number of sex differences that correspond in general to Feldman's observation. For example, in physical, engineering, and life sciences fields in l983, about one-quarter more women than men held university fellowships while significantly larger proportions of men held research assistantships (Syverson and Forster, l983:36). ll4

In l98l the Survey of Doctorate Recipients focused on financial support issues and analyzed primary sources of graduate support by field, with somewhat surprising results. Methods of financing graduate study vary widely by field as well as by sex, but almost all fields were shown to have more outside support for men while 45 percent of women but only 30 percent of men were primarily self-supporting. (Note that the reduction in these proportions from the data cited by Astin, above, is substantially more favorable for men.) Especially large gender gaps in self-support occur in earth, environmental, and marine sciences (l9 percent versus l4 percent), in computer sciences (2l per- cent versus l4 percent), and in medical sciences (36 percent versus 20 percent), all favoring men. In most science fields the dominant sources of financial aid are TAs and RAs. Women held significantly higher proportions of TAs in earth sciences, math, and computer sci- ences, but men held much higher proportions of RAs in those fields (Syverson, l982:l5-l8). The National Commission on Student Financial Assistance (l983:32, 80) noted that such skewing of aid results in a disadvantage to women, who are more likely to be performing teaching chores for the department and losing corresponding research time while male students are able to utilize more of their time for research, benefiting from greater professional socialization in the process. Clearly, to the extent that more women science students are having to rely on TAs, they are being tracked into the sex-typed occupation of teaching. The only field group among the natural and mathematical sciences with an egalitarian support pattern—i.e., few if any sex differences in any of the support categories—was biosciences, where only the dif- ference in the proportion of RAs (5 percentage points) remained. Bio- sciences and psychology, which also has fairly equitable support pat- terns, are the only major graduate fields in the sciences that have both large proportions of women and a high Ph.D. persistence rate for them. Field, sex, and race differences in graduate support patterns have again been reviewed in the most recent Summary Report on new doc- torates (Coyle, l986:24-28), yielding findings that strongly suggest sex- and race-differentiated practices in the awarding of graduate financial aid. Overall, access problems for women in S/E graduate studies have clearly diminished since about l970, when complete or partial exclusion of women from many S/E departments was probably still widely practiced. Since then, most institutions have a stated policy of admitting equal proportions of male and female applicants., which most of them errone- ously designate as "sex-blind." If graduate admissions were to be truly sex-neutral and based only on academic performance, the data suggest that higher proportions of women applicants than of men would be admitted. Questions of equity in financial aid cannot be resolved until a detailed study examines not only field-specific sources and patterns of support, but also possible systematic sex differences in dollar amounts. To the extent that sizable differences in funding patterns exist and that the nature of these differences is such as to impair seriously the quality of some women's educational experience in gradu- ll5

ate school, the effect of such practices is plainly deleterious as well as illegal. In its review of women's progress in higher education since l970, the Russell Sage Task Force notes that financial aid data lead to a curious finding: the fields that require high self-support contributions from all students are also those that have high propor- tions of women, while the well-supported fields have low female parti- cipation, yielding an almost linear inverse relationship. The direc- tion of causality of this phenomenon needs further investigation. Future studies of sex differences in financial aid should attempt to distinguish between support offered to entering graduate students and continuing support for later stages of study, in an effort to shed light on attrition and persistence issues. In this connection the greater success of women in gaining competitive fellowships, presumably because of superior ability and better credentials, should be examined in some detail. In the normal course of events, reliance on continuing support from a somewhat chancy source is probably less conducive to self-confidence than the more certain income from a research assistant- ship. Motivation The literature on women's achievement motivation in general and on professional career-goal formation is enormous, well beyond the scope of this review, but little of this work is of specific utility in interpreting findings on women graduate students in S/E. Earlier in this review, it was noted that the basic career choice is normally made before entering graduate school but that sustained motivation is necessary for persistence. Support for high aspirations may come from a variety of factors in the personal and educational environment, as well as from a realistic expectation of future professional rewards. The voluminous literature on women's achievement motivation has been critically analyzed by Kaufman and Richardson (l982) with partic- ular reference to professional achievement. in brief, these authors traced the development of the two major historic approaches to the study of motivation in women: (l) the establishment in children of an intrinsic, stable motive to achieve and (2) the socialization model that stresses the salience of gender-role learning and the incongruence between traditional female gender roles and professional or public achievement. The authors developed a synthesis that argued for a more fluid interpretation of motivation as a response to sometimes changing or evolving opportunities. The latter would certainly accord better with the observed large-scale movement of women into S/E and other non- traditional fields and the very rapid behavioral changes that resulted in tripling the number of women Ph.D.s in a decade. Feldman's study (l974) of women graduate students illustrates the limitations of both the classic psychological and the sociological ap- proaches to motivation. Examining the stated academic career goals of men and women graduate students (almost l2,000 individuals among the total sample of 33,000 in l969), he found that about 85 percent of the men and about 70 percent of the women (except for 8l percent in so- called "equalitarian" fields) aspired to teach in colleges or universi- ll6

ties; correspondingly more of the women aspired to teach in junior colleges. These sex differences were accentuated in female-majority fields and persisted whether or not a student expected to earn a Ph.D., expressed strong interest in an academic career, or planned to special- ize. Women who had a high intellectual self-image and those who at- tended highly ranked institutions expressed choices much like the men's, but the ambitions of men who rated themselves "not intellectual" or who attended low-ranking institutions were not lowered nearly as much as similar women's. Men in female-majority fields held especially high aspirations, but women did not. Feldman's interpretation of his findings was both superficial and conventional: In order for a woman to aspire to a more prestigious career, she must have those qualities associated with that career choice. If she does not, she is much less apt to choose a prestigious career than her male counterpart. . . . Women who aspire toward university careers are more qualified than their male counterparts. Those who are not plan to end up in junior colleges. On the other hand, men opt for university academic careers whatever their qualifications. (l974:79) He closed that rather extraordinary chapter with the statement that "at the higher-quality institutions, junior college teaching is not a prime objective for students, but these institutions may assume that more women than men will end up in such positions" (l974:l0l). Feldman did not comment on the self-fulfilling nature of that assumption. Feldman's data, even for the "female-minority" fields that include most S/E disciplines, are of limited value not only because the condi- tions they reflect are now largely outdated, but also because the focus on "teaching" in the sense in which he used the term is inappropriate to this field group and because the failure to distinguish between college and university careers is critical. Overall, his conclusions reflect either inability or unwillingness to relate the findings to the reality of academic employment policies in the l960s, to the grad- uate students' entirely accurate appraisal of their likely opportuni- ties, and to the realistic gender differences that derived from that appraisal. In another chapter on dedication to graduate study, Feldman exam- ined a variety of conditions that surely concern motivation, although he seemed not to view them in that light. He found that about one- third of male faculty and fellow students in science fields believed women students to be less dedicated than men (l974:l04), although in fact more women report themselves as subordinating other aspects of life to their work (l974:ll4). More women than men in several science fields said they may drop out because of lack of ability (l974:ll7), although at least 50 percent more of them had high grades as under- graduates (l974:ll3). Women science students described themselves as being much less satisfied than men with their ability to do original work, which Feldman interpreted as lack of confidence but which could equally well reflect different situations within departments (l974:ll8- ll9) or higher standards of originality being held by women. Women ll7

students also were much less likely to have collegial or close working relationships with faculty, and their self-image was accordingly quite low compared to men's (l974:l20-l2l). These findings on women's apparent underestimation of their abil- ity, their relatively low involvement with faculty, and the consequent low level of self-confidence despite continuing high grades are cor- roborated by other investigators (Berg and Ferber, l983:635; Zappert and Stansbury, undated:9). Berg and Ferber found, among a cross- section of graduate students at the University of Illinois, some espe- cially pertinent associations between Ph.D. completion and having been treated as a colleague by faculty, a much less likely event for women than for men. The authors concluded that the motivation of women stu- dents could be better supported by improved mentoring by faculty. Zappert and Stansbury examined a sample of Stanford S/E graduate stu- dents with respect to sex differences in various measures of profes- sional interest and commitment. A "creativity index" showed some dif- ferences: more women felt it essential that a career should allow them to use their skills and knowledge, and more men felt it was important to do seminal work in their fields, to gain social recognition, and to have opportunities for risk-taking. Certain differences analogous to those reported by Feldman emerged in regard to career plans; 62 percent of men and 5l percent of women expected to hold academic appointments, and fewer women than men en- visioned themselves as directors of laboratories, as entrepreneurs, or as corporate chief executive officers. Parallel sex differences oc- curred in overall satisfaction with their graduate programs, degree of responsibility held within a research group, respect for their ideas accorded them by their advisers, and the general quality of relation- ships with faculty. One-fifth of the women and 7 percent of the men reported having experienced some discrimination at Stanford. Again, the authors did not attempt to derive any direct relationship between the women's lower expectations, their realistic assessments of opportunity (e.g., as illustrated by the skewed sex distribution of Stanford fac- ulty), and the manifest differences in the graduate experience. As mentioned earlier, marriage and parenthood, whether existing or anticipated, are widely believed to exercise unfavorable influences on women's graduate study and subsequent careers, usually through making greater demands on women's time than on men's or through constraining women to particular locations. Perhaps because this reasoning seems so obvious, it has not been very thoroughly explored in the few em- pirical studies that actually deal with women S/E graduate students. Zappert and Stansbury (undated:l3) reported that 37 percent of their sample of Stanford medical and S/E graduate students were either married or living with partners, and a total of only 6 percent had children. Among new doctorates in l985, 58 percent of the men and 48 percent of the women scientists and engineers were married (Coyle, l986:49, 5l). Predictably, about twice as many of the women as the men at Stanford reported experiencing difficulty in integrating family and work demands, and women were significantly more likely to antici- pate taking time off for parenting. However, no study has yet at- tempted to correlate marital status and parenting responsibilities ll8

with the more objectively measurable variables that may affect graduate study in these fields. Long overdue are studies that investigate the relationships among such factors as academic achievement (e.g., test scores or GPAs), kind and amount of financial aid, length of time to the Ph.D., marital status including divorce, number of children if any, presence or absence of on-campus child care, and institutional policies that may affect families differently from single people (e.g., whether financial aid is available to both members of a couple). Discussion and Conclusions The large and not yet fully exploited data base on doctoral stu- dents in science and engineering, as well as other academic disci- plines, makes evident the rapid convergence of male and female educa- tional patterns that has taken place since about l970, when artificial constraints on women's higher education were removed by law. Neverthe- less, women remain underrepresented relative to men in most S/E fields while having caught up or surpassed them in behavioral sciences and in some humanities and education fields. This underrepresentation is a growing cause for concern in view of declining interest on the part of men in the traditionally male-dominated areas at a time of heightened awareness of erosion in our international position with respect to both national security and economic competitiveness. The public in- terest requires actions to remedy these developing problems by the more effective inclusion of women and minorities. Objective measures of ability and performance make it amply clear that women suffer no deficits in these dimensions and that explana- tions for lower participation and lower success rates in completing graduate study must be sought elsewhere. Two major areas of interest for further research and possible intervention programs emerge. One of these is the issue of financial support, and the other is the pos- sible effect of improved educational environments (especially with re- gard to more equitable treatment by faculty) and of equalization of ultimate rewards in recognition, opportunity, and earnings potential on women's initial and continuing participation in S/E fields. The rapid and large-scale increases in women's representation should lay to rest any lingering doubts about native abilities; changes in genet- ically determined traits do not occur in thousands of people in the space of a decade. The consistent failure of researchers to explore sex differences in financial aid patterns once their existence had been demonstrated is hard to explain. Such differences are among the potentially most compelling in explaining female attrition from graduate programs at all levels, yet few investigators have pursued the existing leads. Studies seeking to elucidate relationships between funding patterns that track women into lower-level teaching functions and men into pre- ferred faculty careers, or that distinguish on gender lines between those to whom faculty devote full professional attention and those on whom they rely to perform departmental chores, should receive high priority. These issues are complicated by the fact that graduate fac- ll9

ulties are not entirely disinterested parties, since they are respon- sible for the training of those specialists whom they will eventually hire or reject as colleagues. Such a dual role should impose an espe- cially high standard of equity and fairness in the distribution of financial aid, personal attention, and nurturing of high aspirations and in the selection of new colleagues. Massive evidence (see also Zuckerman, this volume) suggests that such a standard has not yet been established or even articulated in most universities. Finally, the well-documented disadvantages women face in employment after the Ph.D. must be explored with special attention to the dele- terious effects they surely exert on the hopes, expectations, and aspirations of women students. An enormous amount of attention has been devoted to what have been regarded as women's intrinsic limita- tions in entering and pursuing careers in science and engineering. Deficits in mathematical ability, in the personality traits considered essential to successful careers, and in strength of motivation and commitment have all been endlessly examined. In contrast, the entire issue of structural barriers within graduate education in the sciences and engineering has been virtually exempt from investigation for more than a decade. The state of research on women's participation suggests it is time to remedy this omission. Bibliography Astin, H. S. l973. Career profiles of women doctorates. In Academic Women on the Move, A. S. Rossi and A. Calderwood, eds. New York: Russell Sage Foundation. Berg, H. M., and M. A. Ferber. l983. Men and women graduate students: Who succeeds and why? Journal of Higher Education 54(November/ December):629-647. Berryman, S. E. l985. Minorities and women in mathematics and science. Paper presented at the annual meeting of the American Association for the Advancement of Science, Los Angeles, May l985. Bickel, P. T., E. A. Hamel, and J. W. O'Connell. l975. Sex bias in graduate admissions: Data from Berkeley. Science l97:39-404. Carnegie Commission on Higher Education. l973. Opportunities for Women in Higher Education. New York: McGraw-Hill. Cole, J. R. l979. Fair Science: Women in the Scientific Community. New York: The Free Press. CEEWISE (Committee on the Education and Employment of Women in Science and Engineering), National Research Council. l979. Climbing the Academic Ladder: Doctoral Women Scientists in Academe. Washington, D.C.: National Academy of Sciences. CEEWISE. l983. Climbing the Ladder: An Update on the Status of Doc- toral Women Scientists and Engineers. Washington, D.C.: National Academy Press. Coyle, S. L. l986. Doctorate Recipients from United States Universi- ties: Summary Report l985. Washington, D.C.: National Academy Press. l20

Coyle, S. L., and P. D. Syverson. l986. Doctorate Recipients from United States Universities: Summary Report l984. Washington, D.C.: National Academy Press. Creager, J. A. l97l. The American College Student: A Normative De- scription". ACE Research Report 6:5. Washington, D.C.: American Council on Education. Feldman, S. D. l974. Escape from the Doll's House: Women in Graduate and Professional School Education. Report prepared for the Car- negie Commission on Higher Education. New York: McGraw-Hill. Fennema, E. l980. Sex-related differences in mathematics achievement: Where and why. In Women and the Mathematical Mystique, L. H. Fox, L. Brody, and D. Tobin, eds. Baltimore, Md.: Johns Hopkins University Press. Folger, J. K., H. S. Astin, and A. E. Bayer. l970. Human Resources and Higher Education. New York: Russell Sage Foundation. Fox, L. H., L. Brody, and D. Tobin, eds. l980. Women and the Mathe- matical Mystique. Baltimore, Md.: Johns Hopkins University Press. Fox, L. H., and S. J. 'John. l980. Sex differences in the development of precocious mathematical talent. in Women and the Mathematical Mystique, Lynn H. Fox, Linda Brody, and Dianne Tobin, eds. Balti- more, Md.: Johns Hopkins University Press. Caret, M. S., and P. Butler-Nailin. l982. Graduate and professional education: A review of recent trends. Paper prepared for the National Commission on Student Financial Assistance. ERIC document no. ED 228 932. Washington, D.C.: The Commission. Harmon, L. R. l965. High School Ability Patterns: A Backward Look from the Doctorate. Scientific Manpower Report No. 6. Washington, D.C.: National Academy of Sciences. Harmon, L. R. l977. Career Achievements of the National Defense Edu- cation Act (Title IV) Fellows of l959-l973. Washington, D.C.: National Academy of Sciences. Harris, A. A. l970. The Second Sex in Academe. American Association of University Professors (AAUP) Bulletin 56(3):283-295. Harris, S. E. l972. A Statistical Portrait of Higher Education. Re- port to the Carnegie Commission on Higher Education. New York: McGraw-Hill. Haven, E. W., and D. H. Horch. l972. How College Students Finance Their Education. New York: College Entrance Examination Board. Hornig, L. S. l984. Women in science and engineering: Why so few? Technology Review (November/December):30-4l. Hornig, L. S., and R. B. Ekstrom. In preparation. The Status of Women in the Humanities. Report to the National Endowment for the Human- ities. Joseph, C. B. l97l. A Report on the First Five Classes of NDEA Title IV Fellows, by Sex. Washington, D.C.: U.S. Government Printing Office. Kaufman, D. R., and B. L. Richardson. l982. Achievement and Women: Challenging the Assumptions. New York: The Free Press. Maccoby, E. E., and C. N. Jacklin. l974. The Psychology of Sex Differ- ences. Stanford, Calif.: Stanford University Press. l2l

Malcom, S. M. l983. Women in Science and Engineering: An Overview. Washington, D.C.: American Association for the Advancement of Science. Mooney, J. D. l969. Attrition among Ph.D. candidates: An analysis of a cohort of recent Woodrow Wilson fellows. Journal of Human Re- sources 3:47-62. National Commission on Student Financial Assistance. l983. Signs of Trouble and Erosion: A Report on Graduate Education in America. Washington, D.C.: The Commission. NSF (National Science Foundation). l982. Academic Science/Engineering: Graduate Enrollment and Support. Washington, D.C.: U.S. Government Printing Office. NSF. l984. Academic Science/Engineering: Graduate Enrollment and Sup- port. Washington, D.C.: U.S. Government Printing Office. NSF. l986a. Academic Science/Engineering: Graduate Enrollment and Sup- port. Washington, D.C.: U.S. Government Printing Office. NSF. l986b. Women and Minorities in Science and Engineering. Bien- nial report. Washington, D.C.: U.S. Government Printing Office. Office of Technology Assessment. l985. Demographic Trends and the Scientific and Engineering Work Force: A Technical Memorandum. Washington, D.C.: U.S. Government Printing Office. Patterson, M., and L. Sells. l973. Women dropouts from higher educa- tion. In Academic Women on the Move, Alice S. Rossi and Ann Calderwood, eds. New York: Russell Sage Foundation. Perrucci, C. l975. Sex-based professional socialization among graduate students in science. In Research Issues in the Employment of Women: Proceedings of a Workshop. Washington, D.C.: National Academy of Sciences. Roby, P. l973. Institutional barriers to women students in higher ed- ucation. In Academic Women on the Move, Alice S. Rossi and Ann Calderwood, eds. New York: Russell Sage Foundation. Russell Sage Task Force on Women's Higher Education Since l970. In preparation. Syverson, P. D. l982. Doctorate Recipients from United States Univer- sities: Summary Report l98l. Washington, D.C.: National Academy Press. Syverson, P. D., and L. E. Forster. l983. Doctorate Recipients from United States Universities: Summary Report l983. Washington, D.C.: National Academy Press. Tidball, M. E. l973. Perspective on academic women and affirmative action. Educational Record 54(Spring):l30-l35. Tidball, M. E. l976. Of men and research: The dominant themes in Amer- ican higher education include neither teaching nor women. Journal of Higher Education XLVII(4):373-389. Vetter, B., and E. Babco, eds. l986. Professional Women and Minor- ities. Washington, D.C.: Commission on Professionals in Science and Technology. Zappert, L. T., and K. Stansbury. Undated. In the Pipeline: A Compar- ative Analysis of Men and Women in Graduate Programs in Science, Engineering, and Medicine at Stanford University. l22

PERSISTENCE AND CHANGE IN THE CAREERS OF MEN AND WOMEN SCIENTISTS AND ENGINEERS- A REVIEW OF CURRENT RESEARCH Harriet Zuckerman Introduction The National Academy of Sciences elected its first woman member just a half century ago. She was Florence Sabin, an anatomist and embryologist. Not only was Sabin a first here, but her entire career was a succession of "firsts." in l902, she was the first woman ap- pointed to the faculty of the Johns Hopkins Medical School, arguably the most distinguished school of medicine in the United States at the time. She was also the first woman to be made a full professor there (l9l7), the first woman elected president of the American Association of Anatomists (l924), and, upon leaving the Hopkins in l925, the first woman to be made a full member of the Rockefeller institute of Medical Research (now the Rockefeller University), that being the same year she was elected to the Academy (Breiger, l980; Rossiter, l982). This array of posts makes it clear that Sabin was an insider and member of the scientific establishment. She was also an outsider and social pioneer, being not just the first but often the only woman to be included in the circles in which she moved. (On insiders and out- siders, see Merton, l972; on social pioneers, see Zuckerman, l987.) Sabin was atypical, if not unique, among women scientists of the time, in the extent to which her work was recognized and rewarded by the com- munity of scientists. By contrast, the historical record shows that many accomplished women were ignored or actively discouraged. Those honors which came to them at all came very late (Rossiter, l982). How much has the lot of women scientists changed since the Sabin era? This review of research on the careers of American men and women scientists will show that this simple question has no simple answer. Rather, as I shall indicate, the pertinent data, drawn from current studies, show three separate but interconnected patterns. First, there are persisting differences between men and women scientists, on average, in role performance and career attainments when viewed cross- sectionally. These differences are almost always in the direction of comparative disadvantage for women and are usually combined with con- siderable intra-gender differences. Second, there are signs of growing convergence between men and women in access to resources, research performance, and rewards—that is, evidence for increasing gender simi- larity over the last decade and a half, especially between younger men l23

and women. Third, there is evidence for growing divergence between men and women of the same professional age in published productivity and in some, but not all, aspects of career attainment—that is, for growing intra-cohort differences as members move through their careers. This report divides into four parts. The first briefly describes the population of American scientists and engineers. The second in- ventories the main findings of research on comparative career attain- ments of men and women, earmarking the three patterns of similarity and difference just noted. The third reviews the principal explana- tions, or "theories," that have been proposed to account for gender differences in careers and assesses how well the data square with these explanations. The last identifies some directions for further inquiry, based on "specified ignorance" (Merton, l987) or what we now know we need to know and why we want to know it. Selected Demographic Characteristics of American Men and Women Scientists and Engineers This review of data on the size, sex composition, and growth rates of the population of American scientists and engineers is not intended to be comprehensive. Instead, it has two purposes: to set in context the more detailed but necessarily more limited findings of research on careers inventoried in the second part of this paper and to provide a cautionary note against the use of simple bivariate distributions of gender and salary, gender and rates of unemployment, or gender and organizational rank in describing the relative status of men and women scientists and engineers. Such bivariate distributions mislead as much as they inform, since they mask marked differences between men and women scientists in professional age, education, and scientific field, these being independently related to salary, unemployment, and rank. These differing distributions of men and women must be taken into ac- count in gauging the extent of gender difference in career attainments. So much then for intent. Where do things stand with respect to sheer numbers? In l984, about four million Americans were working as scientists and engineers; and of these, 5l3,000 (or l3 percent) were women (National Science Foundation, l986:6l; hereafter, NSF)—not a large share but 2.8 times as many as there were a decade earlier, when just l85,000 women (6 per- cent of the total) were working in these occupations (National Science Board, l977:l52; hereafter, NSB).l Indeed, the number of women entering the scientific and technical population has grown even more rapidly than this population as a whole, which almost doubled in the -^emographic data on scientists and engineers are routinely reported by the National Science Foundation in its report, Women and Minorities; by the National Science Board in its biennial Science Indicators vol- umes; and by agencies such as the National Research Council and the National Institutes of Health. l24

same period (NSB, l977:l52). Since the number of scientists and engi- neers usually grows only by increments of young people beginning their careers (few new entrants are older people moving from other jobs), women scientists and engineers as a group are apt to be younger, on average, than men. This inference is consistent with the report that women are twice as likely as men (60 percent versus 27 percent) to have fewer than l0 years of professional experience (NSF, l986:4).2 Women are also less apt than men to hold advanced degrees, and this is so in every field of science and engineering. In all, l9 per- cent of men scientists and engineers hold doctorates as against ll percent of women, but such differences vary considerably from field to field (NSF, l986:l). Gender differences in educational attainments of men and women suggest that women, on average, may hold high-ranking positions less often than men when the doctorate is a required creden- tial for such high rank. Even so, the proportion of doctorates being awarded to women has been growing since l960 and growing particularly rapidly in the last l5 years. By l985, 24 percent of doctoral degrees in science and engineering went to women as against just 7 percent in l970 (Vetter and Babco, l986; see also Vetter, l98l). Such marked increases have occurred against the background of decreases, not just in the relative but also in the absolute numbers of men receiving doctorates (see Fig- ures l and 2). It is not clear now whether the prestige of scientific occupations will change as they become more attractive to women and less attractive to men. What is clear is that such high growth rates in the numbers of doctorates earned by women means that women doctor- ates are considerably younger professionally, on average, than men. Men and women differ also in the fields they choose, and this is so both for holders of doctorates and for those with lower level de- grees. 3 Women are concentrated in the life and social sciences with comparatively small numbers working in the physical sciences and engi- neering (NSF, l986:6l-62, 7l-72) as Figure 3 shows, and this has been so for decades. They comprise just 7 percent of all doctorates in the physical sciences and 2 percent in engineering in contrast to l7 per- cent of doctorates in the life sciences and 22 percent in the social sciences, these composite fields having once been described as the "dispassionate" and the "compassionate" sciences.4 (It is not clear whether gender differences also exist in specialty choice.) There are I have not located any data on large distributions of men and women scientists and engineers currently at work, I rely on inferences of this sort. ^Rossiter (l978) proposes that such differences are responses to the differing job markets in various sciences. Women, she suggests, are more apt to find jobs in new and rapidly growing fields, where short- ages make the hiring of women "tolerable." This hypothesis obviously needs careful examination over fairly long periods. 4This may be another way of identifying fields with comparatively low and comparatively high loading of mathematical analysis. l25

s. 1960 1965 1970 1975 Year 1980 •a- All fields -•- S/E •*• Non-S/E 1985 SOURCE: Survey of Earned Doctorates, National Research Council. Figure l Shares of doctorates earned by women. -a- All fields -•- S/E -o- Non-S/E 1960 1965 1970 1975 Year 1980 1985 SOURCE: Survey of Earned Doctorates, National Research Council. Figure 2 Long-term trends in doctorate production. l26