22 February 2011

Selectivity Matters Only Some Of The Time

Does attending an elite college increase your lifetime earnings?

No, says a new study, but applying to one does, whether or not you are accepted.

The implication is that the sorting process that causes rational students to choose to apply to particular colleges, rather than the quality of the educational experiences that take place in an educational institution, is what drives lifetime success, and that implications that flow from a belief held by students that admissions officers can sort the wheat from the chaff is more accurate than the actual accuracy of admissions officers in choosing the students most likely to succeed.


The study by Stacy Dale and Alan Krueger used looked at 19,000 students, entering college in either 1976 or 1989, controlling for SAT scores, high school grades, college applied to, college accepted by, college attended, and looking at their incomes after college based on Social Security administration records, for the graduates, some of whom were in their 40s or 50s when the study was conducted. I may even be in the study as all of the colleges and universities that I have attended are included in the study. In short, this is a very large study with very accurate data, and as explained below, a quite clever research design.


It turns out that attending an elite college (measured by selectivity) does not increase lifetime earnings for middle class second or later generation college students (who make up the bulk of all college students).

The starting point is the obvious fact that graduates of elite colleges make more money than graduates of less elite colleges. This pattern holds even when you control for the SAT scores and grades of graduates. By themselves, these patterns seem to suggest that the college is a major reason for the earnings difference. . . . They also controlled for the colleges that students applied to and were accepted by. . . . Someone who applies to Duke, Williams or Yale may be signaling that he or she is more confident and ambitious than someone with similar scores and grades who does not apply. Someone who is accepted by a highly selective school may have other skills that their scores didn’t pick up, but that the admissions officers noticed.

Once the two economists added these new variables, the earnings difference disappeared. In fact, it went away merely by including the colleges that students had applied to — and not taking into account whether they were accepted. A student with a 1,400 SAT score who went to Penn State but applied to Penn earned as much, on average, as a student with a 1,400 who went to Penn. . . .

[A] few major groups did not fit the pattern: black students, Latino students, low-income students and students whose parents did not graduate from college. “For them, attending a more selective school increased earnings significantly,” Mr. Krueger has written. Why? Perhaps they benefit from professional connections they would not otherwise have. Perhaps they acquire habits or skills that middle-class and affluent students have already acquired in high school or at home.

This finding is especially noteworthy because the new study included several historically black colleges, like Howard, Morehouse and Spelman, which are not as selective as Penn, Williams or other elite colleges. Students who choose a historically black college over an elite college may be hurting their future earnings potential.

Put another way, this study demonstrates that affirmative action in college admission benefits the intended beneficiaries without harming the individuals who end up at less selective institutions of higher education as a result.

I've seen similar results of this kind of study applied to law school admissions and career success with similar results.

Selectivity and Graduation Rates In General

Then again, these findings may only have validity among middle class, second generation or later college students at colleges that are at least somewhat selective who are basically playing the game of trying to go to a college that is close to the most selective one they can attend, even if this goal isn't the only factor in their decisions. At less selective institutions, where financial limitations and genuinely negative campus cultures can intervene, a college environment can drag a student down, even though any acceptable environment seems to allow students to reach their full potentials.

A "large study . . . has found that students who attend colleges with lower graduation rates are less likely to graduate — which does indeed have a huge effect on their earnings." The results of that study, set forth in “Crossing the Finish Line” by William Bowen (an economist and former Princeton president), Michael McPherson (an economist and former Macalester College president), and Matthew Chingos, a doctoral candidate, looked at records of 200,000 students at 68 colleges, including a much richer selection of less selective schools than the ones studied by Dale and Krueger show selectivity does matter. The study notes that "Among rich countries, only Italy is worse" in terms of graduation rates for enrolled college students, although in fairness, this is in part because the United States has one of the highest rate of college attendance and one of the largest share of adults earning degrees of any country in the world.

The first problem that . . . [they] diagnose is something they call under-matching. It refers to students who choose not to attend the best college they can get into. They instead go to a less selective one, perhaps one that’s closer to home or, given the torturous financial aid process, less expensive.

About half of low-income students with a high school grade-point average of at least 3.5 and an SAT score of at least 1,200 do not attend the best college they could have. Many don’t even apply. Some apply but don’t enroll. “I was really astonished by the degree to which presumptively well-qualified students from poor families under-matched,” . . . They could have been admitted to Michigan’s Ann Arbor campus (graduation rate: 88 percent, according to College Results Online) or Michigan State (74 percent), but they went, say, to Eastern Michigan (39 percent) or Western Michigan (54 percent). If they graduate, it would be hard to get upset about their choice. But large numbers do not. . . .

In effect, well-off students — many of whom will graduate no matter where they go — attend the colleges that do the best job of producing graduates. These are the places where many students live on campus (which raises graduation rates) and graduation is the norm. Meanwhile, lower-income students — even when they are better qualified — often go to colleges that excel in producing dropouts. . . .

[T]he only way to lift the college graduation rate significantly is to lift it among poor and working-class students. . . . it appears to have fallen somewhat since the 1970s. . . . Money is clearly part of the answer. Tellingly, net tuition has no impact on the graduation rates of high-income students. Yet it does affect low-income students. All else equal, they are less likely to make it through a more expensive state college than a less expensive one, the book shows.

The data are illustrated in the following chart:

Students with better grades and GPAs are far more likely to graduate at schools that are more selective than at those with high dropout rates.

A culture that accepts failure at less selective institutions may also be part of the problem, conveying the impression that dropping out is normal, rather than an outcome to be strenuously avoided.

A few colleges, like the University of Maryland, Baltimore County, have intensive programs that have raised graduation rates. The State of West Virginia has begun tying student aid to academic progress, and graduation rates there have risen. . . . When students fill out an online form for federal financial aid, the Obama Education Department now informs them of the graduation rate at any college in which they express interest.

Graduation Rates And Selectivity In Colorado

Colorado specific data are consistent with this general conclusion. As I noted in a blog post that I made a week ago:

For the 2003 cohort at 4 year Colorado institutions, 62.9% of students who did not need remediation graduates in six years, while 29.5% who did need remediation graduated in six years. Students requiring remedial instruction at schools with higher overall graduation rates were more likely to graduate that students requiring remedial instruction at schools with lower graduation rates. For example, 65.0% of students at the Colorado School of Mines who needed remedial classes graduated in six years, while only 19.6% of students at Metro State who needed remedial classes graduated in six years.

The question is begs, of course, is how much of the difference is due to sorting and how much of it is due to the culture of the institutions themselves. It could be that students at the Colorado School of Mines who need remedial classes are more affluent than those at Metro State, and are thus less prone to worry about whether their family's hard earned money should be spent on them when they seem themselves a marginal students who need remedial study. Or, it could be that Colorado School of Mines students need remedial work because the courses they needed to get them on track weren't available despite their strong talent for the material that they didn't study, or that many students who need remedial work there are academically talented students who lag in reading or writing because English is not their native language, who quickly recover from gaps in instruction, while remedial students at Metro State may generally lack academic moxie.

There are indications that this is happening at the high school level, as well as the higher educational level. Students who have been identified as gifted and talented who attend West High School or Lincoln High School in Denver, for example, both of which a high poverty, predominantly minority traditional high school programs are much less likely to attend college and not require remedial education than gifted and talented students who attend North High School in Denver, about a mile away from West, despite the fact that it is also a high poverty, predominantly minority traditional high school with a similar sized pool of gifted and talented students. North is doing something right, while West and Lincoln are doing something wrong. It may not be the fault of the principals and teachers involved who are simply dealing with the situation that has been created for them as well as they can. It may simply be that students at West and Lincoln are even more financially pressed than those at North, something that a simple free and reduced lunch program eligibility rate may not reveal. But, whatever the reason, high school environments do seem to have an impact on outcomes for comparable students in Denver.


Taken together, and in light of previous blogging of educational research that I've done here, these findings suggest the kind of priorities that we should have in public higher education.

* Subsidizing middle class students who want to attend highly selective private colleges and universities rather than the most selective public colleges and universities doesn't provide much value. Yet, a large share of college students are from affluent families.

* Tuition rates have little impact on the likelihood the children in affluent families will attend college or graduate from college.

* Affirmative action in higher education creates significant benefits without causing significant harm for African-American students, Hispanic students, low income students and first generation college students. These students benefit the most from attending more selective colleges and universities in lifetime earnings and in likelihood of graduation.

* The easiest way to increase the number of students who graduate from college in the short term is to increase financial aid support to academically qualified low and medium income students.

* Non-selective admissions policies that produce high college dropout rates undermine the likelihood that academically qualified students at those colleges will fail, and students who would otherwise be admitted by some level of selectivity in admissions have very high rates of academic failure and are more prone to not learn very much while they are in college even if they do graduate.

* Students who are not academically ready for college are more likely to fail at all levels. For example, the standards to determine that a student needs remedial education are basically the same at two year and four year institutions. For the 2006 cohort at 2 year Colorado institutions, 52.7% of students who did not need remediation graduated, while 19.3% of those who did need remediation graduated (from a two year program or with a certificate or after transferring to a new institution). For the 2003 cohort at 4 year Colorado institutions, 62.9% of students who did not need remediation graduates in six years, while 29.5% who did need remediation graduated in six years. But, at four year institutions with some level of selectivity, there is a graduation rate effect on the effectiveness of remedial education.

* Colorado has a system in which every high school graduate or high school dropout with a GED, no matter how academically unprepared will be admitted to some open admissions institution of higher education, and every student attending a public college or university (or one of three private colleges or universities in the state) receives an equal per student state subsidy simply for being from Colorado. This isn't unusual. Almost every state in the United States has this system.

The open admissions, universal equal subsidy system of Colorado and most state's higher education system has expensive flaws. This is a poor model despite it universality.

1. It subsidizes the very large share of college students, particularly at more selective colleges and universities, who are affluent and can afford to pay the full cost of their college education without a state subsidy. Reducing this subsidy would not significant reduce college graduation rates for upper middle class high school graduates students who are not first generation college students.

2. It admits large numbers of unqualified students who have little chance of earning a degree, who also tend to benefit least academically from their experience on average. About half of two year college students and about one in five four year college students at public colleges and universities need remedial education, and significant numbers of students who don't need remedial classes are still marginally qualified academically and likely to not graduate as a result.

3. Admitting large numbers of unqualified students devotes significant state funds to sending students who won't benefit much on average from their education (usually because they don't graduate), this also means that the state helping to create non-discharagable student loan debts for large numbers of students who won't even have the assistance of a degree earned to help pay for those debts making those students worse off than they were when they started. And, this means that a culture of failure is created that reduces the likelihood that academically qualified students will graduate.

4. This system under funds working class and middle class students who are academically ready for college. As a result, these students are much more likely to drop out for financial reasons and are much more likely to attend a college with a high dropout rate where the institutional culture makes it more likely for them to drop out.

Action Items

First, to restate a couple of conclusions from my post last week:

1. We need to change the culture at high school that are doing a poor job of sending a large share of well prepared entering high school students (such as students identified as gifted and talented), or simply to remove students with reasonable prospects of going to college from these high schools all together and remake these schools as non-traditional programs. We are doing these students a grave disservice by sending them to high schools like West High School and Lincoln High School as they exist today in Denver where their opportunities are greatly diminished.

2. We need to have something to offer its non-college bound students besides a watered down college preparatory curriculum that makes it worth the student's time and effort to stay in school and graduate, because the expectation that all students will go to and graduate from college is profoundly unrealistic for a large share of all students by the time that they enter high school, whatever prospects there may have been for this to happen earlier in the educational process. This should probably be focused on vocational programs, but this is a place where school choice among non-traditional programs may be particularly appropriate until there is consensus on what kinds of programs are most beneficial for non-college bound students. Realistically, students who enter high school well below grade level in multiple subjects or with a history of poor academic performance in middle school and earlier will not miraculously catch up in high school.

3. The dominant reason that students need remedial courses when they show up at colleges and universities in the state is weak preparation in math. The good news is that math is much more self-contained as a subject than reading and writing, and much more a product of the educational process than reading and writing which are part of language skills developed every waking hour (and many dreaming hours) for a student's entire lifetime. Beefing up the curriculum and our expectations of our students in math can reasonably be expected to make a dent in this problem.

To these I add these action items for public higher education in Colorado:

4. The best way to serve the goal of increasing access to higher education is to devote most scarce public higher education funds to scholarships that are based on both financial need and academic merit.

5. The state should greatly reduce subsidies for college students who are not in financial need, although perhaps a token subsidy (e.g. $2,000 per year) could be made as a reward to the academically highest achieving students (e.g. students whose grades and test scores put them in the top 10% of high school students) without regard to financial need.

6. The state should set a minimum floor of academic achievement that should apply as a prerequisite to admission for a large share (perhaps 90%) of students at any institution of higher education in the state (e.g. a student has graduated from high school, does not require remedial education, had at combined GPA/test score performance comparable to least a 2.7 GPA in high school/900 SAT/19 ACT, the standard at Mesa State College where about a third of students graduate). While there may be some room for exceptions in exceptional cases, the general rule should be that students who need remedial coursework, or who had a high school GPA and test score combination below the threshold level should not be admitted to any public colleges and universities, even two year programs at community colleges.

7. The state should set a standard of academic achievement significantly higher than the standard that applies to admission to any institution of higher education to achieve a scholarship covering 100% of financial need in grant form to those students whose academic preparation suggests that apart from financial need they have a very high likelihood of graduating roughly similar to the standard for admission to a selective institution of higher education (for example, a high school GPA of 3.3 and combined SAT score of 1100 equivalent). Students eligible for admission but not meeting the higher standard might receive only 50% of their financial need in grant form and the remainder in student loan form.

8. The state should fund, from its K-12 education budget, separate remedial programs for former high school students at little or minimal cost that could make them eligible for admission to a higher educational institution.

9. The state should support, perhaps through community recreation centers, educational and vocational programs designed for former high school students who aren't academically prepared for a college education and for whom going to college is not their goal.

10. The state should stop subsidizing private college and university educations.

Meanwhile at the federal level:

11. We should cease to make non-dischargable federal subsidized higher educational funds available to for profit colleges in programs with low graduation rates. One way to address that would be to set minimum academic standards for odinary student loans and Pell Grants similar to those proposed for public colleges and universities - with eligibility reopened for those who manage to show satisfactory academic performance for a year or two without dropping out. A much more closely monitored program serving a much smaller number of students might be developed for perhaps 10% of all student loan and grant recipients, which would be rationed so that they would remain a small share of all federal higher education assistance at any given institution.

Alternately, some portion of federal student aid held might be held in trust for for profit colleges, and for any institution with a low graduation rate, until a student graduated, with colleges forfeiting and not allowed to charge students for any amounts incurred for students who don't graduate.

1 comment:

Andrew Oh-Willeke said...

Other views on the value of an elite education a linked here and here.