Understanding Interventions That Encourage Minorities to Pursue Research Careers: Summary of a Workshop (2007)

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.

2 Examples of Previous Research R esearchers have examined the processes involved in deci- sions to study science in college, enter graduate school in the sciences, and become a scientist. These research programs originate in a variety of disciplines and can have quite different perspectives, but they also complement each other in explaining the complex processes involved in making educational and career choices. Because many of those interested in these questions are biomedical researchers who may not be steeped in social science viewpoints on these issues, the planning committee constructed an early session in the workshop to provide a varied set of lenses for participants to think more broadly about this kind of work—and to help consider themes for future study. The perspectives offered at the workshop—and in this summary—do not provide an exhaus- tive set, but they help to provide a broader set of questions and approaches for thinking about these issues. Social Cognitive Career Theory Social cognitive career theory (SCCT) is an integrative theoreti- cal framework that explores the psychological and social factors that produce personal interests and lead to choices related to education and careers. The theory is also concerned with the network of factors that affect performance and persistence in a person’s educational and career paths and those that are responsible for an individual’s satisfaction in a particular job.  

 UNDERSTANDING INTERVENTIONS Personal interests are not the only factors that drive educa- tional and career choices and can be trumped by family expectations or other external influences. But interests are “strong motivational drivers of the choices that students make in their educational and career lives,” said Robert W. Lent, professor of counseling and per- sonnel services at the University of Maryland, College Park. Lent described his work on applying SCCT to the issue of expanding the science, technology, engineering, and mathematics (STEM) pipeline, noting that it serves as a template with which to view and develop interventions designed to encourage minorities to enter research careers. SCCT draws heavily from the more gen- eral social cognitive theory of the Stanford University psychologist Albert Bandura. The key construct in Bandura’s work is the concept of self-efficacy—people’s beliefs about their ability to perform specific behaviors or actions. In particular, it refers to domain-specific con- fidence in particular situations, not to self-confidence as a general trait. In the context of science and mathematics, said Lent, self-effi- cacy addresses “the fundamental question: Can I do this thing? Can I, for example, do well in math and science courses in middle or high school? Can I do well in a science or engineering-related major in college?” Self-efficacy beliefs, in turn, derive largely from four sources, according to Bandura’s theory. The first and most important is prior performance—the levels of mastery or failure that people have expe- rienced. “If I have done well in the past at a particular academic subject, for example, I am likely to expect in the future that I can do well in it as well,” said Lent. “Conversely, if I’ve not done so well, my self-efficacy beliefs are going to drop.” The three other sources of self-efficacy beliefs are also important. One is observations of others’ learning or the experience of models, especially models that one perceives as being similar to oneself. “For example, in terms of race, ethnicity, gender, social class, and so forth, viewing our models as being efficacious at things we want to do is a good way of raising self-efficacy—or lowering it, depending on the nature of the model,” Lent said. Another source of self-efficacy beliefs is the social messages that encourage or discourage participation in an activity. Students receive many messages from others and from the mass media that can influence their confidence about a particular activity. “But talk is cheap,” Lent reminded the workshop participants. “Sometimes, if we try to convince people that they are good at things that we are not so sure they are—or that their own performance experiences 

EXAMPLES OF PREVIOUS RESEARCH  disconfirm—then the source of that support may not be credible for very long, and people may not persist.” A final source of self-efficacy beliefs is physiological and affec- tive reactions. For example, if a person is so anxious in taking every math test that he or she does poorly, that person is likely to infer that math is a personal weakness. “So test anxiety can be, in that example, a negative influence on self-efficacy,” said Lent. This is one way in which gender can influence self-efficacy beliefs in science and mathematics, Lent noted. In a general popula- tion of college students, women at the college level and below tend to report significantly lower self-efficacy beliefs at math compared with men. However, the same tendency is not exhibited in more spe- cialized populations, such as engineering students. Also, if women have had similar experiences to men in terms of the four sources of self-efficacy beliefs, they tend to have the same self-efficacy beliefs as men. Interest in Bandura’s theory follows from a number of other factors, including the expectations surrounding particular outcomes. As Lent said, these beliefs “address the question: ‘If I do this, then what will happen? If I major, for example, in science or engineering, or if I choose to pursue a research career, what will be the outcomes? What will be the payoffs for me, and what will be the negative con- sequences? What will the salary be like? What will my co-workers be like? Prestige? Autonomy?’ These refer basically to career values that people want to fulfill.” Another factor is the goals that motivate people to engage in a particular activity or produce a particular outcome, such as trying to get an A grade in a particular math or science course. According to Lent, “Goals address the fundamental question of ‘how much do I want to do this course of action?’” Finally, within the theory, there are various kinds of social, financial, emotional, and other contextual supports and barriers that people encounter while pursuing their goals. These supports and barriers address the question of “‘how will the environment treat me if I try this particular course of action,’” Lent said. For example, “the phrase ‘chilly climate,’ which oftentimes refers to the experience of women and certain minority groups in science and engineering fields, refers to the perception of environmental barriers.” The importance of self-efficacy beliefs is often obvious among students studying science and engineering, Lent noted. For example, he has seen many students who did extremely well in high school lose confidence when they got poor grades on their first college mid- term examination. “All of a sudden their confidence levels plum- 

10 UNDERSTANDING INTERVENTIONS meted, and they were convinced they were in the wrong field,” Lent said. “They had never gotten Bs, or worse, before, and all of a sudden it was time to change majors and career paths.” The ongoing experience of success or failure subsequently modi- fies or stabilizes self-efficacy and outcome beliefs “in a never-ending loop,” said Lent. Changes to these beliefs also can occur through outside influences like “technological advances, parenting, and other life experiences that may formulate changes in interest patterns because of their impact on self-efficacy and outcome expectations.” Each person has what SCCT theorists call “person inputs”— factors like personality, ability, gender, race, ethnicity, disability, and health status. These factors interact with background factors such as social class and the quality of early educational experiences. “Depending on who one is, and what one looks like, the environ- ment may selectively provide or withhold certain opportunities,” said Lent. Lent and others have applied this framework in several major research projects. In a study of students at a predominantly white university, Lent and his colleagues found “that SCCT variables were well predictive of goals and actual persistence in engineering over a three-semester sequence.” This model was equally good at predicting choice and persistence goals in engineering majors when extended to two historically black universities. Lent and his coworkers are now conducting a large-scale longitudinal study of computer science and computer engineering students at multiple predominantly white and historically black colleges and universities around the country. This theoretical work has suggested particular intervention points and approaches, according to Lent. One possibility is to work at expanding vocational interests, especially in high-aptitude areas, and “getting people to rethink areas they might be able to do well at but have prematurely foreclosed upon because they don’t believe they have the ability to do well or don’t know enough about the field to want to pursue it.” Other options are clarifying career goals, sup- porting career goals, strengthening self-efficacy, instilling realistic   R.W. Lent, S.D. Brown, J. Schmidt, B. Brenner, H. Lyons, and D. Treistman. 2003. Relation of contextual supports and barriers to choice behavior in engineering ma- jors: Test of alternative social cognitive models. Journal of Counseling Psychology 50: 458-465.   R.W. Lent, S.D. Brown, H. Sheu, J. Schmidt, B.R. Brenner, C.S. Gloster, G. Wilkins, L. Schmidt, H. Lyons, and D. Treistman. 2005. Social cognitive predictors of academic interests and goals in engineering: Utility for women and students at historically Black universities. Journal of Counseling Psychology 52: 84-92. 

EXAMPLES OF PREVIOUS RESEARCH 11 outcome expectations, and helping people to manage environmental barriers and build effective support systems. Past work has also emphasized how much more could be learned through further research that applies this model. A particular need, said Lent, is for more longitudinal, multiyear, and multisite research. Also, according to Lent, the basic theory needs to be studied in rela- tion to women and underrepresented minorities in STEM fields, and more theory-based interventions and experimental studies are needed. “There has been some of this and I think it holds much promise for the future, but we need much more of it,” said Lent. Lent noted, by the way, that many individuals and groups out- side of academia are interested in applying this approach to the issues they face. In addition to his university position, Lent is a visiting scholar at the U.S. Department of Homeland Security, and he said that the department views this issue as important to national security as well as economic prosperity. “There are some things that we probably don’t want to outsource to other countries,” Lent said. Human Capital Theory Another theoretical perspective, this one rooted in economics, is known as human capital theory. As described by Anne Preston, asso- ciate professor of economics at Haverford College, each individual has a stock of skills, knowledge, abilities, and other characteristics that determine his or her wage-earning potential. Individuals can invest in increases in their own human capital through education, on-the-job training, and other activities. “Human capital theory basi- cally allows us to understand under what circumstances an indi- vidual will decide to invest in further acquisition of human capital and [in] what types,” said Preston. “So you can think of it as a pure cost-benefit calculation made by what we in economics always talk about—the rational and perfectly informed actor.” Of course, as Preston noted, “we do understand that not every individual is totally rational or perfectly informed.” Costs, which Preston said are relatively easy to estimate, are for tangible expenses— such as tuition, room, board, books, and foregone earnings—and they occur at the time of the investment. Benefits, which can include future wages and future income streams, in contrast, can be much harder to predict. In addition, these cost-benefit calculations often require dis- counting future income versus current costs. “Some people value future income differently than others,” said Preston. “It depends on 

12 UNDERSTANDING INTERVENTIONS your current family income, maybe your family structure, the sorts of needs that your family has in terms of income now versus in the future, and the expected duration of the work life.” Some of these factors can differ for different populations. Finally, economists know that people do not always act in per- fectly rational and perfectly informed ways. Methods exist to take a lack of information or irrational decision-making into account, but these methods may introduce additional levels of uncertainty. Human capital theory can be used to provide insights into how interventions might lead to different decisions, Preston said. For example, mentoring programs can increase knowledge and change expectations. Better job placement programs might lead to better returns on an investment in human capital. Fellowships, research assistantships, teaching assistantships, and other forms of financial support can reduce the costs of the investment. Better information about the opportunities that investments give an individual can make a difference. Methodologically, human capital theory is a strategy in which economists quantify variables and seek to determine the relation- ships among those variables. Some of these variables have discrete values, such as whether a person stays in a field or leaves it, or a per- son’s race, sex, or type of school; others are continuous, like wages. Some variables are measured by proxies, as when the number of publications or number of citations are used as measures of research productivity. Preston explained that economists add variables to an analysis with the goal of explaining away the effect. If all vari- ables that can be identified—except for the one under study—fail to explain away the effect, researchers have an indication that the variable under investigation plays an important role. Economists also try to measure the quantitative effects of inter- ventions. If mentoring programs are thought to make a difference, for example, economists will try to analyze whether being men- tored influences the probability of investment in human capital. This could be done for majority and for minority students to see if there are differences in the effects of mentoring. Studies such as this introduce what economists call “selectiv- ity.” If the individuals being mentored differ from those who are not mentored in some important way, the effect ascribed to mentoring may actually arise from personal characteristics, not the mentoring program. Economists can try to reduce these effects using various complex mathematical techniques, but Preston said that “person- ally, I find them not very reliable or satisfying.” An alternative, she said, is “to move from these big national data sets [to create] indi- 

EXAMPLES OF PREVIOUS RESEARCH 13 vidualized data sets.” Approaches such as randomized trials, where individuals are selected to receive or not receive an intervention and the effects of the intervention measured, diminish issues of selectiv- ity; however, they are seldom feasible and may even be unethical in such a setting, where some individuals are prevented from engaging in what is believed to be a positive intervention. Another possibility is to collect data from natural experiments, using existing variation in the population of study. For example, student outcomes could be measured from different schools, some of which have an institutionalized mentoring system and others that do not. Such experiments require thought, time, creativity, and fund- ing, said Preston, but “economists can really make some interesting inroads if they take up this challenge.” Social Identity and Stereotype Threat Claude Steele, director of the Center for Advanced Study in the Behavioral Sciences, professor of psychology and Lucie Stern Professor in the Social Sciences at Stanford University, and his col- leagues have focused their research on two main themes. The first is underperformance in school by groups whose abilities are nega- tively stereotyped in the broader society—an issue closely related to the persistence of members of these groups in pursuing research careers. The second is the set of broader issues posed by diversity. “It is one thing to integrate a school setting or work place,” Steele said. “It is another thing to make that setting a place where everybody seems to flourish—where they feel like they belong.” Unlike many psychologists, Steele stresses the importance of context. “When we talk about schools and other environments of that sort, we tend to think of them as homogeneous environments— environments that are essentially the same for everybody. If there is one thing I hope you get from my remarks today, it is that they are different for people with different identities. The very same rooms with the same pictures on the wall, the same test items, the same teachers, can be very different as a function of social identities that a person has.” Each individual has many different social identities. These iden- tities can be based on age, sex, race, religion, ethnicity, and so on. Different identities generate what Steele calls “contingencies”— reactions by others to a particular identity. “You have to deal with certain things because you have certain identities,” he said. An individual’s social identities can change. During the great migration of African Americans from the southern to the north- 

14 UNDERSTANDING INTERVENTIONS ern United States, an estimated 10,000 to 30,000 African Ameri- cans “passed” from being “black” to being “white,” asserted Steele. “[That’s] what I mean by contingencies,” he said. “They were avoiding the things that went with that identity.” Another example includes changing a foreign-sounding name to one that sounds more American. Some contingencies are threatening. An example might be an African American seeking to excel in an endeavor where members of that group are stereotyped as underperforming. When someone is threatened by the contingencies of a social identity, that person might seek to conceal or disguise that identity. But threats to an identity tend to make it central to your functioning, said Steele. Stereotype threat is a good example of a contingency. Stereo- type threat arises when a person is in a situation where a negative stereotype applies. A good example is women in mathematics. In a series of experiments done by Steele and his colleagues, women and men who were equally skilled in math were given a very difficult math test one at a time in a testing room. Women in this situation tended to underperform. When they experienced the frustration of a difficult test, the stereotype that women have weaker mathematical abilities suggested to women that they may lack ability. Men who are frustrated by the test may also believe that they don’t have the necessary ability, but it’s because of factors other than their male- ness. “So for a woman in that situation, there is extra pressure— especially if that woman cares about math, has high expectations for her performance, or is committed to it,” said Steele. In one recent study, simply mentioning the word “genetics” in the preamble to a math test worsened women’s performance in math. However, when the researchers told the women before they took the test that “for this particular test, women always do as well as men,” the women’s performance was higher than when they were experiencing stereotype threat. Interestingly, the performance of men tends to drop somewhat under these circumstances. “We can be advantaged by stereotypes,” said Steele, describing stereotype   C.M. Steele, S.J. Spencer, and J. Aronson. 2002. Contending with group image: The psychology of stereotype and social identity threat. Advances in Experimental Social Psychology 24: 379-440. C.M. Steele and J. Aronson. 1995. Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology 69(5): 797-811.   S.J. Spencer, C.M. Steele, and D.M. Quinn. 1999. Stereotype threat and women’s math performance. Journal of Experimental Social Psychology 35(1): 4-28.   Dar-Nimrod and S.J. Heine. 2006. Exposure to scientific theories affects women’s I. math performance. Science 314(5798): 435. 

EXAMPLES OF PREVIOUS RESEARCH 15 lift, in which one group can be “on the upside of somebody else’s negative stereotype.” Men may do worse on the test because “it isn’t plausible to them that they lack the ability to do the work. It doesn’t make sense. So the experience of frustration is less. If you take that advantage away from them, . . . then you may see some decrements in performance.” The effects of stereotype threat also were observed among Afri- can Americans taking a test using Raven’s Progressive Matrices, a type of IQ test. When told that the test was to measure IQ, African American students dramatically underperformed compared with white students. But when African American students were told that the test was simply a puzzle, their performance rose dramatically. One remarkable finding from studies such as these is that the strongest students are often more susceptible to stereotype threat. “You have to care about [the domain] to experience stereotype threat,” Steele said. “One protection against stereotype threat is not to care about it. [If you] dis-identify with the domain, then you don’t care that much that your group is negatively stereotyped in that domain because you don’t care that much about the domain.” These studies also emphasize the importance of cues in the environment that accentuate or lessen threats, Steele pointed out. “Cues that signal threatening contingencies foster vigilance,” he said. “They hamper a sense of belonging in the setting, and this in turn impairs learning.” One such cue is the number of other people in a setting with whom you share a social identity. For example, when women are greatly outnumbered by men in taking a math test, they tend to perform worse than if men are absent. This kind of marginalization through small numbers can have a powerful effect on identity threat. The profound segregation that exists on many college campuses can heighten a sense of difference. The effects of cues on attitudes were tested in an experiment performed by Steele in collaboration with Mary Murphy, using stu- dents who were waiting to be interviewed for admission to a sum- mer workshop on science and engineering. While waiting for the interview, they watched a videotape about the summer workshop that showed students working together. In one videotape, men and women were balanced one to one. In the other, men outnumbered the women three to one. The women who watched the video with the unbalanced genders had a much better memory of the inciden-   R.P. Brown and E.A. Day. 2006. The difference isn’t black and white: Stereotype threat and the race gap on Raven’s Advanced Progressive Matrices. Journal of Applied Psychology 91(4): 979-985. 

16 UNDERSTANDING INTERVENTIONS tal details of that videotape. Steele hypothesized that the cues in the videotape were making female viewers aware of their gender identity, which made them more aware of the situation than they would be otherwise. “Think of any time any of you have ever been in a situation where you are one of a kind,” Steele said. “You pay attention.” This awareness has effects not only on memory, but also on physiology. In fact, when the students were hooked up to physi- ological recording equipment (under the pretense that they would need it for a different experiment they would soon undertake), the women who watched the unbalanced videotape had much higher cardiovascular activity than the men. Steele takes several messages away from this research. One is that these kinds of cues and reactions are virtually unavoidable in a diverse society such as ours. “Any diverse setting holds these iden- tity threats,” he said. “This is sort of an American challenge. I think at one level we should be proud of it because we are a society that publicly holds on to the idea that all of society should be integrated. . . . But one of the challenges behind that commitment . . . is making integration work. It is making these settings, these schools, these programs work for a truly diverse population.” Also, these cues do not arise solely from discrimination. On the contrary, he said, they can exist in the absence of discrimination. “These are contextual factors that make identities function in certain ways,” he said. The importance of cues also suggests ways to promote learning. If the cues change, performance can change. The most important change that has to happen, according to Steele, is for women and minorities to have a sense that they belong in a particular setting. “For instruction to work—and for the decisions we want them to make to be made—they have to have a sense of belonging. As a soci- ety, [we have to] understand that that has to come first.” In fact, said Steele, without changing this sense of belonging, other interventions can be counterproductive: “If you push other things, like try to motivate [students], expose them to strong skill-focused programs, without at the same time addressing the sense of belonging, you can really get a backfire effect. Things may not work at all.” One cue is what people say. “What do the university president, the department chair, [and other] people say about the ‘belong- ingness’ of groups? Do they avoid the issue and see it as a minor issue and not something of importance? Or do they really own it and make the proclamation that everybody belongs intellectually in these settings?” he asked. Making the presence of particular groups 

EXAMPLES OF PREVIOUS RESEARCH 17 the norm can relieve the tension in a setting and enable students to feel that they belong. Similarly, a critical mass of people with a particular social iden- tity is also pivotal, claimed Steele. Individuals are always looking around and counting how many other people share their social identities in a particular setting. “People do respond to numbers,” he said. Particular interventions can dramatically shape how students respond to cues. In a study done by Gregory Walton and Geoffrey Cohen at Yale University, African American and white students watched and then discussed a videotape of an African American student talking about how alienated and out of place he had felt at Yale. But the student went on to recount how, during a trip home, his father reminded him what a superb opportunity it was to be attend- ing Yale and that he needed to take advantage of it. The student described becoming active in a singing group and in academics, and he concluded that he was now very happy at Yale and that he was enjoying and learning from Yale’s rich environment. Just watching the videotape and talking about it raised the grade point average of African American students by two-thirds of a letter grade in the subsequent semester. “Why does that work?” asked Steele. “Because it gives [the students] an interpretation of the cues in the environment that [is] hopeful. . . . Everybody has those feel- ings [of not belonging], but if you’re a group that the whole society negatively stereotypes in this way, those feelings are really a weight. So you need an interpretation that makes your sense of not belong- ing normal. This guy in the videotape makes it normal, and then he offers light at the end of the tunnel. Wow.” In another intervention, having African Americans talk with members of other minority and majority groups in informal settings greatly improved their grade point average. “They found out that things that were happening to them were not things that were just happening to black kids. They were happening to every kid. They got the data, the evidence that their experience was not racially based, and then when their experience was not racially based in this environment, the whole environment changed. It wasn’t nearly as threatening. All those cues that might otherwise suggest threat were seen as much less threat.” Steele recounted from his own personal experiences that having an advisor during graduate school who believed in him was enough for him to overcome the many negative   G.M. Walton and G.L. Cohen. 2007. A question of belonging: Race, social fit, and achievement. Journal of Personality and Social Psychology 92(1): 82-96. 

18 UNDERSTANDING INTERVENTIONS cues he encountered. “With this one big cue that said I did belong in that setting, the significance of the other cues tended to wane away,” he recalled. An especially powerful way to undercut stereotype threat is to attack and undermine people’s theories of intelligence, Steele said, citing the work of Carol Dweck and Joshua Aronson. Many Americans tend to think that each individual has a particular level of intelligence and that one cannot perform beyond that level. But others, such as those from Asian and Eastern European cultures, see intelligence much differently. Students in those countries are more often taught that abilities are incremental and can be expanded through learning. They do not see math ability, for example, as something that is fixed and genetically determined but as something that people can improve. “I think this has huge effects on people’s choices of majors and persistence in graduate school,” said Steele. He felt that intervention on this topic would be especially valuable in entry-level, technical, and quantitatively based courses where stu- dents may receive their first sub-par grades, especially with faculty members who discourage students by telling them that many will drop out of or fail their courses. Survey Research Carefully conducted surveys can explore the attitudes, expe- riences, and thought processes that underlie the theoretical per- spectives described above. With his colleague Catherine M. Millett, Michael T. Nettles, senior vice president and Edmund W. Gordon Chair of the Policy Evaluation and Research Center at Educational Testing Service, conducted a 28-page survey of about 9,000 doctoral students at 21 U.S. universities. (The research team used a variety of incentives to achieve a 72 percent response rate, Nettles noted, including a raffle for cash payments.) The survey asked students about their background, undergraduate and doctoral program expe- riences, finances, aspirations, and expectations for graduate study. Conclusions drawn from the survey were published in the book Three Magic Letters: Getting to Ph.D. One critical factor Nettles and Millett examined was how stu- dents are supported during their doctoral education. In particular, they contrasted fellowships (money, tuition, or fee waivers given to students with no expectation of repayment or of services to be rendered), research assistantships (tuition, fee waivers, or a stipend   M.T. Nettles and C.M. Millett. 2006. Baltimore: Johns Hopkins University Press. 

EXAMPLES OF PREVIOUS RESEARCH 19 given to students with the expectation of research services to be ren- dered), and teaching assistantships (tuition, fee waivers, or a stipend given to students with the expectation of teaching services to be rendered). Nettles and Millett found that in the sciences, mathemat- ics, and engineering, African American students were less likely than white students to be research assistants during their doctoral programs, even when background characteristics and student expe- riences are taken into account. Yet being a research assistant can have a profound effect on a student’s experiences in graduate school. For students with a research assistantship, Nettles pointed out, “we observe an increase in students’ social interactions with peers, their academic interac- tions with faculty, their interactions with their faculty advisers, their presenting papers and publishing articles, and their overall research productivity.” Somewhat surprisingly, a research assistantship did not influence students’ time to degree, their overall satisfaction with their doctoral programs, or social interactions with faculty. Nettles noted that universities often use fellowships to attract students to their institutions. While fellowships can be attractive to prospective students, they can have other consequences once students arrive on campus. Because students on fellowship are not always engaged in research or teaching activities from the beginning of graduate work, Nettles said, fellowships “can lead to the social isolation or the neglect on the part of faculty of students who are not actually engaged in the production of [that teaching and research]. . . . This is not to suggest that fellowships are not a good idea, but I think that what universities are experiencing is trying to figure out the right balance.” Another critical factor identified in the surveys is whether stu- dents have a mentor. Nettles distinguished sharply between an advisor—who acts in an official capacity to give a student advice about academic programs or coursework—and a mentor—who is a faculty member to whom a student turns for advice about intellectual and academic processes as well as general support and encourage- ment. One of the good messages to emerge from the survey, Nettles said, was that race was not a major factor in whether a doctoral stu- dent had a mentor (possibly the same person as a faculty advisor). Furthermore, of the students who had mentors, three-quarters were able to find them within the first year of their doctoral experiences. Having a mentor influences social interactions between students and faculty, unlike having a research assistantship. Having a mentor also influences the rate of scholarly publishing, degree completion, 

20 UNDERSTANDING INTERVENTIONS and even time to degree. However, it did not influence satisfaction with doctoral programs or whether students left the program. A third key finding that emerged from their study was the importance of research productivity. Publishing in a refereed journal is a strong measure of this productivity, but the study showed that many other measures of research productivity are also important, such as presenting a paper at a research conference, publishing a book chapter, or being a member of a roundtable discussion at a professional meeting. As Nettles said, “many students pursuing research careers get on the train in different places.” Over half of the students surveyed had presented a paper at a conference, published an article in a refereed journal, published a chapter in an edited vol- ume, or published a book. Publishing in a journal “has become an extremely important endeavor for students,” Nettles said. “In fact, many people feel that they can’t complete [their degrees] without doing this because their first entry into the academic profession is going to be enhanced by their performance in conducting this activity.” However, the percentage of African American students publish- ing refereed journal articles in science and mathematics was signifi- cantly lower than for other groups (although that was not the case in engineering). Again, this was true even after controlling for factors such as student backgrounds and experiences. Before doing the study, Nettles thought that research productiv- ity might compete with time to degree because students would be devoting time and attention to producing articles and publishing. However, “we found just the opposite,” he said. Publishing articles actually was associated with an increased rate of progress in their doctoral programs and reduced the time to degree. Research on Existing Interventions Existing intervention programs can have research components that produce broadly applicable information. An example is the Alli- ance for Graduate Education in the Professoriate (AGEP), funded by the National Science Foundation. Yolanda S. George, deputy director for education and human resources at the American Association for the Advancement of Science (AAAS), which has provided evalua- tion capacity-building activities and research resources for the AGEP program, explained that the goal of AGEP is to increase the number of underrepresented minority students pursuing advanced studies, obtaining doctorate degrees, and entering the professoriate in STEM fields, including the social sciences. 

EXAMPLES OF PREVIOUS RESEARCH 21 AGEP has identified several factors that facilitate progression of minorities into STEM post-secondary studies: • Taking high-intensity and high-quality advanced high school STEM courses • STEM pre-college programs • Post-secondary support programs in core STEM courses • Financial aid packages that reduce debt burden • STEM pre-graduate-school bridging programs. The institutions that participate in AGEP “are expected to engage in comprehensive institutional cultural changes that will lead to sustained increases in the conferral of STEM doctoral degrees, significantly exceeding historical levels of performance,” George explained. She discussed several of the important lessons AGEP has demonstrated in seeking to achieve this goal. One lesson, accord- ing to George, is that admission and selection committees need to be conscious of diversity issues. The AGEP program tries to have a diversity coordinator or diversity-conscious faculty member sit in on admissions and selection. “You will get a behavior change if you get an advocate there,” said George. AGEP programs have also found that following up with applicants and linking financial aid to admissions helps with recruitment. At the same time, AGEP has found that it is important to work closely with university administrators on what can and cannot be done with recruitment and retention programs. George said, “You have to start talking to counsel about diversity-conscious and legally defensible student admission selection criteria, financial aid, and programs before you get that letter from that group that is threaten- ing to shut you down.” Furthermore, these discussions need to be ongoing, said George, since challenges will continue to arise. AGEP has conducted meetings and workshops to explore par- ticular topics. For example, a 2003 meeting on mentoring found that relatively little was known about mentoring specifically for STEM students. “We know that STEM core mentoring appears to be more prevalent in the after-school programs at the middle and high school level, but the level of systematic STEM career and workforce mentoring is not high in undergraduate research programs,” George said. However, support networks for women, including students, in STEM areas in academia, industry, and government are useful in   AAAS created a Science Mentoring Research website that followed on the 2003 meeting: <http://ehrweb.aaas.org/sciMentoring/>. 

22 UNDERSTANDING INTERVENTIONS helping to balance family and career, negotiating organizational or departmental challenges, and advancing in a career. George also observed that, through its program evaluation capacity-building project, AAAS has helped AGEP awardees build comprehensive evaluation and assessment infrastructures to exam- ine their graduate education enterprises. The framework for mak- ing change includes collecting and using disaggregated data for decision-making and leadership development within the faculty and administration. The goal of AAAS’s AGEP program is “to help the leaders in these projects, [who] are the deans and provosts in some cases, faculty members, and people who run the program, to figure out how to evaluate and assess the infrastructure in order to get the types of effects that they want,” George said. A particularly important task is to help faculty and administrators understand the research that has been conducted so that they can engage faculty members in the process of institutional change. Other Research Initiatives Several other important lines of research were mentioned more briefly by presenters and attendees at the workshop. Two described here are conducted by current grantees of the Efficacy of Interven- tions program; additional interventions and research studies are discussed elsewhere in this summary.10 For example, Reba Page, professor of education at the University of California, Riverside, conducts ethnographic studies of mentor- ing, journal clubs, research in labs, and so forth to understand how those components of intervention programs play out in practice. She wants to know “not what do people tell us they are, not what does the brochure tell us they are, but what do they actually look like in real time, as people, students and teachers together, enact the com- ponents.” By studying these situations and the processes they entail, Page is able to examine “the assumptions that undergird those pro- cesses and what holds them in place, and what we might want to target if we wanted to change them.” A prominent question in her work is why outcomes seem so resistant to change. The conclusion she has drawn is that outcomes depend not only on the culture of science but on the culture of the broader society. To understand sci- ence, including science education, “we have to see that science is embedded in our society,” Page said. Another line of research focuses specifically on the attitudes 10  See, in particular, Chapter 4 for discussion of initiatives by educational stage. 

EXAMPLES OF PREVIOUS RESEARCH 23 and preferences of students. Merna Villarejo, professor emerita at the University of California, Davis, has asked students in interviews about the motivating factors that caused them to make particular career decisions. Students who went to medical school tended to say that they want to give back to the community. But “that is not what researchers say,” Villarejo observed. “The most frequent answer for researchers for ‘why did you choose your profession’ is ‘because I really love science; it just turns me on; it is exciting; it is great.’” According to Rick McGee, associate dean for faculty affairs at Northwestern University’s Feinberg School of Medicine, another distinguishing characteristic was between students who wanted a fairly predictable future and those who were willing to live with more uncertainty. The students most likely to go into research were the ones who said, when asked about their future, “‘I don’t know, I might be doing this, I might be doing that, I might do this for awhile, I might do that for awhile.’ . . . They really are quite different people,” McGee said. As Daryl E. Chubin, planning committee member and director of the AAAS Center for Advancing Science and Engineering Capacity, said, many kinds of investigations can produce information needed to advance minorities in research careers. “Where does knowledge come from? We know it comes from data and we know it comes from research. But it also comes from evaluation and it comes from technical assistance and it comes from first-person reports. . . . The challenge here is to learn from all of these interventions and then try to apply that in our own context.”