Attrition from Science
Minority interest and persistence in the hard sciences is highly relevant to a fully integrated society, but racial preferences used at the undergraduate level plausibly cause mismatch effects. Because of the rigid structure and competitive nature of engineering and physical science programs, students that intend to major in science could find themselves at a disadvantage if they enrolled at a school where their incoming credentials are weak relative to the rest of their class, and might drift to other majors or out of school altogether. If so, disproportionate numbers of minorities will get bumped off the science track.
Project SEAPHE has conducted preliminary research on the effects of mismatch on persistence using data from the University of California, the University of Michigan, and a number of other sources. Prof. Richard Sander is presenting these findings at a
September 12th hearing of the U.S. Civil Rights Commission.
Prior research on the subject of minority representation in the sciences has found that there is no disparity between underrepresented minorities and whites in terms of initial intention to major in the hard sciences at the point they apply for college.
Interest in Science Among High School Seniors
| Comparison group |
Proportion of students of each race intending to major in science |
| White |
Black |
Hispanic |
Asian |
| University of California - average |
34.7% |
37.5% |
52.6% |
| University of California - 2004-06 |
37.3% |
40.5% |
57.1% |
| HERI-CIRP data (Astin et al.) |
27.3% |
34.2% |
35.7% |
52.6% |
| Rogers Elliott data (4 elite institutions) |
41.4% |
44.2% |
44.0% |
55.0% |
However, minority representation in the hard sciences dwindles at every stage after college enrollment. Black and Latino attrition from science during college is much greater than that of whites and Asians, and the gap only widens when looking at continuation to graduate school, and research and teaching.
The literature below provide some background on the subject.
Alexander W. Astin and Helen S. Astin
(1992)
Abstract
Astin and Astin use longitudinal data on over 27,000 college students at 388 colleges to conduct a comprehensive study on an array of factors that affect interest and persistence in science. The authors find that entering academic credentials (particularly in math) are consistently the strongest predictor of students’ persistence in science. Credentials are also good predictors of initial interest in science. Pre-admissions credentials are also the strongest predictors of performance on the GRE and MCAT tests, suggesting that post-graduate performance is largely constrained by pre-collegiate credentials.
more...
The authors were surprised to discover that African American and Hispanic students were more likely to intend to major in hard science than white students. But these racial groups have much higher rates of attrition during college than white so that the proportion actually receiving science degrees is a little higher for whites than the under-represented minorities. The authors are not able to explain why underrepresented minority groups suffer such large losses of attrition. (Pg. 50-51.) But students leaving the sciences tend to achieve lower grades. However, the post-graduation outcomes (GRE and MCAT scores) reflect large racial disparities. African-American students perform less well on these tests than would have been expected from their pre-college credentials. So the test score gaps that already exist between African-Americans and their white peers at the beginning of college are widened at the end of college.
The authors find that freshmen with a greater proportion of science major friends are significantly more likely to persist in their science major. This peer effect is the largest of the environmental effects that the authors studied. The manner in which science is taught also has large effects on persistence in science during and after college, as the schools that encouraged self-directed research and teaching assistance produced students more likely to stick with science. Written student evaluations also seem to keep students on track.
SEAPHE Comments
This study predates the serious and organized study of mismatch, so the authors did not take into consideration how a student’s SAT-Math scores and other important predictors relative to their peers affects persistence in science. Circumstantial evidence corroborates mismatch—that low grades result in attrition from the majors, and that African-American students’ GRE and MCAT scores were worse than whites who had entered college with similar credentials. This remarkably complete database, currently held in the HERI archives, would be ideal for testing mismatch effects on standardized post-college tests like the GRE and MCAT. Thus far, we have not been able to make arrangements to use the dataset.
One note: In the executive summary at the beginning, the authors state that “white students show larger proportionate losses from scientist-practitioner careers during college than do other racial/ethnic groups.” This is misleading; whites are less likely to be scientist-practitioners because they are much more likely to be scientist/professors or engineers.
Rogers Elliott et al.
(1996)
Abstract
The authors used individual-level data for a single year (1988) from four highly selective universities to determine what accounts for the racial disparity in the award of science degrees. The authors ruled out differences in initial interest as an explanation because black and Hispanic enrollees to the colleges sampled had very high initial interest in science (higher than whites), despite having lower standardized test scores and high school GPAs than the other racial groups. The disparity in ultimate degree awards was the result of attrition from the science majors, which affected black students much more than any other racial category. Only 33.8% of the black students in the sample that had an initial interest in science ended up completing a degree in science, as compared to 55% for Hispanics, 61% for whites and 70% for Asians. Drop out rates from school were similarly lopsided.
more...
The authors hypothesize that, because science courses are difficult, fast-paced, and impersonal, an individual student’s standing in terms of credentials within their institution is a very important predictor of persistence in the field. Since minority students admitted through preferential admissions programs will enter with lower pre-college credentials, they are put at a disadvantage compared to their white and Asian peers. Circumstantial evidence of this theory is displayed in Table 4, which reports aggregated data on the percent of science degrees awarded broken down by SAT-Math terciles for eleven private institutions. The table suggests that a student with an SAT-Math score below 600 would be much more likely to attain a science degree at the least elite institutions (Institutions J and K) than at the most elite (Institutions A-C). For example, a student with an SAT-Math score of 600 would be at the top of the distribution at Institution J and have a 55% chance of completing a science degree, but would be in the bottom third of their class at Institution A and have only a 15.4% chance of completing a science degree. The article goes on to assess academic support programs that seem to have proven benefits for the minorities that participate in them.
SEAPHE Comments
Elliott et al make a compelling case for the need to study how mismatch affects persistence in science at the undergraduate level, but the authors were not able to make within-race comparisons to test whether the observed racial disparity in persistence was caused by the relative strength of their incoming credentials instead of some other factor that correlates with race (stereotype threat, lack of critical mass.) The sample available to the authors was surely too small to contain enough matched minority students to do a convincing study of mismatch that could control for race and mismatch separately. Also, the four colleges for which the authors have individual-level data are highly selective and therefore atypical; it will be interesting to see if initial interest and persistence in science for under-represented minorities have similar patterns at non-elite institutions.
