Saturday, June 30, 2018

Affirmative action at America's top tier private colleges and universities ... updated 9/29/18

Last update: Satirday 9/29/18
The advocates for Asian applicants who are currently challenging Harvard University's admissions policies would surely concede that all of the freshmen admitted in a particular year to a top tier university should be Ethiopian or Polish or Peloponnesian Greek if Ethiopians, Poles, or Peloponnesian Greeks were the most qualified applicants that year.

Executive Summary -- TLDR ... See "Update" comments at end of this post
I was motivated to write this note by last week's news reports that the Trump administration was reversing the support that the Obama administration had given to affirmative action. These reports were confirmed by an extensive review in this morning's NY Times (7/4/18) of the official announcements from the Departments of Justice and Education.

I was concerned that most readers outside of academia don't understand how affirmative action really works at the nation's top tier universities so they suspect that the Black and Hispanic students who were admitted to these institutions via affirmative action were not as academically qualified as the White or Asian applicants who were rejected.
  • My discussion begins with a description of a rubric that an admissions office at an imaginary Great University might follow, a rubric that is easy to understand but complex enough to capture the most important features of the admissions process at a real top tier university. The rubric would enable Great U's admissions office to reduce the mountain of applications it received down to a manageable pool of the academically qualified applicants who were most likely to become highly successful in their careers in the years after they graduated.
  • The rubric would also guide the admissions office in its subsequent application of affirmative action policies to the pool of academically qualified applicants i.e., policies that increased the admissions prospects for three groups of applicants: legacies, applicants from financially stressed families, and applicants who were members of underserved minority groups
  • The preferential admission of legacies, the sons and daughters of alums of Great U, is its most important form of affirmative action, most important in the sense that Great U admits far more legacies than financially stressed students or minority students 
  • Enrollment data from the nation's five wealthiest universities, where wealth was measured by the size of their endowments, was examined. Enrollment data from a sixth leading institution whose endowment was much smaller than the top five but whose admissions standards were much higher was also examined. A smaller endowment and higher admissions standards don't leave room for affirmative action preferences to have substantial impact on its enrollments. So this institution's enrollments were used as benchmarks to gauge the impact of affirmative action policies at the five richest institutions
  • There were strong indications that the admission of legacies at the five richest institutions was the most likely cause of their admitting a smaller percentage of Asian students than might otherwise have been expected.
The Asian challenge to Harvard's admission policies will ultimately become a challenge to Harvard's right to give preferential admission to the academically qualified young relatives of its alums, alums whose donations and other support enabled Harvard to pay the high salaries that secured the employment of some of the world's finest scholars and provided them with cutting edge infrastructure to support their teaching and research. Without this generous support from its alums Harvard would not be the preeminent institution to which Asian applicants are clamoring for increased admission.

+++++++++++++++++++++++++++
+++++++++++++++++++++++++++

A. Simple Model
But what should "most qualified" really mean? My sense is that many, if not most readers outside of academia have vague notions that the "most qualified" applicants should be the ones who earned the highest grade point averages (GPAs) in high school and/or received the highest scores on their SAT or ACT exams. These notions go hand in hand with suspicions that institutions that employ affirmative action policies rob qualified applicants of well-deserved admissions and give these coveted slots to unqualified applicants from minority groups. I preface my discussion with four caveats:
  • First, never having worked in the admissions office of a top tier private university, I have drawn on my administrative experience in less rarified academic environments to construct a simple model of the admissions process at an imaginary Great University. Hopefully my model will enable readers from outside of academia to understand why the identification of the "most qualified" applicants is more complicated than merely selecting the ones who have the highest GPAs and SAT/ACT scores. 
  • Second, I confine my discussion to private universities because public universities are not as free to adopt the kinds of affirmative action policies that are implemented by private universities, e.g., Harvard. 
  • Third, I'm a Black American alum of Harvard's Graduate School of Arts and Sciences. Readers should therefore stop reading right now ... or ... (my preference) finish this note before deciding whether my race and/or my degree have biased my judgments ... :-)
  • Finally, I offer no defense for Harvard's ineptitude in rejecting many Asian applicants primarily because they failed to satisfy a mysterious "personality" requirement. 

The "most qualified" applicants to Great University
From Great U's perspective the "most qualified" applicants are the ones who will provide the most help for Great U's efforts to become an even greater university 20 to 30 years from now. Most students can't help Great U until decades after they graduate because it takes time to reach the top of their professions -- 20 years, 30 years. But at that point the most successful alums will be positioned to provide Great U with substantial contributions to its endowment, enhanced visibility, and political connections.

Quantitative and qualitative metrics
Good grades in high school plus high scores on SATs are good predictors of a student's success in his or her first two years in college, but less reliable predictors of success thereafter. SAT tests were only designed to make reliable short-term predictions of a student's success while in college. That's all.
    Predicting high levels of success 20 to 30 years after graduation is precisely the challenge that faces Great U's admissions office. It must try to identify the students who, with the benefit of the excellent education plus the powerful network connections they will receive from Great U, are most likely to become successful enough to return substantial benefits to Great U. 

    Admissions rubrics and quantitative indicators 
    The model was framed as a rubric because rubrics enable us to encapsulate  different kinds of data about applicants into comparable profiles that can be ranked against each other. Real colleges and universities might use other processes to achieve equivalent results. The rubric evaluates simplified student profiles, i.e., profiles that only contain a few pieces of information about each applicant. Profiles for real colleges and universities would contain a lot more information, e.g., ACT scores and other equivalent measures of academic skills.
    • Suppose that our rubric assigned a maximum of 160 quality points to each profile. Further suppose that the profiles that would be considered for admission would have to receive at least 115 points. 
    • Further suppose that Great U could only admit 2500 students per year. Then Great U's incoming class would be the 2500 applicants whose profiles were assigned the highest number of quality points beyond the minimum 115.
    • Applicants who were unlikely to graduate should not be admitted. There are no remedial programs at private top tier universities like Great U.  All applicants are required to present high enough GPAs and SAT scores to "guarantee" that they are academically qualified, i.e., that they already possess the prerequisite knowledge and study skills required to graduate. 
    • Let's assume that our Great U requires that all applicants have at least a 3.6 GPA (where max GPA = 4.0) and score at least 700 on their math and reading/writing SAT exams, respectively (where max SAT scores = 800). Applicants who achieved these minimums would receive 70 points from our rubric. Applicants who do not achieve these minimum grades and scores will be rejected.
    • A real college or university could estimate these minimum cutoff values by conducting a statistical analysis of its student records for the last ten years or so. The analysis would determine the GPAs and SAT scores that best predicted 95 percent of the students who actually graduated. Less selective institutions might might opt for lower grades and scores that predicted lower percentages. Reminder: even perfect 4.0 GPAs and 800 SAT scores will not provide 100 percent guarantees that every student with these perfect marks will graduate.
    • Unlike Great University, some highly selective universities that were striving for 95 completion rates might decide to admit students whose grades and SAT scores predicted lower completion rates. These institutions would implement remedial programs that provided lower rated applicants with the prerequisite knowledge and study skills that were already possessed by its other students. Nevertheless, these institutions would still be vulnerable to charges that they admitted students who were less qualified than some of the students they rejected. 
    • The rubric awards additional quality points to applicants who earned higher than the minimum GPA and SAT scores. For example, applicants who scored 750 on their SATs might receive an additional 5 points; those who scored a perfect 800 might receive an additional 10 points. But no more than 30 additional points would be awarded for higher GPAs and and the two SATs. To be specific, applicants who presented a 4.0 GPA, an 800 math SAT, and an 800 read/write SAT would receive the maximum 30 additional points.
    • As consequence, the rubric assigns at most 70 + 30 = 100 points for GPAs and SAT scores to an applicant's profile, well below the 115 point minimum required to become a contender for admission. No applicant would be considered for admission solely on the basis of GPAs and SAT scores. Why? Because these quantitative metrics by themselves are unreliable predictors of long-term success at the highest levels. 

    Admissions rubrics and qualitative metrics 
    The admissions office of Great U would then examine the qualitative material in the admissions package of the applicants who met the minimum GPA and SAT requirements ==> letters of recommendation, notes from interviews with Great U's recruiters, descriptions of hobbies, interesting summer internships, participation in competitive sports, research projects, application essays, etc, etc, etc. 

    The office would assign quality points based on indicators of creativity, grit, self-esteem, capacity to work productively with colleagues, and any other characteristics that would increase an applicant's likelihood of becoming highly successful in whatever careers they embarked on 20 to 30 years after they graduated. In our rubric the office would assign anywhere from zero to a maximum of another 30 additional points to each applicant's profile. At that point each profile would be valued somewhere between 70 quality points and 70 + 30 + 30 = 130 quality points. Students who earn at least 115 quality points are now eligible for admission.

    Three forms of non-remedial affirmative action for eligible applicants

    • Note: In its 2003 Gratz v. Bollinger decision, the U.S. Supreme Court rejected the rubric used by the University of Michigan because the university awarded points to minority students whose qualifications as individuals had not been evaluated. As consequence, some students might have received the minimal points required for admission only because they received affirmative action points. By contrast, the present rubric only awards affirmative action points to students whose profiles have already earned the minimum 115 points required for admission. 
    Great U's rubric assigns a final round of quality points to eligible applicants who fall into one or more of the following categories:
    • Legacies are relatives of Great U's alums, e.g., typically daughters and sons. As a private institution, Great U depends on substantial donations from its alums to cover shortfalls between tuition received and operating expenses. More importantly, alums contribute to endowed chairs, financial aid programs, and capital expenditures for new dorms, labs, athletic facilities, computers, campus networks, libraries, and classroom buildings. Alums are more likely to make generous donations if their young relatives receive special consideration. This rubric assigns 10 quality points  to academically qualified legacies. These additional points will not guarantee the admission of legacy applicants; the points will just make their admission more likely. 
       
    • Financially pressed applicants are of interest to Great U because its quest for continued excellence impels it to recruit the most academically qualified applicants, regardless of their family's income. However, applicants from lower income families would not have been able to afford as many out-of-classroom learning opportunities as their more affluent peers. So they won't have as many research projects, exciting but unpaid summer internships, extensive hobbies, etc, etc, etc in their application packages. Accordingly, the rubric awards academically qualified, but financially pressed applicants with 10 quality points to make their profiles more competitive.

      Admitting academically qualified applicants from financially stressed families is senseless if the applicants can't afford to pay Great U's tuition, fees, and other expense. So Great U also offers them generous financial aid packages that covers these expenses.

       
    • Underserved racial/ethnic minorities includes non-White applicants who are Hispanic, Native American, Native Hawaiian, or Black. Asian applicants are usually not included in this category. Whereas financially pressed applicants could not afford as many out-of-classroom educational opportunities as more affluent applicants, members of underserved minorities are often denied access to these opportunities just because they are members of underserved minorities. Worse still, they are often deprived of access to schools that provide as many in-class learning opportunities for the same reasons. Therefore the rubric awards 10 quality points to academically qualified underserved minorities in order to make their profiles more competitive. 
    The goal was to provide a rubric that was easy to understand, but complex enough to illustrate the limited role of GPAs, SAT scores, and affirmative action in admissions decisions at Great U. That's why it only included three forms of affirmative action, omitting some other well-known categories.  For example:
    • Daughters and sons of celebrities -- Children of presidents, governors, senators, Nobel prize winners, superstar athletes, Oscar winning actors/actresses, etc, etc, etc 
    • Prodigies -- Applicants who are celebrities in their own right, e.g., young movie stars, best selling authors, rock stars, classical music virtuosos, and Olympic champions;
    • Athletes -- Applicants who play one or more sports well enough to become members of Great U's teams
    Academically qualified applicants who were prodigies or the children of celebrities would provide Great U with substantial boosts to its visibility right now, rather than 20 or 30 years from now. The present value of these immediate boosts might be worth 20 quality points. On the other hand, athletes who were not world class prodigies but were good enough to play on Great U's (winning) teams also provide immediate boosts to Great U's visibility, but not nearly as much as prodigies. So athletes might only receive 10 quality points. 

    Diversified enrollments
    There is an obvious tradeoff between Great U's use of affirmative action to enhance the admissions of its legacies and its use of affirmative action to diversify its enrollments. More of one means less of the other. Legacies guarantee continued high levels of support from its most successful alums. But in a world that grows increasingly diverse, the most successful leaders in any field 20 or 30 years from now will be the alums who learned early and well how to work productively with diverse groups of colleagues and employees. So it is in Great U's long term interest to cultivate diverse enrollments.

    Great U's diverse enrollments could be likened to a diversified investment portfolio. In this analogy, academically qualified legacies are the low risk investments that yield guaranteed returns. Their family connections plus Great U's excellent education plus Great U's connections provide strong assurances of their future success. By contrast, academically qualified students from financially stressed families and/or underserved minorities lack the family connections that will facilitate their future success, so their recruitment represents a higher risk for Great U. Their recruitment is also more costly. Financially stressed applicants will also need financial aid; and many, if not most, minority applicants will also be financially stressed.


    B. Fitting the model to reality
    Table 1 (below) contains data about the five private institutions that had the largest endowments in 2016 ... plus data about the California Institute of Technology because it will provide useful points of reference.

    Table 1. Five Private U.S. Universities Having the Largest Endowments ... Plus Cal Tech

    Undergraduates
    Harvard
    Yale
    Princeton
    Stanford
    M.I.T.
    Cal Tech
    1
    Enrollment
    9,915
    5,472
    5,400
    7,034
    4,524
    979
    2
    % White
    43
    46
    43
    36
    35
    29
    3
    % Hispanic
    11
    12
    9
    16
    15
    12
    4
    % Black
    7
    7
    8
    6
    6
    1
    5
    % Asian
    17
    17
    21
    21
    26
    43
    6
    % White-Asian Gap
    -26
    -29
    -22
    -15
    -9
    14
    7
    % White 6yr grad rate
    97
    97
    98
    96
    96
    93
    8
    % Hispanic 6yr grad rate
    96
    99
    95
    94
    90
    77
    9
    % Black 6yr grad rate
    97
    96
    94
    92
    85
    100
    10
    % Asian 6yr grad rate
    96
    98
    99
    95
    94
    93
    11
    SAT Math 25%
    730
    730
    720
    700
    770
    780
    12
    SAT RW 25%
    730
    730
    710
    690
    720
    750
    13
    % Pell grants
    11
    14
    16
    15
    17
    12
    14
    Endow ($millions)
    37,616
    25,543
    22,291
    22,223
    13,475
    2,079
    15
    Endow Rank
    1
    2
    3
    4
    5
    42
    Note: All data in Table 1, except endowments, is for 2016 and came from College Navigator; endowment data for 2015 came from IPEDS Largest 120 Endowments in 2015

    Lessons learned from the table 1:
    Note: The undergraduate enrollment percentages in rows 2 through 5 for each university don't add up to 100 percent because the table only includes data for the nation's four largest racial/ethnic groups; its figures do not include percentages for students from smaller minorities, students of mixed parentage, or students whose racial/ethnic backgrounds was unknown.
    • According to rows 7 through 10, the richest private institutions are graduating around 95 percent of their students in six years. Affirmative points granted to legacies, financially pressed students, underserved minorities, and other categories have not impeded the admission of academically qualified applicants.
    • More specifically, the table shows that Black students are graduating at more or less the same 6 year rates as other students, including Asians, at the five wealthiest universities. Evidently, the Black students are academically qualified to attend these institutions, no matter how loudly some misguided affirmative action critics may scream to the contrary.
    • The narrow range of the percentages of enrolled racial/ethnic groups is noteworthy: Whites 35% to 46%, Hispanics 9% to 16%, Blacks 6% to 8%, and Asians 17% to 26%.
    • Row 13 shows the percentage of undergraduates who received Pell grants. These grants are only awarded to students from financially stressed families, the kinds of students who would have qualified for affirmative action points using the rubric discussed in the previous section of this note 
    • At this point we can no longer ignore the elephant sitting quietly at the end of the table: the California Institute of Technology, a/k/a Cal Tech. Most of the numbers in this last column are head shakers. Row 1 shows that Harvard's undergrad enrollment (9,915) is ten times as large as Cal Tech's (979) ... Row 13 shows that Harvard's endowment ($37,616,000,000) is more than eighteen times as large as Call Tech's ($2,079,000,000) ... But Row 11 shows that CalTech's 25th percentile for the SAT math exam was 780, 10 points higher than M.I.T's, and at least 50 points higher than the other four universities. Not shown in the table is the fact that CalTech's 75th percentile for math was 800. In other words, 25 percent of Cal Tech's incoming class in 2016 scored perfect 800's on their math SATs. 
    • Another surprise: Rows 7 and 10 show that Cal Tech's six year graduation rates for White (93%) and Asian (93%) students were slightly lower than their graduation rates at the five richest institutions, but the graduation rate for its Black students (100%) was the highest rate in the entire table. Of course, the fact that this peak is based on a Black enrollment of only 10 students (1 percent of 979) throws a little shade on our celebration; but better the highest graduation rate for this Fab 10 than the lowest ... :-)
    • Cal Tech's most important distinction has been saved for last. Row 6 shows the gap between White and Asian enrollments. All entries are negative except for Cal tech's +14% because Asians represent a lower percentage of every other institution's total enrollment (Row 5) than Whites (Row 2). At Cal Tech, Asian students represent 43% of total enrollment vs. only 29% for White students. On the one hand, Harvard's endowment (measured in millions of dollars) per undergraduate is almost twice as large as Cal Tech's, i.e., $37, 616 / 9915 = 3.8, vs. $2,079 / 979 = 2.1.  On the other hand, 75 percent of Cal Tech's students score above 780 on math SATs and 25 percent scoring perfect 800s. Taken together, there's little room for alums to have the power to make legacies play a significant role in Cal Tech's admissions. In other words, Cal Tech can be regarded as a natural experiment. Its enrollments are what admissions at top tier institutions might look like if legacies and other forms of affirmative action had minimal influence.
    • What are the actual percentages of legacy enrollments at these institutions? It's difficult to say because they don't publish their rubrics nor their admissions data. Some observers believe that legacies represent about 30 percent of Harvard's enrollment based on a survey of the incoming class in 2017 by the Harvard Crimson, its school newspaper. But the Crimson stated that "Of the roughly 1,700-member class, 853 freshmen responded, representing roughly 50 percent of the class. The Crimson did not adjust the survey results for any possible selection bias." In other words, the survey was not random; we can't be sure that its respondents reported their legacy status accurately; nor can we be sure that the legacy share of non-respondents was as low as the reported share of the respondents.
    • Based on their access to Harvard's admissions files as part of their lawsuit's disclosure process, the advocates for Asian students at Harvard have claimed that if admissions were solely based on grades, test scores, and other materials in the files, then Asian students would represent more than 40 percent of Harvard's enrollment, instead of 17 percent. Cal Tech's 43 percent Asian enrollment strengthens the plausibility of this claim.

    C. Questionable Assumptions
    Daniel Patrick Moynihan -- a sometime Senator from New York and a sometime member of Harvard's faculty -- was one of the first to observe that Black Americans were not the primary beneficiaries of the 1960s Civil Rights legislation that outlawed segregation and thereby bent the arc of America's moral universe towards justice. The primary beneficiaries were Asian Americans. The Senator's observation underscored the bitter irony that Asian Americans were on the sidelines while Dr. King and his brave Black followers risked their lives in the marches and other demonstrations that ultimately produced resounding legislative victories. Nevertheless Asian Americans enjoyed a far larger share of the spoils, especially in higher education. Time after time when affirmative action programs that briefly expanded Black access to higher education in public institutions were struck down by the courts, Asian shares of enrollments surged while Black shares fell ... but so did White shares, and usually by larger percentages than the Black shares.

    This brings us back to the simple rubric that was created to explain the admissions process at a Great University, like Harvard. Friends and colleagues who claim some access to the inner workings of Harvard's admissions have assured me that there is no "rubric". GPAs, SATs, ACTs, and other scores are given serious consideration; but there is no formal multi-step process such as has been described. No matter. An "implicit" rubric could still be inferred from Harvard's enrollments from one year to the next. 

    However my friends and colleagues' remarks suggest the possibility of a major flaw in my implicit rubric ==> its assumption of fixed quality points for affirmative action categories, especially for legacies. Given Harvard's history, we can safely assume that the overwhelming majority of its legacies are White. If legacy points were fixed, we would expect greater variability in the percentage of Whites, Hispanics, Blacks and Asians in Harvard's enrollments from one year to the next. But if legacy points were adjusted from year to year so as to retard reductions in the legacy shares of Harvard's enrollments, there would be less variability in the shares of all four categories. So let's take a look at some enrollment data.

    Tables 2a, 2b, 2c, and 2d (below), show enrollment shares of Whites, Hispanics, Blacks, and Asians for the last seven years at two institutions: Harvard and Cal Tech. Once again Cal Tech serves as a reference point for what enrollment shares might look like if legacy and other affirmative considerations were minimized.
    • The first seven columns display the institution's name and the percentage shares of racial/ethnic groups from 2016 back to 2010. 
    • The last column displays the standard deviations of these percentages
    • The last column of the third row displays the ratio of the standard deviation at Cal Tech (first row) to the standard deviation at Harvard (second row)

    Table 2a. Enrollment % Shares for White Students
    Institution
    2016
    2015
    2014
    2013
    2012
    2011
    2010
    StdDev
    Cal Tech
    29
    27
    28
    30
    31
    35
    37
    3.70
    Harvard
    43
    45
    48
    48
    48
    49
    48
    2.16







    Ratio
    1.71

    Table 2b. Enrollment % Shares for Hispanic Students 
    Institution
    2016
    2015
    2014
    2013
    2012
    2011
    2010
    StdDev
    Cal Tech
    12
    12
    12
    10
    10
    8
    6
    2.31
    Harvard
    11
    10
    10
    9
    9
    8
    8
    1.11







    Ratio
    2.08

    Table 2c. Enrollment % Shares for Black Students 
    Institution
    2016
    2015
    2014
    2013
    2012
    2011
    2010
    StdDev
    Cal Tech
    1
    1
    2
    2
    2
    1
    1
    0.53
    Harvard
    7
    6
    6
    6
    6
    6
    6
    0.38







    Ratio
    1.41

    Table 2d. Enrollment % Shares for Asian Students 
    Institution
    2016
    2015
    2014
    2013
    2012
    2011
    2010
    StdDev
    Cal Tech
    43
    45
    44
    42
    40
    39
    39
    2.43
    Harvard
    17
    17
    17
    16
    15
    15
    14
    1.21







    Ratio
    2.00
    Note: All data in Tables 2a, 2b, 2c, and 2d came from IPEDS 

    Lessons learned from Tables 2a, 2b, 2c, and 2d:
    • The standard deviations of the enrollments at Cal Tech ranged from 1.41 to 2.08 times as high as the standard deviations of the enrollments at Harvard between the years 2010 and 2016. These indicators of greater variability are consistent with an hypothesis that Harvard used legacies and other affirmations more extensively than Cal Tech.
    • Table 2b. The ratio was largest for Hispanics (2.08). Whereas Hispanic enrollment doubled at Cal Tech, from 6 percent to 12 percent, it only increased by 3 percent at Harvard, starting at 8 percent and ending at 11 percent. 
    • Table 2d. Asian enrollments had the second largest ratio (2.00). Starting at 39%, almost three times as high as Harvard's 14% in 2010, Asian enrollments at Cal Tech rose then declined, but nevertheless finished 4% higher at 43%. By comparison, their enrollments at Harvard steadily climbed from 14% to 17%, a position that was 26% below their position at Cal Tech in 2016. 
    • Table 2c. The ratio was smallest for Black students (1.41). Whereas Black enrollment rose, then fell at Cal Tech, at Harvard it remained fixed at 6 percent for six years, then rose to 7 percent in 2016. Whereas such constancy is commonplace in physics and chemistry, it is remarkable with regards to most aspects of human behavior.
    • Table 2a. White enrollments declined every year at Cal Tech by different amounts from one year to the next from 37% in 2010 to 27% in 2015, then rose to 29% in 2016. White enrollments at Harvard were more or less 48% until 2014, then they dropped by 3% in 2015 to 45% and by another 2% in 2016 to 43%  As per the 1.71 ratio, White enrollments at Cal Tech were more variable than their enrollments at Harvard, but its two year decline in 2015 and 2016 could be signaling Harvard's commitment to lower White enrollment shares for the foreseeable future. 
    In an old movie, whose name I can't recall, a jaded aristocrat gave something like the following advice to his poor, but ambitious young friend: "The upper crust of our society would rather ride two to the saddle than give up the hunt. So given a bit of a shove, they always make room." Now that Asian advocates are shoving and Harvard is moving aside, is Harvard moving fast enough?

    D. Conclusions
    The anti-affirmative action factions inside and outside the Trump administration who are encouraging the Asian challenge to Harvard's admissions haven't read the numbers. As can be seen from Table 2b and 2c, in the Fall 2016 semester Harvard's Hispanic and Black students only constituted 11% + 7% = 18% percent of its enrollment. If the rejected Asian applicants had better grades and test scores and other measurable qualifications than, say, half of these academically qualified Hispanic and Black students, Asian enrollment would increase by 9%. According to Table 2c, this gain would only bring Asian enrollment up to 17% + 9% = 26%, which is well below the 40 percent goal their advocates have referenced. 
    • Readers should understand that my concession of half of the enrollments is based on my assumption that Asian advocates might believe they could discount some of the qualitative characteristics of the Black and Hispanic applicants that Harvard's admissions office might have valued, e.g. leadership and grit, as demonstrated by successful participation in competitive sports ... and commitment to public service, as demonstrated by after-school volunteer activities in neighborhood community organizations.
    Assuming that Harvard's academically qualified legacies are 30 percent of its enrollment, then its legacies are the biggest obstacles by far to higher Asian enrollment. If rejected Asian applicants outscored half of the legacies, Asians would achieve a 15 percent increase in their enrollment. In summary, it is only by displacing half of the academically qualified Hispanic + Black + legacies that Asian challengers can get to their 17% + 9% + 15% = 41% > 40 percent goal.

    The Asian challenge to Harvard's admissions is not about bogus "personality" metrics. It's about Harvard's right to give preference to academically qualified applicants who are financially stressed and/or underserved minorities over equally qualified Asian applicants. But ultimately it will be about Harvard's right to give preference to the academically qualified young relatives of its alums, alums whose donations and other support enabled Harvard to pay the high salaries that secured the employment of some of the world's finest scholars and provided them with cutting edge infrastructure to support their teaching and research. Without this generous support from its alums Harvard would not be the preeminent institution to which Asian applicants are clamoring for increased admission.

    Not being a lawyer, I have no idea what the courts will eventually decide. Perhaps some kind of Solomonic compromise will be achieved on the courthouse steps that adjusts Harvard's rubric to yield fewer legacies and more Asians. Otherwise the heavy handed intervention of the courts will be unavoidable. Hopefully, a precedent-setting compromise achieved outside the courts will not diminish the enrollments of financially stressed students, of Black students, and of other underserved minorities at Harvard nor at the nation's other top tier private institutions. 

    Roy L Beasley, PhD
    DLL Editor
    Harvard University, GSAS 1975


    UPDATE ...Monday 7/30/18
    Articles published in the NY Times and Inside Higher Ed this morning discuss lengthy depositions of Harvard admissions officers, notations in Harvard's admissions files, and Harvard's response to the lawsuit. In my opinion, these anecdotal disclosures are consistent with my statistical analysis in this blog note. Asian Americans represent a much smaller share of the U.S. population than Black Americans or Hispanic Americans, but their share of Harvard's admissions is much larger and has grown at a much faster pace. So the primary obstacle to further growth is the substantial number of predominantly White American legacies who are given preferential admissions.

    P.S. Articles providing background information about Edward Blum, the activist behind the current lawsuit against Harvard and other anti-affirmative action and anti-voting rights cases, have been published by the Boston Globe (2018), NY Times (2017), and Mother Jones (2016) 


    UPDATE ... Monday 8/6/18
    Inside Higher Ed just published an editor's review of an assessment of Harvard's admission of legacies and athletes that was conducted by the U.S. Department of Education in 1990. The assessment, obtained by Inside Higher Ed via a Freedom of Information request, covered Harvard's admissions during the 1980s. Although the Department found no evidence of discrimination against Asians, the comments of some of Harvard's admissions officers that were quoted in the assessment suggest that legacies and athletes were sometimes academically inferior to other students who were admitted. 

    While this might still be true today, Table 1 (above) of this note shows that Harvard's 25th percentile for the math SAT scores of entering freshmen is 730. In other words 75 percent of its full-time, first time undergraduates scored at least 730 on the SAT math exam. That's about two standard deviations above the national mean. Table 1 also shows that the 6-year graduation rate of its White students is 97 percent. These stats plus the Harvard Crimson's estimate that 30 percent of Harvard's enrollment are (mostly White) legacies suggest that Harvard's admissions standards are probably much higher today than back in the 1980s. Harvard may still be admitting a few "dumb legacies" and a few "dumb jocks" ... but no more than a few ... :-)

    Update ... Sunday 9/2/18
    Yesterday the U.S. Department of Justice declared its formal support for the  lawsuit against Harvard. According to this morning's reports in Inside Higher Ed, New York Times, and the Washington Post, the Justice Department did not address the impact of Harvard's most powerful affirmative action program, its preferential admission of legacies, on its other affirmative action efforts. 

    Harvard's response attacked the Justice Department's monolithic stereotype of all Asian applicants as achievers of the highest possible ratings on all admissions metrics, not just grades and standardized test scores. These Asian "model minority" stereotypes also imply correspondingly derogatory stereotypes about Black and other minority applicants as being uniformly inferior to Asian applicants on all metrics. (DLL Editor's note -- Here's a link to a page that links to documents filed by organizations, including Asian American groups, supporting Harvard)

    In my opinion, the Justice Department's brief reflects its Attorney General's misunderstanding of the non-remedial strategy employed by Harvard's admissions office, as described in "Part A" of this report, a strategy that only gives preferential advantage to Black and other minority applicants who are otherwise qualified for admission. It also reflects the Attorney General's long-standing personal bias against Black Americans, a bias that precludes his admission of the possibility that some Black applicants might be more qualified than some of the best Asian applicants, and more importantly, more qualified than some of the best White applicants. 

    Update ... Thursday 9/27/18
    Surely no one who has been closely following this tacky saga was surprised by yesterday's announcements from the U.S. Departments of Justice and Education (covered by NY Times, Inside Higher Ed, and Chronicle of Higher Edthat they are bringing their anti-Black, anti-Hispanic/LatinX, pro-Asian, White-legacy-neutral, anti-affirmative action five ring circus to New Haven. Yes, Yale will now be facing a lawsuit like Harvard's, filed on behalf of Asian plaintiffs and supported by the Trump administration's dynamic duo of higher ed. 

    The only thing that is surprising in this new Age of Data Science is that a specious line of reasoning that would be tossed aside in the exploratory stages of a properly conducted data science investigation should gain our courts' attention not once, but twice in high profile lawsuits. Obviously, introductions to data science have not become required courses in our nation's best law schools ... yet.

    It should be noted that the Asian American Coalition for Education has implored Justice and Education to investigate all members of the Ivy League. Again, no surprise.

    Update ... Saturday 9/29/18
    Yesterday a federal judge decreed that the case against Harvard University would move to a bench trial (no jury) in October, according to reports in Inside Higher Ed and the NY Times. In so doing the judge rejected pleas from both sides that the case be decided by a summary judgement from the court without a trial because summary judgements can only be issued when both sides are more or less in agreement on the fundamental facts of the case -- which they definitely are not in this case. Each has disputed at least half of the facts asserted by the other. 

    Perhaps neither side is operating within a data science framework. Modern data science decrees that after the data has been wrangled into manageable formats, the data must be thoroughly explored in order to determine which facts are the most important. Only then can an analyst determine which statistical and/or computational procedures will yield the most plausible explanations of why the data are the way they are. 

    There is also the possibility that one side has presented facts, whereas the other has presented "alternative facts". But what if both employ regressions and/or other statistical procedures -- tools of the trade for most analysts, but "sociological gobbledygook" for most judges, as per the lamentation of Chief Justice Roberts in a recent case. Then neither side will provide convincing explanations to a single judge, to an appellate panel, or to the Supreme Court itself. By contrast, data scientists never forget that the final stage of any investigation is the delivery of compelling presentations to non-data scientists, e.g., jurists.

    1 comment:

    Thank you!!! Your comments and suggestions will be greatly appreciated ... :-)