Plugging the Leaks in the Scientific Workforce

Much more needs to be done to reverse the high rate of attrition of both men and women early in their scientific careers.

In response to the dramatic decline in the number of U.S.-born men pursuing science and engineering degrees during the past 30 years, colleges and universities have accepted an unprecedented number of foreign students and have launched aggressive and effective programs aimed at recruiting and retaining underrepresented women and minorities. Since 1970, the number of bachelor’s and doctoral degrees earned by women and minorities has grown significantly. Despite these efforts, however, the science workforce remains in danger. Although we have become more successful at keeping students in school, we have paid relatively little attention to the success and survival of science graduates–regardless of race or gender–where it really counts: in the work world.

The numbers documenting occupational exit are striking and alarming. Data collected by the National Science Foundation (NSF) in the 1980s (Survey of Natural and Social Scientists and Engineers, 1982-1989) reveal that roughly 8.6 percent of men and 17.4 percent of women left natural science and engineering jobs between 1982 and 1989. A study that follows the careers of men and women who graduated from a large public university between 1965 and 1990 (the basis of my book) further confirms this two-to-one ratio. For science graduates with an average of 12.5 years since the highest degree, 31.5 percent of the women who had started science careers and 15.5 percent of the men were not employed in science at the time of the survey. Estimates from more recent NSF surveys conducted in the 1990s (SESTAT 1993-1999) give similar trends for more recent graduates and further show that, for women at the Ph.D. level, occupational exit rates from the natural sciences and engineering are double the exit rates from the social sciences.

This magnitude of attrition from scientific jobs is especially troubling at a time when, even outside the scientific community, there is a growing awareness that a productive and well-trained scientific workforce is essential to maintaining a technologically sophisticated, competitive, and growing economy. In addition, exit from the scientific workplace is often wasteful and inefficient for the people involved. Individuals who have personally paid for a scientific education often turn to occupations in which their learned skills are not nearly as valuable. The social return on educational investments by the federal government also falls, and institutions that lose scientific employees cannot benefit from their often extensive investments in training.

A better understanding of why people leave scientific careers should ultimately lead to changes in the science education process and in the scientific workplace: modifications that will reduce attrition by both improving the information flow to potential scientific workers and making the scientific workplace more hospitable to career men and women. Such a body of knowledge is also likely to result in workplace enhancements that make science careers more attractive to high-performing educated men and women. Therefore, understanding exit is not only a good defense against attrition but also a valuable component of the strategy to increase the attraction and desirability of science.

The four major reasons for leaving science cited by survey respondents in the study are lack of earnings and employment opportunities, inability to combine family with a scientific career, lack of mentoring, and a mismatch of respondents’ interests and the requirements of a scientific job. A secondary reason involves the high rate of change of scientific knowledge, which leads to many temporary exits becoming permanent as skills deteriorate from lack of use. The factors behind exit separate along gender lines, with men overwhelmingly leaving science in search of higher pay and career growth and women leaving as a result of one of the other three factors, which often contribute to an overall sense of alienation from the field. Policy prescriptions can be organized according to the four factors, but because the factors are interrelated, any one policy action is likely to address multiple causes of exit. Similarly, the policy prescriptions need not be directed toward increasing retention of one gender or the other, because any proposal that enhances the attraction of scientific careers will benefit all participants in science.

Unmet expectations

Unmet salary and career expectations have become an important issue for U.S. scientists in the past 40 years. Early career progress has increasingly stalled, with multiple postdoctoral positions replacing permanent employment and scientific salaries dwarfed by those of other professionals such as doctors and higher-level business people. At the same time, financial success as a goal in itself has become more attractive and, in the 1990s, increasingly attainable in the management professions. Because a large portion of the scientific labor force is employed by government and nonprofit organizations, it is unlikely that salaries, especially at the high levels, will ever be competitive with top managerial salaries. To combat unmet expectations, information about careers must become more comprehensive and up to date. Students choosing scientific majors must know what types of careers they are being prepared for and what salaries, opportunities, and responsibilities they can anticipate.

Ideally, a government agency such as the National Science Board or a professional association should periodically conduct workforce surveys by field, with reports on job options, salaries, and salary growth for scientists with differing levels of education, within differing fields and specialties, and in varying cohorts. The cost and time of such studies can be drastically reduced by using Web technology. Reports on the studies should then be disseminated to all institutions of higher education, so that individual departments can post the results on well-publicized career Web sites for their students. Updating the studies frequently could help keep students well-informed as they progress through their studies.

Once students go into professions with their eyes open, the match between the individual and the career is more likely to be successful. People choosing science careers will be those who value scientific work enough to forego income earned elsewhere. However, even with excellent information, there still will be individuals whose needs and preferences change during their lifetimes, so that they may feel the need to leave science for higher-paying occupations. Improving information collection and flow will not solve the problem of unmet salary expectations completely, but it will go a long way to reduce its severity.

Second and equally important, pay and benefits for postdoctoral positions must be set at acceptable levels. In 2001, the annual salary for a first-year postdoc funded through the National Institutes of Health (NIH) was just over $28,000. Furthermore, outside of NIH there is a lot of variability in pay across fields and institutions. Most postdoctoral scientists are in their late 20s through their mid-30s, a time of life when many individuals are forming families. Low pay can create stress. With the increasing dependence on postdoc positions for early employment opportunities, especially in the biological sciences, low pay is discouraging young scientists from pursuing Ph.D.-level careers. Because many postdoctoral positions are financed by federal grants from NSF, NIH, and the Department of Defense, it is up to these organizations and the science community to educate Congress about the importance of acceptable salaries and to budget for them. The situation has improved slightly with NIH’s commitment to increase annual stipends for entering postdocs to $45,000 over a number of years. As of 2004, annual stipends for first-year postdocs had climbed by $7,600 to $35,700. But this one-time increase will not be enough. A regular review is needed to ensure adequate salaries for the scientific elite.

There is little evidence that imaginative career development and compensation schemes are being used in the scientific workplace.

Well-thought-out and imaginative compensation schemes and career trajectories can be important tools for motivating and retaining existing employees, but there is little evidence that these tools have been wielded in the scientific workplace. Compensation-for-performance schemes are notoriously difficult to design in organizations that are not driven by profits and for employees who work in group settings and whose satisfaction is not tied solely to income. Because scientists find satisfaction in a host of nonmonetary attributes that include prestige, creative freedom, intellectual recognition, and responsibility, such attributes can be used to reward performance. But desired performance must be articulated and measured with care, and rewards must be continually reevaluated for relevance to the employees targeted. Deferred benefits or benefits that grow with seniority are elements of a compensation scheme that would encourage a continuing employment relationship. Because steep career trajectories and greater opportunities are luring scientists into management jobs, scientists seem to want not just more money but also the promise of broadened responsibilities as their tenure with an employer increases. Designing compensation schemes for scientists that reward both good performance and longevity might go a long way toward quieting complaints about the lack of opportunity in scientific careers. Here, private companies with more flexibility in how they spend their resources should take the lead, but the government and nonprofit organizations will have to follow suit in order to stay competitive in the labor market.

Balancing career and family

Family issues arise at different stages of family formation for scientists with different career aspirations. The issue of job location for the married couple is often a stumbling block for Ph.D. scientists who are anticipating an academic career, whereas master’s- and bachelor’s-level scientists, whose jobs are not so specialized, can find jobs in business and government in vibrant urban areas, although they often have trouble combining work and small children because of the more rigid work hours and policies that these jobs entail. Policy to address family issues, therefore, needs to come in a variety of forms.

Dual-career issues are especially thorny for Ph.D. scientists for a number of reasons. First, universities are geographically dispersed. Second, because of large space needs, universities are often built in non-urban areas that do not have vibrant labor markets outside of the university. Third, the early Ph.D. career, which often coincides with marriage and partnership, frequently requires several geographical relocations before a permanent job is secured. Finally, the compromises of the dual-career marriage are disproportionately made by female scientists, who are more likely than their male counterparts to be married to an employed professional and who are likely to be younger and less established than their spouses. Relocating universities is obviously not an option. Still, especially within out-of-the-way university communities, there can be stronger efforts to employ spouses of desired job candidates. Currently, such efforts are most often observed for star candidates, and often the spouse’s job offer is a step down in the career trajectory. Increasing the coverage of such efforts and ensuring that job opportunities for spouses are attractive on their own terms would help ease the problems. However, these programs can only be successful with considerable administrative support, because departments do not usually have the know-how or resources to put together a joint package alone.

We need to reexamine the requirement that Ph.D. scientists make a number of geographical moves in the early stages of their careers as they learn from different scientists in graduate school and postdoctoral appointments. With the increasing ease of communicating and traveling, long-distance collaboration and short-term collaborative research experiences might substitute for numerous geographical relocations. The extent of this substitution will necessarily differ by discipline and is likely to depend on the type of lab work performed and the extent to which researchers are tied physically to their laboratories. Because scientific career paths are well established and deeply entrenched in the scientific culture, change is not going to come easily. Furthermore, change will not come about at all unless it is supported by leaders of the scientific community.

Discipline-based associations, together with the National Academy of Sciences, should commission panels to study alternative ways to teach Ph.D. scientists. In the biological and health sciences, biotech firms seem to be offering alternative career paths already. Many firms will hire an employee after graduate school, providing a postdoctoral position that often leads to a permanent position. These relatively permanent employment opportunities in urban settings create solutions to dual-career problems. Elizabeth Marincola of the American Society for Cell Biology and Frank Solomon of the Massachusetts Institute of Technology have proposed creating staff scientist jobs in university laboratories for scientists who are looking for a more permanent and predictable employment situation Although both of these options would be helpful, there is concern that because the female scientist is more likely than her male counterpart to find a solution to the dual-career marriage, there is a risk of a two-tier work force in which women take the predictable and permanent jobs and men choose the riskier and more prestigious academic route. Leaders in the academic community need to address these issues regarding the academic career path, because past experience has shown that such a gender-based allocation of scientific talent has not been conducive to attracting women into scientific pursuits.

Policies that help to balance the demands of child rearing and a scientific profession are likely to improve the quality of life and the productivity of all scientists who take on both career and family responsibilities. Employing institutions have many options to improve the quality of life of working parents, including but not limited to maternity/paternity leave, increased flexibility of work hours, telecommuting, unpaid personal days for childhood emergencies, a temporary part-time work option, and onsite day care. These reforms are crucial for the success of working parents in all areas of employment, not just in science workplaces, and if media coverage of workplace benefits can be trusted, such reforms have become more commonplace throughout the economy since the early 1990s.

Although Ph.D. scientists in academia often find that the flexibility and autonomy that these policies create help to coordinate child-rearing demands, the flexibility is often an illusion in the early years when, working for tenure, the scientist is putting in 60-to-70-hour weeks. For these scientists, such childcare benefits improve the quality of the individual’s work life but do not diminish the work time necessary to attain tenure. A policy increasingly being considered in academia–giving a parent extra time on the tenure clock for each child born during its duration–allows the working parent the opportunity to make up for some of the research time lost to early childhood parenting and to spread the time in research over a larger span of calendar years.

Together, these policies will help with the day-to-day strains of working parents, and mothers in particular, but they are not enough. Scientists employed at research universities believe that taking extra time off after giving birth and stopping the tenure clock is the kiss of death to one’s career (as the study that is the basis of my book makes clear), and the small number of faculty who do take advantage of this benefit are overwhelmingly female. Scientists do not trust that these activities will be viewed neutrally in the tenure decision. The result is that some women delay childbearing until after the tenure decision, which can be a risky strategy for a woman who wants a family; others take only a minimal maternity leave and return to work to compete as if they were childless; and still others take advantage of the benefit and hope that the gains of the extra time will outweigh any negative perceptions. The distrust felt by these women means that if such benefits are put in place in a college or university, the administration must stand by them and make sure that those who control tenure decisions support them as well. A special committee should be set up to review each tenure case in which the individual has taken advantage of a childcare-related benefit that gave the parent extra time away from teaching or the tenure clock. Ensuring that such activities are not penalized during promotion decisions is paramount to the success of working parents.

More generally, there are two important issues in the realm of work and family. The first is that there is a predominant feeling in the scientific community (and in society generally) that child rearing and careers are in direct conflict and that one has to be compromised for the other. Second, expectations are that women will make this compromise. Because employers assume that women will eventually take time off to care for children, they are likely to give them reduced opportunities early in the career. Once career options are lessened, the decision to put child rearing ahead of work is much easier. Thus, the prophecy becomes self-fulfilling. Claudia Goldin’s finding that only 13 to 17 percent of the college-educated women who graduated in the late 1960s through the 1970s had both a family and a career by age 40 is striking evidence of the fulfillment of these expectations These two issues are difficult to address, because both are based on longstanding cultural norms concerning work, family, and gender roles. The U.S. workplace encourages competition and rewards stars with money, prestige, and opportunity. Technological developments that have recently increased labor productivity have had little impact on the child-rearing function, which offers no acceptable substitute for adult/child personal contact; therefore, child rearing is becoming increasingly expensive to U.S. employers. Because child rearing does take time from work and career development, even for full-time employees, the stars in the U.S. workplace in fields as diverse as business, science, and the arts are not likely to be men or women who spend a lot of time with children and family.

Both issues will become less problematic when men start taking on an increased share of childcare. Once this happens, childcare will be given higher status, and policies to help balance work and family will be given more attention. Furthermore, men and women will be treated much more equally in the labor market. If, in some ideal world, 50 percent of the child-rearing responsibilities were taken on by men, employers would not have differential expectations about the long-term commitment to work of men and women. Women and men would be given the same career opportunities leading up to childbirth and before the child-rearing choices have to be made. Although there may have been some change in the gender allocation of childcare during the past 30 years, data reveal that men still take on only a small portion of child-rearing responsibility. Even men who might be interested in staying home with children for a spell often resist taking advantage of policies such as paternity leaves, which they feel send the wrong signals to employers. Therefore, change will only occur if upper-level management in these employing institutions gives credible promises that there will be no negative repercussions in response to decisions to take advantage of childcare benefits.

Other advanced countries, Sweden most dramatically, have national policies aimed at equalizing male and female participation in both child rearing and work. Sweden’s Equal Opportunity Act of 1992 requires employers to achieve a well-balanced sex distribution in many jobs and to facilitate combining work and family responsibilities. Paid maternity or paternity care, at 80 to 90 percent of the salary, is mandated for 12 months. Sweden falls short of requiring men to take some part of this 12-month leave, but statistics show that about 70 percent of fathers take some time off and that these leaves have recently been getting longer. Given the contentiousness that marked congressional debate and approval of the Family and Medical Leave Act of 1993, it is unlikely that this type of workplace policy will be replicated in the United States.

The mentoring gap

Lack of good mentoring is more problematic for women than men, because women are less likely to be mentored than men and because the effects of mentoring on retention and performance are greater for women. Sex disparity in mentoring is greatest in academic institutions, where mentoring tends to be quite informal and thus arises naturally between male professors and male students. With more female professors, female students may find that developing a mentoring relationship is becoming easier. However, because the sex ratios of science professors continue to be highly unbalanced, formal mentoring programs for female science students, which have been growing in number during the past 10 years, should continue to be set up and supported in all academic institutions. Then women who are having trouble developing a personal relationship with a professor can be directed to professors or graduate students who are willing to take on the role of mentor. A variety of universities now use a program called multilevel mentoring, in which a junior biology major may mentor a freshman and also be mentored by a postdoc. Such a program creates a network of women to whom individuals can turn with questions. Social occasions for participants have also been successful in making the relationships more personal and developing ties with a whole community of women in science. These activities need not be limited to women although, because of the ease with which men seem to develop these relationships in academe, female mentoring programs may be sufficient.

In industry, men and women are equally likely to be mentored, and mentoring relationships generally develop in organizations in which mentoring is the cultural norm or where formal mentoring programs have been put in place. Again, for these institutions, mentoring programs are most likely to take hold when upper-level management puts its weight behind them.

The individual/field mismatch

Mismatches between an individual’s interests and the requirements of the scientific career are addressed in some of the policies advocated above. Good career counseling for degree recipients in the different scientific disciplines is likely to ward off bad matches that result from uninformed expectations. Mentoring relationships and well-developed networks of scientists with similar interests are likely to increase the personal connections that a given scientist makes with other scientists, thus reducing feelings of isolation. The trend toward interdisciplinary work during the past 20 years should give the individual scientist the opportunity to choose areas of work in which the science itself can be connected to a bigger picture. NSF and private foundations such as the Alfred P. Sloan Foundation have taken the lead in funding broad multidisciplinary research efforts. However, universities have historically had fairly rigid disciplinary boundaries. In order for scientists to feel free to participate in these interdisciplinary projects, the reward and promotion processes of employing institutions may have to be restructured to value this type of research.

In response to the finding that permanent exit is higher for men and women who are in fields that are changing at rapid rates, institution-sponsored skill update and training programs can help alleviate stresses associated with change. NSF sponsors programs for women who have left science to help them rebuild skills for reentry. These types of programs help ensure that temporary exit remains temporary. Training programs and skill updates are especially important in academic institutions, in which separation is not an option for tenured employees who feel that their skills have become out of date. Organizations such as the Mellon Foundation have been instrumental in supporting programs of career development for professors at all levels in liberal arts colleges. Many private companies do not engage in wide-scale training of existing employees in new techniques and knowledge. Companies may be comfortable with the loss of older employees who are not willing to update their skills because new employees, fresh out of the university, already have the updated skills and are cheaper than more senior employees. But if the pool of new hires becomes insufficient to replace this attrition, companies will have to face this issue head on.

Pursuing a science career should not be a matter of choosing hardship and sacrifice. In addition to interesting and challenging work, science careers should offer a strong support network, the possibility of having a real family life, an income throughout the career that allows a comfortable family lifestyle, and possibilities for continuous advancement and development. Currently, many scientists feel that science careers are falling short in one or more of these dimensions, both in absolute terms and in relation to alternative careers that are attracting bright and talented young men and women. The full scientific community in combination with government policymakers must mobilize for change in the scientific workplace. The future of the United States as a world power depends on their success.

Your participation enriches the conversation

Respond to the ideas raised in this essay by writing to [email protected]. And read what others are saying in our lively Forum section.

Cite this Article

Preston, Anne E. “Plugging the Leaks in the Scientific Workforce.” Issues in Science and Technology 20, no. 4 (Summer 2004).

Vol. XX, No. 4, Summer 2004