Thanks Matt for another thoughtful piece on an important issue. As a graduate of a selective college and a former teacher at a less selective institution, I agree that selection is often a much more powerful force than the impact of actually attending a college. However, because that is so, I think you (and us all) need to think more carefully about what is the "treatment effect" of going to college because there is a lot going on in those years.
It was noticeable to me in the college where I taught that many many students did not want to be there. Outside of the students who were there for its strong technical education in engineering (a minority) most were much more interested in social events, part time jobs, volunteer work, than school. And I think they were right. There is little content in a liberal arts/social science curriculum that is specifically necessary later in life. Most of my students intended to work in "helping professions" and I am ashamed to admit that there is little in either the psychology or sociology curriculum that will truly be relevant in work as compared to the policies and procedures of the individual workplaces.
So what is the benefit of attending college? There is additional maturation both physical and social - the differences between an 18 year old and a 22-24 year old can be pretty big. There are skills of self regulation - learning how to manage deadlines, juggle priorities, work with others, and supervisors. There is also an additional selective filter, given that many people don't complete a college degree - college is partially a test of endurance - of course it also is testing family finances, resilience, health and many other factors that don't seem legitimate as determinants of future success, but may in fact be correlated through unmeasured "third" variables anyway.
Finally, as you note college finances are complex. In your analysis, you are focusing on colleges as undergraduate educational institutions. They are also R&D hubs, industry partners, entertainment providers, and providers of graduate professional school education. When you say that we should give more resources to colleges serving those who will benefit the most, I get the idea in general, but the devil is in the details. When I was on faculty, state government, which paid less and less of college costs, only wanted to know about class room contact hours not acknowledging that much of faculty time was "sold" to grant funded research projects that brought in indirect revenues that supported research infrastructure that supported graduate students, staff, institutional prestige etc.
The college enterprise is pretty clearly something that just doesn't work for a lot of people. And yet our society has pretty much bought into a model that was created as a kind of "finishing school" for the children of economic and social elites who were pretty likely to be successful anyway. Perhaps we need to go back to some fundamental analysis of what we think is required for job and life preparation, what high school education should provide, what college is for and for whom, and what alternatives such as life-long learning, vocational learning, on-the-job training might look like.
That chart of University of California campuses left out UC Eugene. (Sorry; local Oregon joke. In 2020, more than 5,300 students at U of O were from California.)
The case for mismatch theory seems stronger than you suggest here. Check out this podcast with Duke University economics professor Peter Arcidiacono who breaks it down further: https://bloggingheads.tv/videos/59966
I can't figure out how to work the comments section now it's so huge so hopefully REF will see this- what I meant about learning fast is similar to a point David R. made somewhere on this page lol. I am a very fast learner - I'm good at memorizing and recognizing patterns, I had a very rigorous academic upbringing from when I was a toddler, and I'm confident in my intelligence. So picking up new skills, subjectively speaking, seems to be easier for me than it is for some other people. However, because I was precocious my whole childhood, I didn't learn other good skills like "not procrastinating" or "persevering through challenges" and a bunch I'm probably forgetting bc work was exhausting today. So the example I often give is I was an exceptionally good writer of essays in the sixth or seventh grade relative to my peers, but by the end of high school the other high performing students caught up. I think there could be a correlation between people who learn quickly and people who are more intelligent overall but I find I am over-praised for the quick part when people who get there over a month instead of a day may end up being better writers or Mandarin speakers or whatever it is, in a tortoise and hare way. You need to be quick *and* have follow through, I think. Or that's the ideal. I am definitely just now learning how to do the second part and I'm almost thirty.
My position is uniformally unpopular for college admissions. I love quotas.
Take the US population. Identify racial demographics. Set criteria cap for selective universities based on demographics with a 10% variance.
Everyone hates this. Over-represented minorities will hate it Chinese, Indians, Jews. Conservative White gentiles will hate it for the explicit set asides for African Americans and since it won't do much for them. Hispanics voted against Affirmative Action in California. Many African Americans would like it.
But I do think it is essential for US stability. People do care that people like them have some authority in society. I think liberals and conservatives are obtuse about because they have narrow conceptions of meritocracy. Conservatives somehow convinced themselves racism doesn't matter. Liberals are only concerned with harms of minority underrepresentation.
Will we also set quotas for IQ distribution? Height? BMI? Political party? Left-handedness? Literacy?
How will you handle people who don’t identify with a single race? Will you break down people of Chinese origin v Vietnamese? ADOS v African immigrants? Catholics v Protestants?
Apologies if you were being sarcastic and I missed it.
I actually do think you need to delineate between the broader categories such as African descendents of Slaves and African immigrants. What I would look at determining delineation is how much barriers exist between the groups (income, intermarriage, existing education levels) to see if it is necessary.
No one particularly cares about left-handness so I wouldn't measure that. But people do care that people who look them, belong to the same religion are represented in positions of authority. Societies that have elites that differ from the general population, either due to significant minority under-representation or significant majority under-representation, are societies more likely to be unstable.
And since the US is already a heterogeneous society and I want broadly open borders going forward, I consider it essential to stability to ensure the American elite looks roughly like the average American.
I don't think you'll be able to convince under-represented groups to suck it up and stop whining. Nor will you convince them their under-representation is due to merit, as conventionally defined by elites.
Bleemer’s paper specifically notes “About one-third of UC students transfer from community colleges rather than enrolling as freshmen. Because affirmative action was likely less impactful for those applicants and because of limited data availability about those students’ academic background (prohibiting selection correction on observables), transfer applicants are not directly analyzed in the present study, though freshman applicants may enroll at a community college and transfer to UC later.”
To my mind, this is a shame because the community-college transfer population makes a lot of sense for testing elements of Mismatch theory without an explicit focus on race. Setting racial considerations aside, I do think there are plenty of instances in which universities admit marginal students for reasons primarily motivated by revenue-generation (CC transfers being a prominent example).
It would be interesting to see how the life trajectories of these transfer students were impacted relative to other students who attend less-selective colleges, as well as to those who attend 4-year colleges out of high school. Does the finding that for black and Hispanic students “the estimates of the return to college selectivity remain large, even in models that adjust for unobserved student characteristics” extend to the community-college transfer population? If not, something other than inefficient meritocratic sorting seems to be driving these outcomes.
The top x% thing seems definitely better than the current flavors of "diversity" but what about guaranteed transfers from community college? That could have similar effects and help deal with the high school quality problems, since those usually impact the first year or two of college.
I feel like it's worth mentioning on the "credentialism" issue with the short listers that Candace Jackson-Akiwumi still went to Yale Law School. Princeton (which doesn't have a law school) was her undergrad.
Am I the only person who thinks we should just have quotas at colleges so that the ethnic/class composition bears some resemblance to that of the country? (And make sure that the quotas account for race/class intersectionality, so that elite colleges aren't just meeting their requirements with rich non-White kids.) Sure, that's going to mean lots more "otherwise qualified" rich White and Asian students don't go to Ivies so that more (still pretty damn smart) Black and Hispanic kids can, but honestly, who cares? The material benefits to the disadvantaged kids would be huge, increasing class diversity at those schools would make them less ideoogically extreme, and there's all sorts of benefits to representation that I think people often underrate (Matt often likes to cite the paper that found that zip codes with more female inventors produced more female inventors in the long run, because girls saw people like them being inventors and thought it was more of a possibility).
Yes, I know the current court would say this is super-duper unconstitutional, but in my mind this beats "colorblindness" and current affirmative action practices.
I think you meant this as a gotcha, but I am honestly not opposed to quotas based on height or things like that. If something is a genuine axis of non-meritocratic privilege, then we should try to equalize along it.
This column should start with the massive defunding of public universities throughout the country. What happens to the relatively small number of undergrads at a handful of "elite" schools matters far less to the country and to the general availability of opportunity than outcomes at public universities. The quality differences in California universities, for example, come from resource constraints (and CA is ahead of every other place). This defunding of the public universities is one of the sources of American decline (though it may be a symptom as well as a cause).
Second, we gone nuts over "merit" in some linear way as the basis for admission to the few places that are, in any meaningful way, selective. One of the great WASP leaders towards inclusion, Yale's Kingman Brewster, focused on "leaders." Of course that would select for strong intellectual talent but it could open to way to consideration of many other important factors in trying to help the country identify and support future leaders.
The issue with Top % admission is it's often taken away later. For example, to increase the average SAT scores for a state school and make it more competitive nationwide. This policy would need to be implemented at a federal level to make a real difference where it's needed most.
I'm confused by the chart adapted from Table 5 in the Ardiciacano paper.
The first thing to note is that the great reduction in Black/Hispanic enrollees is virtually entirely from the elimination of race as a category, with none from elimination of legacy/athlete preferences (which you can see from the actual Table 5 (http://public.econ.duke.edu/~psarcidi/legacyathlete.pdf)). It gives a distorted picture to lump all three categories together (granted that Milan is just following what the authors did).
Secondly, that Table 5 shows minimal impacts on white enrollment from elimination of legacy and athlete preferences. This conflicts with other things I've read (>40% of white enrollees were one or other of those categories) and indeed, seems to conflict with what the authors themselves write: "We find that a white typical applicant with a 10% chance of admission would see a five-fold increase in admissions likelihood if they were a legacy; more than a seven-fold increase if they were on the dean’s interest list; and that they would be admitted with near certainty if they were a recruited athlete." How they reconcile that statement with a ~5% decline in the number of white enrollees by eliminating those two preferences is beyond me.
The arguments I’m currently seeing on the center-left against affirmative action seem like a bit of a straw-man to me. The idea that affirmative action was supposed to make up for past injustices never made any sense since it ignores all the prejudice against Americans of Asian descent, and glosses over all the class injustices in our nation’s history.
Nevertheless, I still support affirmative action. And this is because it helps ensure that the demographic breakdown of the nation’s elite has some resemblance to the underlying distribution in the public at large. This is beneficial because a diverse elite really does appear to help underrepresented groups (see Pande 2003 for example). Also, the events of the past five years suggest that it is important to a whole lot of people that there are people who look like them in top companies and in power. I don’t buy that this problem can be solved by choosing more candidates from lower rated schools. Or rather I think this approach is fine for a federal judge appointment, but it won’t help when it comes to filling tech, finance, consulting, and other high prestige, high pay jobs. Hiring is hard and companies will always take the short cut of favoring candidates from elite schools, because there is some value to that signal.
To put it another way, a society with an elite that is vastly different from the underlying population seems like a much more unstable society. For the sake of harmony, I’d rather take the affirmative action hit.
Thanks Matt for another thoughtful piece on an important issue. As a graduate of a selective college and a former teacher at a less selective institution, I agree that selection is often a much more powerful force than the impact of actually attending a college. However, because that is so, I think you (and us all) need to think more carefully about what is the "treatment effect" of going to college because there is a lot going on in those years.
It was noticeable to me in the college where I taught that many many students did not want to be there. Outside of the students who were there for its strong technical education in engineering (a minority) most were much more interested in social events, part time jobs, volunteer work, than school. And I think they were right. There is little content in a liberal arts/social science curriculum that is specifically necessary later in life. Most of my students intended to work in "helping professions" and I am ashamed to admit that there is little in either the psychology or sociology curriculum that will truly be relevant in work as compared to the policies and procedures of the individual workplaces.
So what is the benefit of attending college? There is additional maturation both physical and social - the differences between an 18 year old and a 22-24 year old can be pretty big. There are skills of self regulation - learning how to manage deadlines, juggle priorities, work with others, and supervisors. There is also an additional selective filter, given that many people don't complete a college degree - college is partially a test of endurance - of course it also is testing family finances, resilience, health and many other factors that don't seem legitimate as determinants of future success, but may in fact be correlated through unmeasured "third" variables anyway.
Finally, as you note college finances are complex. In your analysis, you are focusing on colleges as undergraduate educational institutions. They are also R&D hubs, industry partners, entertainment providers, and providers of graduate professional school education. When you say that we should give more resources to colleges serving those who will benefit the most, I get the idea in general, but the devil is in the details. When I was on faculty, state government, which paid less and less of college costs, only wanted to know about class room contact hours not acknowledging that much of faculty time was "sold" to grant funded research projects that brought in indirect revenues that supported research infrastructure that supported graduate students, staff, institutional prestige etc.
The college enterprise is pretty clearly something that just doesn't work for a lot of people. And yet our society has pretty much bought into a model that was created as a kind of "finishing school" for the children of economic and social elites who were pretty likely to be successful anyway. Perhaps we need to go back to some fundamental analysis of what we think is required for job and life preparation, what high school education should provide, what college is for and for whom, and what alternatives such as life-long learning, vocational learning, on-the-job training might look like.
That chart of University of California campuses left out UC Eugene. (Sorry; local Oregon joke. In 2020, more than 5,300 students at U of O were from California.)
Bahaha. Thanks for the number though, proves its "California school" reputation is kind of real.
The case for mismatch theory seems stronger than you suggest here. Check out this podcast with Duke University economics professor Peter Arcidiacono who breaks it down further: https://bloggingheads.tv/videos/59966
I can't figure out how to work the comments section now it's so huge so hopefully REF will see this- what I meant about learning fast is similar to a point David R. made somewhere on this page lol. I am a very fast learner - I'm good at memorizing and recognizing patterns, I had a very rigorous academic upbringing from when I was a toddler, and I'm confident in my intelligence. So picking up new skills, subjectively speaking, seems to be easier for me than it is for some other people. However, because I was precocious my whole childhood, I didn't learn other good skills like "not procrastinating" or "persevering through challenges" and a bunch I'm probably forgetting bc work was exhausting today. So the example I often give is I was an exceptionally good writer of essays in the sixth or seventh grade relative to my peers, but by the end of high school the other high performing students caught up. I think there could be a correlation between people who learn quickly and people who are more intelligent overall but I find I am over-praised for the quick part when people who get there over a month instead of a day may end up being better writers or Mandarin speakers or whatever it is, in a tortoise and hare way. You need to be quick *and* have follow through, I think. Or that's the ideal. I am definitely just now learning how to do the second part and I'm almost thirty.
My position is uniformally unpopular for college admissions. I love quotas.
Take the US population. Identify racial demographics. Set criteria cap for selective universities based on demographics with a 10% variance.
Everyone hates this. Over-represented minorities will hate it Chinese, Indians, Jews. Conservative White gentiles will hate it for the explicit set asides for African Americans and since it won't do much for them. Hispanics voted against Affirmative Action in California. Many African Americans would like it.
But I do think it is essential for US stability. People do care that people like them have some authority in society. I think liberals and conservatives are obtuse about because they have narrow conceptions of meritocracy. Conservatives somehow convinced themselves racism doesn't matter. Liberals are only concerned with harms of minority underrepresentation.
https://americanaffairsjournal.org/2021/08/why-are-racial-problems-in-the-united-states-so-intractable/
This reminds me of this article, worth a read I think
Will we also set quotas for IQ distribution? Height? BMI? Political party? Left-handedness? Literacy?
How will you handle people who don’t identify with a single race? Will you break down people of Chinese origin v Vietnamese? ADOS v African immigrants? Catholics v Protestants?
Apologies if you were being sarcastic and I missed it.
I'm not being sarcastic.
I actually do think you need to delineate between the broader categories such as African descendents of Slaves and African immigrants. What I would look at determining delineation is how much barriers exist between the groups (income, intermarriage, existing education levels) to see if it is necessary.
No one particularly cares about left-handness so I wouldn't measure that. But people do care that people who look them, belong to the same religion are represented in positions of authority. Societies that have elites that differ from the general population, either due to significant minority under-representation or significant majority under-representation, are societies more likely to be unstable.
And since the US is already a heterogeneous society and I want broadly open borders going forward, I consider it essential to stability to ensure the American elite looks roughly like the average American.
I don't think you'll be able to convince under-represented groups to suck it up and stop whining. Nor will you convince them their under-representation is due to merit, as conventionally defined by elites.
Bleemer’s paper specifically notes “About one-third of UC students transfer from community colleges rather than enrolling as freshmen. Because affirmative action was likely less impactful for those applicants and because of limited data availability about those students’ academic background (prohibiting selection correction on observables), transfer applicants are not directly analyzed in the present study, though freshman applicants may enroll at a community college and transfer to UC later.”
To my mind, this is a shame because the community-college transfer population makes a lot of sense for testing elements of Mismatch theory without an explicit focus on race. Setting racial considerations aside, I do think there are plenty of instances in which universities admit marginal students for reasons primarily motivated by revenue-generation (CC transfers being a prominent example).
It would be interesting to see how the life trajectories of these transfer students were impacted relative to other students who attend less-selective colleges, as well as to those who attend 4-year colleges out of high school. Does the finding that for black and Hispanic students “the estimates of the return to college selectivity remain large, even in models that adjust for unobserved student characteristics” extend to the community-college transfer population? If not, something other than inefficient meritocratic sorting seems to be driving these outcomes.
using zip codes as a proxy for diversity, as Roberts suggested, might be a decent workaround for elite institutions too.
Thanks so much for staying on this issue. It matters and your audience has some agency over it.
The top x% thing seems definitely better than the current flavors of "diversity" but what about guaranteed transfers from community college? That could have similar effects and help deal with the high school quality problems, since those usually impact the first year or two of college.
I feel like it's worth mentioning on the "credentialism" issue with the short listers that Candace Jackson-Akiwumi still went to Yale Law School. Princeton (which doesn't have a law school) was her undergrad.
Am I the only person who thinks we should just have quotas at colleges so that the ethnic/class composition bears some resemblance to that of the country? (And make sure that the quotas account for race/class intersectionality, so that elite colleges aren't just meeting their requirements with rich non-White kids.) Sure, that's going to mean lots more "otherwise qualified" rich White and Asian students don't go to Ivies so that more (still pretty damn smart) Black and Hispanic kids can, but honestly, who cares? The material benefits to the disadvantaged kids would be huge, increasing class diversity at those schools would make them less ideoogically extreme, and there's all sorts of benefits to representation that I think people often underrate (Matt often likes to cite the paper that found that zip codes with more female inventors produced more female inventors in the long run, because girls saw people like them being inventors and thought it was more of a possibility).
Yes, I know the current court would say this is super-duper unconstitutional, but in my mind this beats "colorblindness" and current affirmative action practices.
why on race instead of just wealth or by height. https://www.apa.org/monitor/julaug04/standing#:~:text=The%20findings%20suggest%20that%20someone,found%20by%20psychologist%20Timothy%20A.
I think you meant this as a gotcha, but I am honestly not opposed to quotas based on height or things like that. If something is a genuine axis of non-meritocratic privilege, then we should try to equalize along it.
It is wrong to discriminate on the basis of race. Always
This column should start with the massive defunding of public universities throughout the country. What happens to the relatively small number of undergrads at a handful of "elite" schools matters far less to the country and to the general availability of opportunity than outcomes at public universities. The quality differences in California universities, for example, come from resource constraints (and CA is ahead of every other place). This defunding of the public universities is one of the sources of American decline (though it may be a symptom as well as a cause).
Second, we gone nuts over "merit" in some linear way as the basis for admission to the few places that are, in any meaningful way, selective. One of the great WASP leaders towards inclusion, Yale's Kingman Brewster, focused on "leaders." Of course that would select for strong intellectual talent but it could open to way to consideration of many other important factors in trying to help the country identify and support future leaders.
But how do you propose to divert resources away from the most selective schools? Did I miss something here?
The issue with Top % admission is it's often taken away later. For example, to increase the average SAT scores for a state school and make it more competitive nationwide. This policy would need to be implemented at a federal level to make a real difference where it's needed most.
I'm confused by the chart adapted from Table 5 in the Ardiciacano paper.
The first thing to note is that the great reduction in Black/Hispanic enrollees is virtually entirely from the elimination of race as a category, with none from elimination of legacy/athlete preferences (which you can see from the actual Table 5 (http://public.econ.duke.edu/~psarcidi/legacyathlete.pdf)). It gives a distorted picture to lump all three categories together (granted that Milan is just following what the authors did).
Secondly, that Table 5 shows minimal impacts on white enrollment from elimination of legacy and athlete preferences. This conflicts with other things I've read (>40% of white enrollees were one or other of those categories) and indeed, seems to conflict with what the authors themselves write: "We find that a white typical applicant with a 10% chance of admission would see a five-fold increase in admissions likelihood if they were a legacy; more than a seven-fold increase if they were on the dean’s interest list; and that they would be admitted with near certainty if they were a recruited athlete." How they reconcile that statement with a ~5% decline in the number of white enrollees by eliminating those two preferences is beyond me.
The arguments I’m currently seeing on the center-left against affirmative action seem like a bit of a straw-man to me. The idea that affirmative action was supposed to make up for past injustices never made any sense since it ignores all the prejudice against Americans of Asian descent, and glosses over all the class injustices in our nation’s history.
Nevertheless, I still support affirmative action. And this is because it helps ensure that the demographic breakdown of the nation’s elite has some resemblance to the underlying distribution in the public at large. This is beneficial because a diverse elite really does appear to help underrepresented groups (see Pande 2003 for example). Also, the events of the past five years suggest that it is important to a whole lot of people that there are people who look like them in top companies and in power. I don’t buy that this problem can be solved by choosing more candidates from lower rated schools. Or rather I think this approach is fine for a federal judge appointment, but it won’t help when it comes to filling tech, finance, consulting, and other high prestige, high pay jobs. Hiring is hard and companies will always take the short cut of favoring candidates from elite schools, because there is some value to that signal.
To put it another way, a society with an elite that is vastly different from the underlying population seems like a much more unstable society. For the sake of harmony, I’d rather take the affirmative action hit.