07 June 2010

Meritocracy v. Privilege In College Admissions

Richard D. Kahlenberg, in a Washington Post op-ed picked up by the Denver Post, headlined "5 myths about college admissions, has a little problem. It debunks myths that are true. It is one of those lies, damned lies and statistics stories.

The Facts

Some of its facts are either true or not profoundly far from the truth.

* "[A]t the most selective universities and colleges, 74 percent of students come from the richest quarter of the population, while just 3 percent come from the bottom quarter." A large share of low income college bound kids need remedial education, particularly in math, which is not a recipe for getting into a selective college.

* "Legacy preferences — an admissions boost for the children of alumni — increase a student's chances of admission by, on average, 20 percentage points over non-legacies."

Those estimates may be somewhat overstated, however. Large numbers of legacy applicants (usually a majority) are still rejected. The statistics on the subject compare the overall admission rate of the college to legacy admission rates, without adjusting for academic ability. But, a lot of people with little hope of admission apply to Ivy League and other highly selective colleges as "reach" schools, while it is likely that fewer Legacy applicants who a weak academically do. Clearly being a legacy does provide an admissions edge, but that edge is not quite as pronounced as it seems comparing only raw numbers. Also, the total numbers of legacy admissions, on the order of 10%-30% of students at the most highly selective institutions, with only some gaining admittance because of their legacy status, and far fewer at less selective institutions, number in the high thousands or low tens of thousands out of a couple million college students a year.

Despite legacy admissions, controlling for academic ability, high socio-economic class is actually a disadvantage in the private college admissions process.

* "[There is a] 784-point difference in SAT scores between the most advantaged and the most disadvantaged students. All other things being equal . . . there was a 56-point difference between black and white students. Most of the rest of the gap was the result of socioeconomic factors. . . . at highly selective institutions, for students within a given SAT range, being a member of an underrepresented minority increases one's chance of admission by 28 percentage points. That is, a white student might have a 30 percent chance of admission, but a black or Latino student with a similar record would have a 58 percent chance of admission. . . . students from poor families don't receive any leg up in the process — they fare neither better nor worse than wealthier applicants."

* "At the University of North Carolina at Chapel Hill, for example, a generous financial aid program, the Carolina Covenant, was instituted in 2004. Under its terms, low-income students are not required to take out loans as part of their financial aid. . . the program has been successful in . . . boosting the graduation rate among low-income students. Traditionally, low-income and working-class students drop out at much higher rates than do higher-income students, as financial worries and jobs with long hours distract from their studies. . . . the Carolina Covenant raised the four-year graduation rates of low-income students by almost 10 percent. . . . [The program applies to students] eligible for federal Pell grants, 90 percent of which go to students from families making less than $40,000 a year[.]"

The statistics cited in the op-ed addressed retention, but finances also greatly influence whether kids attend college at all:

High academic achievement kids in high income families have a 97% chance of going to college, while it is 78% for low income kids. For those in the second quartile of academic achievement kids in high income families have a 90% chance of going to college, while it is 63% for low income kids. For those in the third quartile of academic achievement 85% chance of going to college, while it is 50% for low income kids. For those in the bottom quartile of academic achievement kids in high income families have a 77% chance of going to college, while it is 36% for low income kids.


Increased grant based financial aid is one of the few proven ways that both college attendance and college retention can be improved significantly for children from less affluent families, without changing how kids are educated before they apply to college.

* "If a more selective school and a less selective school enroll two equally qualified students, the more selective school is much more likely to graduate its student. Future earnings are, on average, 45 percent higher for students who graduated from more selective institutions than for those from less selective ones, and the difference in earnings is widest among low-income students."

* "[S]tudents are going to college at a higher rate than ever before . . . white student representation declined from 79 percent to 58 percent at less selective and noncompetitive institutions between 1994 and 2006, while black student representation soared from 11 percent to 28 percent."

Analysis

The big problem, with Kahlenberg's reason is that he asserts that "Rich kids can't possibly be 25 times as likely to be smart as poor kids, so wealth and connections must still matter" and points to the use of the SAT in college admissions as the culprit. The trouble is that there are huge class differences in academic ability at the time kids apply to college, that he himself demonstrates. (I do not whole heartedly embrace the hereditary theory of IQ for reasons expressed here, but almost none of the criticisms apply to measures of academic ability as young adulthood where however one arrived where one did at that point, when the source of the differences that do exist doesn't matter because a person has been born, raised and educated in a way that can't be changed then.)

There is every reason to believe that SAT's are a reasonable approximation of academic ability (i.e. merit) and smarts (i.e. IQ). Few tests are better validated and numerous careful studies have validated the relationship (see, e.g., here and here). The fact that class differences are profoundly more statistically significant than racial differences in SAT scores, also suggests that ethnic or culture bias in SAT scores is modest and that this bias is not only compensated for, but overcompensated for, in the admissions process.

Indeed, he cites evidence that there is a 784-point difference in SAT scores (which have three subscores that are scored on a 200-800 point range summed for a total of up to 2400 from three subscores).

A point difference this large is roughly the difference between an average score at about the 87th percentile and an average score at about the 13th percentile; it is hardly a statistical fluke. Kahlenberg doesn't say so explicitly, but he implies that high scoring "advantaged" kids are in the top quartile of family income, and that the low scoring "disadvantaged" kids are in bottom quartile of family income. By the time that they take the SAT and apply for colleges, there is an immense difference between rich kids and poor kids in academic ability that the SAT measures. In fact, the actual average SAT score results are not different at a statistically significant level from the results that would result if a student's SAT percentile and family income percentile were always identical. Thus, a huge part of what looks like class privilege in college admissions which he quantifies can be traced to differences in academic ability.

Now, this is a matter of averages. There are some affluent kids who are academically weak, and there are some low income kids who are academically strong. The actual admissions rates by socio-economic class largely reflect that fact. But, the starkness of the socio-economic class exclusiveness in high education also show that a large share of outcomes are pretty close to the mean.

In order words, Kahlenberg has pretty much disproven his claim that "Admissions officers have figured out how to reward merit above wealth and connections" with his own data.

The assertion that "universities that have abandoned legacy preferences — or never used them — have plenty of alumni donors" also proves too little. The question is whether the legacy admissions influence how much is donated, not whether it alumni donation rates depend upon legacy admissions alone. And, it fails to recognize other admissions preferences that are commonly and quietly given to the rich for future donor potential, in the absence of a legacy preference, and the other non-need blind admissions (often subtle, like limits on the supply of financial aid packages to eligible students).

Instead, he has shown that apart from race and legacy status, that college admissions are largely meritocratic, that race and legacy status are much smaller factors than academic merit, and that selective colleges are picking the kids most likely to succeed. Indeed, in our unfair world, a college or university devoted to producing the most successful graduates might be less meritocratic and favor those of higher socio-economic classes more than the existing system.

Similarly, while Kahlenberg claims that it is a myth that "Generous financial aid policies are the key to boosting socioeconomic diversity," in fact, he demonstrates that the mix of financial aid afforded to low income students with academic ability does provide considerable benefit. He also fails to note that class mobility is much greater in countries with free public higher education. Kahlenberg also doesn't squarely address the question, implicit in his analysis, of whether low income kids with high academic ability more often go to less selective institutions because they are cheaper.

Yes, Amherst and Harvard can admit more low income kids if they want and have the pick of the cohort. There are some low income kids are have incredible academic ability, even if it would not normally be so great that it would lead to admission at their schools, and the most selective schools in the country can choose to admit them. But, this just reshuffles where a comparatively small number of academically talented low income kids are admitted.

The bottom line is that the lack of diversity based on socioeconomic class in American colleges and universities has a large and (in a meritocratic system) irreducible component that flows from huge, social class linked disparities in average academic ability at the time of college admission.

The affirmative action in admissions for low income kids with poor SAT scores that Kahlenberg is arguing for in his op-ed wouldn't necessarily help much either.

Academically ill prepared students who are admitted to colleges and universities have a very high dropout rate: "92 percent of students with low grades planning to earn an A.A. failed to do so—even higher than the 86 percent of those who abandoned their plans to earn a BA." Most college students who drop out because they are academically qualified do so very early on in their studies: "“It’s a pretty good rule of thumb that you will lose half of the people you will lose — either physically or psychologically — by the end of the first semester . . . you don’t find a lot of people flunking out for academic reasons after the second year."

Community colleges are generally open admissions institutions, and they have very lower retention rates as a result:

At Morgan Community College, only 30% are still in college a year after they start. The Community College of Aurora, the Community College of Denver, Trinidad State Junior College, Otero Junior College, Front Range Community College and Colorado Northwestern Community College all lose more than half of their students in the first year.


Admitting academically unprepared low income students to colleges and universities, only to have them drop out in short order doesn't make our society more meritocratic.

Reducing the importance of legacy admissions and improving financial aid can both reduce socio-economic biases in higher education, but there is only so much that they can do unless merit based college and university admissions based on academic ability are abandoned.

Is Our Society More Meritocratic Than It Used To Be?

Note that the existence of merit based college admissions, and the link between class and academic ability may well be related. College and university admissions became much more merit based in the 1960s. The children of the first cohorts of people for whom college admissions were made largely on the basis of merit are the ones going to college now.

Rich kids with weak academic ability from the 1960s on were likely to be downwardly mobile. Low income and middle income people with academic ability from the 1960s on were likely to be upwardly mobile. Socio-economic class membership has been much more closely tied to academic ability in the last generation than ever before in U.S. history. Our society is starting to look like that of ancient China where social advancement depended mostly on an ability to do well on Confucian philosophy examinations. When social class is closely tied to college admissions, is it so absurd that kids of the most academically able parents do best academically, while kids of the least academically able parents do least well academically?

Meritocracy is college and university admissions, and social class determination closely linked to education, fundamentally strengthens social class lines, by preventing "rotten heirs" from being kept in high social classes, and by preventing able people born to modest means from being stuck in the social class into which they were born. These twin flaws have been the sticking point in meritocratic reforms of rigid caste and class systems.

When Do Academic Achievement Gaps Emerge?

Of course, when you acknowledge that "by the time that they take the SAT and apply for colleges, there is an immense difference between rich kids and poor kids in academic ability that the SAT measures," the question that immediately follows is "when does this disparity arise?"

The evidence on this point is more clear than one might expect. It arises very early. Pre-school quality is statistically linked to a 47% difference in college attendance by age 24. While early academic ability is not destiny, "By [age] 5, it is possible to predict, with depressing accuracy, who will complete high school and college and who won’t.” There are immense socio-economic class disparities in academic performance (demonstrated, for example, by CSAP scores) by the middle of the third grade, and nobody seriously supposes that those differences were not already well progressed by the time children start the first grade. Students who start behind are unlikely to catch up and disparities increase slightly over the course of twelve years of education, for the most part, although most of the disparities are in place as soon as they are measured. Not surprisingly early disparities manifest in college attendance and retention which is closely related to academic preparation.

The early point at which academic disparities arise also belies the conclusion that actual material wealth is driving childhood disparties in academic achievement. The income divide between families that will be in the top quartile and bottom quartile when their kids are seventeen years old is typically considerably smaller when the kids are young. High income earners have lifetime incomes biased strongly toward the ends of their careers. Low income earners have incomes that increase more gradually over their lifetimes.

Social class disparities in academic ability probably are influenced a little by a child's K-12 education, but the impact is far more modest than most people would believe it to be.

There probably is more to academic outcomes by the time kids are applying to college than genetically inherited IQ. But, the overall impact of a child's genes and early childhood on later academic performance overwhelms other factors in kids who otherwise live typical lives, attend typical schools, and grow up with their parents.

1 comment:

Dave Barnes said...

Andrew,
Thank you for a cogent demolition of a crap DP article.