Professor Lynne Billard visited Australia in 1994 and gave some public lectures on women in academia. These lectures received some attention in the media and so the Gazette invited Professor Billard to write an article on these matters. The resulting article follows.
Lynne Billard is University Professor and Professor of Statistics at the University of Georgia. She obtained her PhD in 1969 from the University of New South Wales and has held academic positions in Australia, Britain, Canada and the USA.
Professor Billard's publications range over many issues including AIDS and epidemics, sequential analysis, time series and inference, with applications in agriculture, biology, epidemiology, education and social sciences.
She has served on US national committees including an Advisory Committee for the National Science Foundation, a Panel on AIDS and a Panel on Microsimulation Modeling of Social Welfare Policy.
Further, she has undertaken many gender and racial studies both for the University of Georgia and, in an advisory capacity, for other universities, and has several publications on gender related issues.
At present, Professor Billard is President of the International Biometric Society, and will serve as President of the American Statistical Association in 1996.
Legislation designed to guarantee equal opportunity for all in the workplace in Australia was enacted in 1984. It is still too soon for enough hard data, as opposed to anecdotal evidence, to emerge for any definitive statements on the success or otherwise of these legislative goals. However, sufficient data have accummulated in countries, such as the United States which passed comparable laws in 1972, to provide clear evidence of the impact of the implementation of these legislative intentions. What little information that is available for Australia and other countries set on this path, is not inconsistent with its United States counterpart data. Therefore, we report herein on progress in the United States allowing those trends to serve as a harbinger of what we might expect will transpire elsewhere. More importantly however, there were observed to be early gains in the first ten years or so after the legislation was passed followed by a levelling off well short of the goals of parity despite a perception that equality had been attained. Therefore, those of us elsewhere have the opportunity to learn from this U.S. experience to try to develop procedures and processes to ensure that no such premature levelling off occurs in our situations.
We shall look briefly at tertiary level educational attainments leading to academic job opportunities, and hence movement up the promotion and tenure ladder and salary levels. Since progress naturally tends to revolve around work accomplishments, how that work is evaluated differentially for men and for women will also be considered. Where appropriate and available, data will be that for the mathematical sciences including statistics. More complete details together with data for other disciplines can be found in Billard (1994).
Contrary to long-held beliefs, women perform just as well as do men, as reflected by their grades in mathematics at high school (see Table 1). However, women are less likely to take as many courses in mathematics (see Table 1) probably because of the direct influence of these contrary beliefs. (The same results pertain for the physical sciences.) On the other hand men consistently score higher than do women on standardised test scores such as the Scholastic Aptitude Test for Mathematics (SAT-Math), Quantitative and Analytical components of the Graduate Record Examination (GRE-Quant, GRE-Anal); see Table 2 for these scores for undergraduate majors in the mathematical sciences. However, when controlled for course taking, this gap narrows. Despite the fewer years of preparation and the lower SAT scores which play a role in university admission, roughly the same number of men as women graduate with a bachelors degree in the mathematical sciences.
Further, Hornig and Tidball (1979) have shown women have better scholastic records at both the undergraduate and graduate levels. Yet, there is considerable attrition in the proportion of women subsequently graduating with a masters degree and with a doctoral degree (see Table 3). Clearly factors unrelated to abilities are playing a more dominant role in influencing a woman's choice or opportunity to enter a graduate program, her retention and the completion of her degree. Throughout the 1970s and 1980s the increase in percentage of women earning the doctorate was largely due to a decrease in numbers of men; so if it is assumed whatever additional factors were discouraging men from embarking on a graduate program also discourage women, then on balance there are indeed more women entering and completing this degree. This is most encouraging, though clearly there is still room for improvement.
Prior to the 1980s, this persistence or lack thereof in graduate school might have been in part explained by inequitable levels of financial support. However, those disparities appear to have disappeared through the 1980s; so that clearly other less tangible reasons must explain the still visible and real higher attrition rates for women. However, there has been an increase in numbers of both men and women in the early 1990s with 23.8% (288) of the doctorates having been awarded to women in 1993. For mathematics, this was 26.6%, for probability and statistics 21.4%, and for applied mathematics 20.7% (see Fulton, 1994).
3.1 Employment: The most definitive study on employment characteristics and the subsequent movement through the promotion and tenure ranks is the matched triad study of Ahern and Scott (1981). Although this study reports on prevailing conditions as of 1979, other (unmatched) data reporting on comparisons of the 1980s and 1990s are not inconsistent with the conclusion of the Ahern and Scott study, thus still lending great importance to its original results. Their study surveyed approximately 50,000 doctoral graduates from the United States in the years 1940-79. Triads consisting of two men and one woman were matched by year and field degree, type of doctoral granting institution, race, and where appropriate type and years of employment. There were four cohorts, those receiving their degree in the years 1940-59, 1960-69, 1970-74 and 1975-78. Data for various classifications are presented in Table 4 as they pertain to the mathematical sciences.
In all cohorts, despite starting out with equivalent credentials, more men than women are in full time employment positions and more women are in part-time positions or are un(der) employed than are men. The same pattern persists in the 1990s as evidenced from the 1993 Annual AMS-IMS-MAA Survey (Fulton, 1994) in which we see that of those hired, 61.5% of the women but only 21.4% of the men are in part-time appointments over all institutions. If we exclude those institutions for which the highest degree awarded is a masters degree (group M) or has a bachelors degree (group B), which traditionally have a higher proportion of part-time appointments, we observe that for doctoral institutions (Groups I, ..., V) 9.0% of the men and 35.8% of the women occupy part-time positions.
Table 5 shows the percentage of all positions (men and women) which are filled by women, for various faculty status across the group categories of the mathematical science departments. Here, departments are predominantly or exclusively mathematics except for Group IV (Statistics) and Group V (Applied Mathematics). We observe that in general the women are more likely to be in the lesser ranked institutions. Thus, of the tenured faculty in the top ranked doctoral mathematics departments, 4.9% only are women (and 95.1% are men). This percentage for women increases to 5.9% for Group II departments, to 6.5% for Group III, to 10.9% for Masters departments, and 15.1% for Bachelors departments; likewise, for the other faculty status classifications. The anomaly is that for non-tenure track positions, 30.3% of those in Group II departments but only 10.9% for Group III are filled by women. (In effect, more of the ``Group III'' nontenure positions given to women were appointed to ``Group II'' positions.) This same anomaly occurs when looking at the changing status within a given institutional Group. In general, for each Group, larger proportions of the women than occur for men are increasingly in the lower ranked positions. Thus more men are tenured, while more women are in nontenured positions.
If it is assumed that the recent doctoral graduates and therefore recent doctoral hires are all in the tenure-track-untenured or the nontenure-track positions, by comparing the percentages who are women [in Table 5 column (a) and/or (b) for mathematics only, (c) for doctoral institutions] with the 26.6% who were 1993 recipients of doctoral degrees in mathematics [or by comparing the column (d) figures for all the mathematical sciences with the 23.8% doctorates overall to women], it is clear that women are still not being appointed to academic positions at a rate commensurate with their rate of earning the doctorate degree. The lag becomes more pronounced as the rank of the departmental group increases. A possible exception is statistics with the proportion of women in the untenured ranks slightly higher than the proportion of 1993 graduates who were women. The numbers for Applied Mathematics are too small for any meaningful conclusions. However, when we compare these figures with those of Table 6 which give the percentages of the men and percentages of the women within gender and within all faculty (i.e., men and women combined) for the mathematical sciences for four-year and doctoral institutions we see that by 1989 there is some progress, most of this progress occurring in that more women than previously are receiving initial appointments even if not yet at a comparable rate at the better institutions.
3.2 Promotion and tenure: Let us first consider the matched triad study; see Table 4. We observe that for all cohorts women are more likely to be frozen in rank below the Full Professor level. For the 1940-59 cohort, it might be reasonable to assume that few would be expected to continue moving up the ranks at or after the time of the 1972 antidiscimination legislation. However, by the prevailing standards of the 1970s, it is not unreasonable to expect that most of the 1960-69 cohort had the expectation to have attained the Full Professor rank by 1979. Nevertheless, more men (one and a half times as many) than women have been promoted to this rank, and more of the men (97%) have received tenure than have women (80%). Likewise, for the 1970-74 and 1975-78 cohort, more men (one and a half times) have been promoted and more have been tenured, despite starting out with equivalent credentials.
Three commonly and widely held beliefs to explain this slower rate of progress were investigated by Ahern and Scott (1981); all three were found to be myths. The first was that women's slower rate of progress was due in part to her holding appointments predominantly in teaching as distinct from research institutions. Instead, those with primarily teaching responsibilities progressed faster than those with research duties as well; but still in each case men fared better than women. Thus, for the 1970-74 cohort, men (women) at a teaching institution were promoted to Associate Professor at a 64% (48%) rate while those at a research institution were promoted at a 60% (32%) rate.
The second myth was that women's childbearing responsibilities resulted in slower progress up the academic ranks. To the contrary, married women with children fared best (51% promoted), followed by married women without children (41%), then single women without children (37%) and single women with children (33%). The corresponding figures for men were 66%, 51%, 53%, and 80%, respectively, where these figures cover all disciplines. One possible explanation is that married women have links with the male networks through their husbands; this is especially so in the sciences including the mathematical sciences; see Gibbons(1992). The third belief, also unsubstantiated by the data, at least for the matched 1970-74 cohort, is that women are less mobile than men. Rather, more women (28%) than men (19%) changed jobs, but women tended to move laterally while men moved to better positions (i.e., higher ranks and/or higher salaries).
The rank and tenure status data for mathematical science faculty in doctoral and four year institutions in 1989 (of Table 6)show that while there is a slight improvement in the movement up these ranks, very little has changed from the 1979 picture as seen in the matched triad study. Indeed, other data covering all disciplines over doctoral institutions show that after an upward trend through the 1970s, there was a levelling off through the 1980s.Table 7 shows this trend for tenure; comparable data for rank is available only from 1988, but that too has been essentially unchanged in the subsequent years from that shown for 1992.
3.3 Salaries: As for ranks, so do women's salaries lag behind those for men. Part of the explanation is that women tend to start at lower salaries than do men with comparable records. Thus, it is hard for them to catch up. This occured for the matched triad study (see Table 4) where even for the most recent cohort a salary gap has appeared; this despite the more obvious non-discriminatory pressures of the 1972 enactments. The figures of Table 8 covering all doctoral institutions and all disciplines show this has persisted over the years. Indeed, for the Associate and Assistant Professor ranks where any legislative constraints or encouragements should have had most impact, we observe that, rather than narrowing, the salary gap has instead widened since the early 1970s. Since faculty do move at least into the Associate Professor rank, explanations along the lines that men have been employed longer than women are clearly invalid. Longevity arguments may be more applicable at the Full Professor rank; however, they are not the entire reason. At one major institution across all ranks only 68% ($5880 of the $8700 average difference) could be explained by longevity (or other nongender related factors), with the remaining 32% being apparently due to gender differences (see, Billard et al., 1994). As for the myths developed to explain slower promotion and tenure rates, so the myth that part of the salary gap is due to women's voluntary choice such as career interruptions did not hold up in Ferber and Kordick's (1978) study.
Promotion and tenure including appointments to higher levels and the concommitant salary structures are all tied in various ways to an evaluation of work performance and accomplishments. It is here that subtle, albeit unintentional forms of inequitable biases come into play. Menges and Exum (1973) concluded that there was ``considerable bias shielded by ambiguous criteria and standards; while Theodore (1986) sheds light on the issue by discussing the effects of ``shifting the criteria''.
Some studies have shown that men have published more than women over their careers; for example, Cole (1979), and Fish and Gibbons (1989). Others have been able to explain part of this difference by the fact that women were more likely to be in teaching positions, and/or in disciplines where books rather than papers were published; see, for example, Astin and Bayer(1973). There is by now a large number of studies addressing this issue; see Billard(1994) for a more detailed review.
The substantial literature covers many aspects and many disciplines. Behind these studies is the prevailing conclusion that women's work is deemed to be of lower quality and of less significance than that performed by men (see Persell,1983; Cole and Zuckerman,1984; Chamberlain,1988). One study which illustrates this phenonenom most clearly is that of Paludi and Bauer(1983).
In the Paludi and Bauer study, 180 men and 180 women reviewers were each asked to review comparable papers of which one-third each were purported to be written by a John T. McKay, a Joan T. McKay, and a J.T. McKay. The mean average ratings (one to five, with one being the top rating) are shown in Table 9. Papers believed to be written by Joan T McKay were given an average rating of 3.0 by both men and women reviewers. However, papers believed to be written by a man, John T. McKay, received considerably better ratings, a 1.9 average rating from the male reviewers and a 2.3 rating from the women reviewers. Reviewers thought J.T. McKay was a woman trying to disguise her female identity and the ratings reflect this belief. Since it is assumed reviewers are genuine about their work, these differential ratings for work perceived to be done by men and women support the belief that there are indeed cultural pressures influencing our impressions and evaluations of men's and women's work, however unintentionally we may be allowing this to happen. Other studies such as the Fidell (1970) and Lefkowitz (1979s) reveal the same differences.
Sandler(1986) concluded that despite the perception that inequities had been removed, parity was still a distant goal. This conclusion remains valid today. There have been considerable gains, most notably in women's access to undergraduate and graduate education. However, once a woman enters the workplace, she soon discovers that her male counterparts are moving ahead of her. She is making some progress, but much more slowly than for equivalent work from men. Furthemore, if and when she does move up the ladder, she continues to discover that the gap between her gains and rewards widens as her own accomplishments relative to those gains increases. Or, as Persell(1983) concluded, there is a positive correlation between performance and rewards for men but a negative correlation for women. This lack of equitable progress is intrinsically linked with (cultural) perceptions that a woman's work is deemed to be of less importance. It is this feature of the landscape that needs to be addressed before true equity will prevail. Otherwise, as in the U.S., the initial gains will likely level off before parity has been achieved. Only with determined effort will the playing fields be levelled and open to all.
Thanks to the School of Mathematical Sciences at the Australian National University where this report was prepared are acknowledged.
University of Georgia