A common, sexy, commonsensical claim is that having women/minority role models will motivate other women and minorities into being high-achieving.
I was pleasantly surprised to find that this claim has been repeatedly tested (and not at all surprised that this research is never cited when people talk about the importance of said role models).
"For many years academic researchers, governmental officials, and the lay public have assumed that the number and type of role models available to a student will have an important influence on that student’s performance in school and selection of a career. These assumptions became highly visible in the 1978 Supreme Court case of Regents of the University of California y. Allan Bakke (438 U.S. 265, 1978)...
[Speizer] reviewed studies on mentors and sponsors. Speizer found that most of the relatively few studies that claimed to show an influence of role models have severe methodological problems, which bring their conclusions into doubt.
In her conclusion, Speizer wrote: “Role models, mentors, and sponsors are concepts which still need to be defined and studied. Despite their almost universal acceptance, there is very little supportive evidence for their validity. Until methodologically sound studies are conducted on large, randomly selected populations, these concepts should be considered as suggestive rather than proven” (Speizer 1981, p. 712).
Our more recent search of the large literature on role models found virtually no studies of the type that Speizer called for over twenty years ago. Most papers on role models simply assume the importance of same-gender and same-race/ethnicity role models on a large variety of dependent variables. An example is Janice C. Bizzari’s “Women: Role Models, Mentors, and Careers” (1995), in which she states, “According to recent studies, evidence suggests, potential can be denied or lost for women in certain male-dominated careers for lack of women mentors or role models in the field” (p. 145). The studies she cites are either qualitative or nonsystematic.
There are a few exceptions, in which the concept of role model is looked at systematically using relatively large data sets. All of these studies have been done by economists, and most appear in an issue of Industrial and Labor Relations Review edited by Ehrenberg (1995). One study that did not appear in the Ehrenberg volume is that reported by Mark O. Evans (1992). Evans conducted a study of the influence of same-gender and same-race role models on how much students learned in high school economics classes. Evans used data from the Joint Council on Economic Education’s National Assessment of Economic Education Survey con ducted in the spring of 1987. From that survey he had information on 2,440 students who took a high school economics course, including the results of a test of economic knowledge, which Evans used as his dependent variable. From his careful study, Evans concluded that there is no evidence of same-gender role model effects but that black students do slightly better on the test when they have black teachers, particularly when their mothers do not have a college degree. The effect, although statistically significant, is relatively small. It should he noted that Evans (like the studies reported later from the Ehrenherg volume) considered the race and gender of the teacher, but did not determine whether the students looked upon the teacher as a role model.
Ronald Ehrenberg, Daniel Goldhaber, and Dominic Brewer (1995) reanalyzed the data collected in the 1960s by James Coleman (see Chap ter 1) to see if matching the teacher’s and student’s race had any effect on the amount that students learned. Their main conclusion was that African American teachers did not help African American students and under some conditions had a negative effect on how much white students learned. Donna Rothstein (1995), a student of Ehrenberg’s, used the “High School and Beyond” data set (a longitudinal study conducted by the National Center for Educational Statistics) to investigate the effect of attending a women’s college on labor market and educational outcomes. She found that when appropriate input variables were controlled, single- gender schools conferred no advantage on young women. But she also found that among those who attended women’s colleges the percentage of female faculty had a small positive effect on postcollege wages...
None of the studies in the Ehrenberg collection, although all based on large systematic data sets, determined whether the students considered their teachers to be role models. Despite this fact, the papers in the Ehrenberg collection do offer some interesting and suggestive findings.
Harvey Rosen and Brandice Canes (1995) studied how the number of female faculty in science and engineering fields might influence the number of women entering the fields. They correctly point out that a cross-sectional analysis would not answer this question because the same factors that make a particular field attractive to women professors might make it attractive to women students. Instead, Rosen and Canes chose to look at change over time in the number of women professors. They found no evidence that this variable had any effect on the number of women who majored in the sciences.
Sara Solnick (1995) studied female students at women’s colleges and at coeducational institutions and found that the women at the all-women schools are more likely to shift from “traditionally female” majors to either neutral or traditionally male majors. She found no evidence to support her second hypothesis: that women who begin in traditionally male majors would be more likely to persist in those majors if they attend a women’s college. Solnick’s research does not specify the mechanisms through which the women’s colleges have their influence. indeed her results could easily be due to the self-selection of particular types of students into women’s colleges.
Donna Rothstein (1995) examined how the percentage of female faculty at a college or university might influence the likelihood of women’s attaining an advanced degree and later earnings. She found that the pro portion of women had no direct effect on earnings hut that it had a small indirect effect on earnings through the proportion of women who go on to attain an advanced degree. Again, in order to understand fully these findings we would have to know more about how the students attending colleges having a high proportion of women as faculty might differ from those who attend colleges with a lower proportion of women as faculty.
In a study conducted by Rothsrn and Ehrenberg (1994) the authors found that “attendance at a HBCU substantially enhanced the probability that a black college student would receive a bachelor’s degree within seven years after starting college; however, on average, it had no apparent effects on the student’s early career labor market success (as measured by 1979 earnings) or the student’s probability of enrolling in graduate school” (Ehrenberg 1995, p. 484). jill Constantine (1995) asked the same question but uses earnings data from a later period in the career history of the students she studied. She found that attending an HBCU does have a significant influence on later earnings. The reasons for the difference in findings between the two studies are unclear. However, Ehrenherg (1995) points out that the 1986 wave of the NLS-72 data, which Constantine utilized in her study, was a subsample of the original sample and oversamples college graduates who earn more money than nongraduates. This may account for the difference between her findings and those of Roth stein and Ehrenberg.
Finally, Ehrenberg, Goldhaber, and Brewer (1995) used data from the National Educational Longitudinal Study of 1988 to determine whether the race or gender of the teacher has an influence on how much students learned. They found that the gender and race of the teacher have no significant influence on how much the student learns but that they do influence how the teacher evaluates the student. Thus white female teachers were more likely than were white male teachers to evaluate their white female students highly. They noted that their data provide no evidence that black students will learn more when they have black, rather than white, teachers. They add that there are two interpretations that can be given to their data:
At face value, our findings may be interpreted in either of two conflicting ways. On the one hand, if it is argued that what is crucial is how much students learn in classrooms, one might conclude that teachers’ race, gender, and ethnicity per se do not matter. On the other hand, if it is argued that teachers’ subjective evaluations of students mirror the encouragement they provide these students and the “track” on which they place the students or to which they encourage them to aspire, our results suggest that in some cases teachers’ (race, gender, and ethnicity) do matter. (Ehrenberg, Goldhaber, and Brewer 1995, p. 560)
Our general conclusion remains the same as when J. R. Cole published his book on women in science: SO far, there is no systematic evidence that same-gender or same-race/ethnicity role models have significant influence on a range of dependent variables that they are assumed to influence, including occupational choice, learning, and career success."
--- Increasing Faculty Diversity: The Occupational Choices of High-Achieving Minority Students / Stephen COLE, Elinor G. Barber
This undermines what is the key rationale for having gender/racial quotas, and a key rationale for affirmative action.