Chancellor Berdahl's letter on UC Berkeley's admissions policy
A recent Los Angeles Times story about a draft analysis by UC Regent John Moores of UC Berkeley's 2002 admissions process reported that marginally academically qualified students were admitted at the expense of more highly qualified applicants. In an Oct. 6 letter printed below, Chancellor Robert M. Berdahl tells UC President Robert Dynes and the Board of Regents that Moores' draft analysis contains some erroneous assumptions and observations and that the Los Angeles Times reporters failed to include "essential contextual information" that faculty and staff provided them during many hours of interviewing.
UC President Robert Dynes also wrote to the Regents about the article, and the UC Office of the President created a Q&A to address questions about UC admissions raised by the newspaper story and the Regent's analysis.
October 6, 2003
To: Robert C. Dynes
President, University of California
I appreciate this opportunity to share with you, and the Board, additional information about UC Berkeley’s admission policy and process. We hope you will find this information helpful in thinking about the issues raised by Regent Moores’ recent draft analysis. In this letter I address the fundamental questions we believe Regent Moores’ document raises: what are the goals of our admission policy and what constitute legitimate and educationally sound ways of reaching those goals?
As you know, the Los Angeles Times published a story on Saturday on Regent Moores’ report. Although faculty and staff from UC Berkeley and the Office of the President spent many hours talking with the authors of that story, the reporters failed to include essential contextual information necessary to understand some of the data cited. This letter also addresses some of the fundamental misunderstandings in the Times article.
I. The Policy Context for Berkeley Admissions
Deciding which of thousands of well-qualified applicants to admit is certainly one of the most important and difficult tasks that we as faculty and administrators face. Fortunately, campuses are given strong and effective guidance on this topic—from the state in the Master Plan, from The Regents in Regental policy, and from the faculty in specific admission criteria and policy guidelines articulated at both the systemwide and campus levels.
The state, through the Master Plan, dictates that admission should be based most fundamentally on academic criteria: only the top 12.5% of California graduates are considered qualified for UC admissions. UC determines eligibility based entirely on traditional quantitative criteria: grades in required academic courses, SAT II scores, and ACT or SAT I scores.
Regents’ policy states that “Mindful of its mission as a public institution, the University of California…seeks to enroll, on each of its campuses, a student body that… demonstrates high academic achievement or exceptional personal talent, and that encompasses the broad diversity of…backgrounds characteristic of California.” Regents policy further specifies that, when selecting among eligible applicants, campuses should take into account “multiple measures of achievement and promise, while considering the context in which each student has demonstrated academic accomplishment.”
Taken together, these policies indicate that (1) academic qualifications should be given priority; (2) academic achievement and potential should be evaluated based on multiple measures; and (3) campus selection processes should be designed to provide some level of access to low-income and educationally disadvantaged students. These principles are further elaborated in the University’s undergraduate selection criteria and BOARS’ “Guiding Principles for Comprehensive Review.”
Berkeley’s admission policy is fully consistent with both Regental and faculty policy. Academic factors dominate campus admission decisions. Berkeley has room for roughly one-fourth of its applicants in the fall term and approximately 30 percent when fall and spring are combined, as they are in Regent Moores’ analysis. For California applicants to the College of Letters and Science whose GPA’s and test scores are in the top one-quarter of the Letters and Science applicant pool, the admission rate is 89 percent. For students with only one of those factors above the 75th percentile, the admission rate drops to 55 percent. Conversely, students whose GPA’s and test scores fall below the 50th percentiles have an admission rate of only 7 percent.
The academic quality of students admitted to Berkeley is higher than it has ever been. For example, data provided to The Regents in the September 2003 BOARS report on comprehensive review indicate that, for each of six quantitative academic indicators shown, the profile of Berkeley admits has risen in both of the last two years. This is true throughout the breadth of the admitted pool. That is, academic indicators at Berkeley are not rising simply because students at the “top” of the admitted pool have higher grades and test scores than they did five years ago, but because at each point in the full distribution of the admitted pool, these indicators continue to increase.
At the same time, it must be understood that academic criteria—and particularly the single criterion of SAT I total score—are not absolute. Berkeley considers a broad range of factors for each applicant and the total picture the student presents determines his or her admission score. On average, applicants’ GPAs and test scores correlate very highly with the admission score they receive (as is evident in Regent Moores’ analysis). But for any individual applicant some factors will be higher and some lower than the average for that score. More important, in accordance with Regental policy, Berkeley does endeavor to preserve some level of access for low-income students. Because SAT I scores, in particular, are very highly correlated with family income and education level, it is likely that some students with otherwise strong academic and personal qualifications will present relatively low SAT I scores. As a result, while applicants with higher test scores are much more likely to be admitted than those with lower scores, a small number with lower scores will be admitted. This outcome is not unusual—it is true on all UC campuses and at all selective institutions across the country.
II. Comments on Specific Aspects of the Report
Regent Moores acknowledges that his analysis is highly preliminary. As such, it contains some erroneous assumptions and observations we believe will benefit from additional explanation and analysis. We welcome the opportunity to work together to examine each assumption and finding in greater detail. To that end, below we identify and briefly discuss several areas where we believe additional work is particularly warranted.
A. The Consistency of the Reader Score in Admission Decisions
One of the most troubling findings in Regent Moores’ analysis is an apparent lack of congruence between the “read” scores students receive and their admission or denial. This finding is based on an incomplete analysis of the data.
All Berkeley applicants are evaluated and scored in the same way, using a single, consistent set of criteria and implementing guidelines. However, like all UC campuses, Berkeley operates multiple, simultaneous admission processes. Therefore, all applicants are not competing directly against one another for the same spaces. Rather, they are separated into different pools according to the college (and, in come cases, the specific major) to which they have applied, their residency status, their eligibility for special programs (e.g., athletics, augmented review), etc. Read score cut-offs for admission vary depending on the program for which the student is being considered. Thus, a student applying to one major within the College of Engineering may be admitted while an equally qualified applicant with the same score may be denied in another major within the same college. When differing admission targets for different programs are taken into account, reader scores and admission decisions are fully consistent.
B. The Role of the SAT I in Admission Decisions
In any process involving multiple criteria, selected applicants will not appear “uniform” when compared based on a single criterion. In this specific case, focusing on the SAT I total score as the dominant measure of academic “quality” is especially problematic. Research conducted on UC applicants and enrolled students demonstrates that, of the three major quantitative indicators of academic preparation (high school GPA, SAT II scores, and ACT/SAT I scores), SAT I scores are the least predictive of first-year success at UC. The relatively greater reliability of GPA and SAT II scores are officially reflected in UC’s Eligibility Index, which weights GPA most heavily—consistent with its relative power in predicting UC success—and ACT/SAT I least heavily.
Consistent with this research, faculty policy at Berkeley explicitly instructs readers to give greatest weight to the complete high school record—including not only grades earned, but courses taken and the pattern of achievement over time. It also instructs them to give greater consideration to SAT II scores than to ACT/SAT I’s. Consequently, for students assigned the same overall read score, we would logically expect SAT I scores to vary more than any other numerical factor.
This being said, Berkeley has nonetheless carefully investigated the cited cases of students with very high scores who were denied and those with more modest scores who were admitted. In the case of denied students with high scores, we found that in virtually every case one or more of four factors was at work. Either (1) the students had withdrawn their applications and were thus coded as “non-admits” when they had not in fact been denied admission; (2) they were out-of-state applicants, for whom, in accordance with faculty policy, the campus establishes higher standards than for in-state students; (3) their GPA’s and other academic factors were below average for Berkeley admits; or (4) they had applied to one of three very highly competitive majors in the College of Engineering. In fact, for California residents with SAT I scores above 1400 who did not apply to one of those three majors and whose GPA’s were not below average for the Berkeley admit pool, the admit rate was 98 percent.
Similarly, for students with SAT I scores below 1000, all indicated impressive overall academic and personal achievements. Nearly half ranked in the top 4 percent of their graduating class and many provided ACT scores which, when translated to an SAT I equivalency scale, were well above 1000. Additionally, most had achieved this level of excellence despite multiple, significant socioeconomic and educational challenges. Most important, first-year performance data for these students indicates they are doing well at Berkeley: not one has left due to academic deficiency.
C. The “Subjectivity” of the Berkeley Process
The admission processes at all UC campuses employ some element of “subjectivity.” Readers examine evidence in the application relevant to a specific criterion—e.g., leadership—and evaluate that evidence relative to their sense of the overall qualifications of the total applicant pool. Their judgment may be reflected in a number of points awarded for that factor, in a general score the applicant is given for “personal achievements,” or, as it is at Berkeley, in a unitary score that ranks the applicant relative to all others in the pool on all criteria.
Berkeley’s process is unique among the campuses in that each reader assigns a file a single score, rather than two, three, or more scores that are then combined to reach a decision. It is inaccurate to suggest that this assessment is not grounded in extensive data, as well as in written policy, established practice, and highly informed professional judgment. For example, when evaluating each applicant, Berkeley readers review not only the entire application, but also a data summary that includes 58 separate numerical data points about that student and his or her educational circumstances. In evaluating each applicant’s academic record, readers not only examine quantitative indicators, they also study closely the actual transcript, which shows course-taking patterns, course-by-course grades (as opposed to averages), and trends in academic performance. Our faculty adopted this approach because they believe it is a more informed, rigorous, and accurate evaluation than the formulaic one Berkeley previously employed.
My colleagues on the faculty and I welcome the opportunity to provide additional information to The Regents and we look forward to a comprehensive analysis and discussion of these issues.
Robert M. Berdahl