It’s Complicated—In Fact,It’s Complex

Major Findings

The most important contribution of the Ceci et al. report is its separation of science and mathematics into fields that are spatial-mathematics intensive (geoscience, engineering, economics, mathematics/computer science, and the physical sciences—GEEMP fields) and those that are not (life science, psychology, and social science—LPS fields). Men are overrepresented in GEEMP fields, and women are overrepresented or at parity in LPS fields. Thus, any program of study or proposed explanation of gender differences in mathematics and science needs to specify the field of study. This distinction should change the nature of future research. We can no longer talk about gender gaps in science, technology, engineering, and mathematics (STEM fields) as though they are homogeneous across disciplines. The spatial-mathematics designation adds economics, a social science, to the list of fields that have a preponderance of men, and it moves life science to the other side of the ledger in which women either predominate or are achieving at rates that are comparable to those of men. Additionally, this division focuses attention on spatial and mathematical skills as areas that are critical in understanding male-versus-female participation rates in academic science.

The authors have not hesitated to challenge some popular beliefs, most notably the idea that sex discrimination in hiring and promotion in the academic community can explain a reasonably large portion of the gender gap in academic careers. Their conclusion is likely to be met with criticism and denial because it is a central theme in research on gender gaps in scientific careers. It is important to note that the authors do not deny that discrimination exists, but it is more likely to occur earlier in the career paths of women and men, not at the time they enter (or attempt to enter) academic careers. If this finding holds up over time, researchers will have to move their focus on discriminatory practices down to younger ages, when children are developing their academic and career interests. If this pending controversy leads to the collection of better data, it will help to move our understanding of the gender gap forward.

Another major strength of the Ceci et al. report is the authors’ examination of many types of explanations, including those that are rooted in biology and the social system, and the conclusion that their explanatory power varies across developmental and historical periods. The research literature on questions about gender ratios in careers is vast, but there are few studies that have crossed levels of analysis and actually examined the question of gender gaps at the biological, individual, social, and cultural level simultaneously. Of course, no single researcher can be knowledgeable about the influences that occur at all of these levels, but teams of researchers can track individual development and choices using the multiple lenses of psychology’s subdisciplines. Future research will need to integrate data from all of these levels.

The Broader Questions

Ceci and his coauthors ask a deceptively simple question: “Why are women underrepresented in some STEM disciplines?” But it is a question with many corollaries, including, “Why are men underrepresented in education, social work, nursing, and other helping professions?” The question of why there are lopsided gender ratios in academia is a smaller part of a larger question: Why are most occupations, including those of truck drivers, secretaries, chief operating officers, home real estate agents, and auto mechanics, to name a few, largely segregated by sex? All questions about gender segregation in the workplace need to account for the way gender ratios have changed, especially in the last 10 to 20 years. How can we understand the dramatic changes in the past few decades, such as (near) gender parity in medical-school graduation rates, the large female advantage in veterinary-school graduation rates, and increases in the proportion of women in all areas of academia, even areas in which they remain underrepresented?

In understanding gender gaps, some variables can be useful across occupations and others will be specific for one or few occupational gaps. For example, it is likely that the male advantage in upper-body strength is important in determining why men are much more likely to be furniture movers than women are, but upper-body strength is not important in understanding the predominance of women in clerical positions or of men in theoretical physics. The fact that women have the majority of caregiving responsibilities for children and the elderly is surely an important factor in understanding the preponderance of women in kindergarten through high school education, with school days ending in the late afternoon and summers off from work, but this is unlikely to be an important factor in explaining the gender ratio in clerical work because such workers traditionally have little flexibility in how or when they work. In short, the questions about gender gaps in academic science and in other occupations are complex, and thus would benefit from the growing field of complexity science.

Complex-Systems Analysis: A New Look at an Old Problem

Perhaps the best summary of reasons for lopsided gender ratios in achievement in academic science is that “it’s complex!” The answer to questions about gender ratios will depend on a multitude of variables that combine in complex, nonlinear ways. Psychologists can look to our sister disciplines and borrow an analytic approach that is designed to help us understand complexity that is embedded in a system. A systemic approach allows us to take a holistic worldview with the understanding that “everything” is connected. Complexity science is based on a set of defining attributes for complex systems. It is a field that is growing in popularity among our sister disciplines but has not been widely used in psychology.

Complex-systems analysis is one approach that offers the promise of enhancing our understanding of gender ratios in academic science and other careers. It has been used to model and predict behavior in areas as diverse as the flight patterns of flocks of birds, the distribution of consumer goods in a community, the action of microorganisms in a human body, and the way ant colonies find food, to name a few. The underlying idea is that we need to consider the system in which a phenomenon operates. Like gender ratios in occupations, each of these systems has outcomes that emerge from the actions of independent “agents.” Think of every female and male as an independent agent who makes myriads of choices and is subject to a countless array of environmental pressures that influence career outcomes. For example, some children are naturally active, early crawlers who explore their environment and thus learn about movement in space. Other infants who develop motor skills many months later may learn to verbalize their desire for a toy rather than moving toward it. Small initial differences in early development and experience can be maximized or minimized depending on how environmental forces respond to these differences. The child who asks for a toy rather than getting it for himself may be talked to more often because that seems to be his preferred mode of interacting. Throughout life, individual, social, and cultural variables interact in complex ways.

Another premise of this complex-systems approach is that small initial differences in the actions of independent agents can have large effects. Individual choices in course-taking patterns, leisure activities, role models, and opportunities for learning can result in large gender ratios at a later point in time. In a complex system, the independent agents are connected and influence each other. In the context of the development of gender gaps in the workplace, the individual choices of boys and girls depend on the choices that other children are making. So, if many girls spend their time on stereotypical female activities (think about getting their nails done or spending long stretches of time doing their hair), these actions influence the choices of girls, who learn that these are desirable activities, and of boys, who learn to eschew these activities and use their time for other activities—often activities such as single-shooter video games or basketball practice. Many male-typical activities are more likely than activities engaged in by girls to develop spatial skills. Although there are female-typical spatial activities, such as home decorating and sewing, these activities are less popular among girls than the male-typical spatial activities are among boys.

Complex systems are adaptive—they respond to changes. This central feature of complex systems is what makes them distinct from systems that are merely complicated. The agents in a complex system are self-organizing, which means that there is no leader telling individual boys and girls about the sorts of choices they must make in order to create or maintain gender ratios in the workplace—academic or otherwise. For example, changes in marriage demographics, with people (on average) marrying at later ages than in earlier generations and an increasing percentage of adults remaining single, are likely to affect the career options that men and women pursue. Salaries tend to be higher in the spatial- and mathematics-intensive fields of academic science than in the life sciences. As the salary differential climbs, these more lucrative choices should attract more men and more women and, in the process, may alter gender ratios as the most qualified compete for desirable jobs.

There are also tipping points in complex systems. In general, occupations that are predominantly female tend to pay less than those that are predominantly male, so as the ratio of females in a discipline rises, one prediction is that salaries will decline as the proportion of women exceeds a tipping point. Mathematical models can be generated that demonstrate the effects of different tipping points, so we can visualize what might happen to salaries as the percentage of women in civil engineering, for example, moves to 30%, 40%, or 50%. Complex systems adapt via feedback mechanisms, which can be used in simulations. Positive feedback would continue to increase the number of women going into GEEMP fields; negative feedback would create an equilibrium in which the percentage of women and men would reach a steady state (not necessarily 50%, but some fixed percentage). Thus, with this model, we could model various tipping points to determine, for example, what would happen if the tipping point at which the number of women in GEEMP fields began to decline were 60% or some other value, or if we could attract more women to these fields if the proportion of women faculty were raised by varying amounts. There is a long list of possible models that could be tested with a complex-systems approach.

A complex-systems approach to the question of gender gaps in academic science allows us to model assumptions about important variables and how they combine. This sort of understanding can help us design more effective interventions, if that is a desirable goal (an important question about values and assumptions), and to predict how gender gaps will wax or wane in the coming decades. Ceci and his coauthors have laid out a framework of critical variables that affect gender ratios. Personal decisions about how to use these data will range from “do nothing” to “design interventions that would eliminate gaps.” Regardless of where individuals stand on philosophical questions about the desirability and alterability of gender gaps, we are now better informed about “what is.” Questions about “what should be” are more difficult to answer.