Elite/Ivy grads really do earn more? (new study)

Couple other things. Whether or not someone would see a StateU doctor is an individual choice. Whether or not someone would like to pay a Harvard grad more than a StateU grad is also a personal choice. But that’s really not relevant. What’s relevant is what the markets would pay. That’s what the study assesses. So why talk about personal choices and not talk about how the findings of the study can be harnessed to make better financial decisions for kids when it comes to college admissions?

That’s not the only substantive argument that has been raised.

Also, I think you mischaracterize the causation argument—it’s not that there is something non-measurable causing the gap, it’s the possibility that there is something that was not measured that is causing the gap.

That’s not a small difference.

That “something” could be the main thing, has 60%, 70%, or 80% the influence. Until researchers partition the influence of each factor and put a weight on it I won’t take their conclusions without questioning.

Think about what influences a kid’s final height. CC people may say, milk, water, exercise, sleep, whatever. Tell us how important each is. Tell us genetics contributes 70+%. Maybe the “something” we talk about here is the genetics in college admission or school differences.

Guess what? The genetics of admissions are not great mystery. If you play cello at the level of Yo-Yo Ma you don’t need to be valedictorian or have 780 on each of your SAT’s, as long as your grades and scores demonstrate that you are capable and motivated to do college level work AND will benefit from the kind of intellectual environment that a place like Harvard wants to provide.

But every Tom Dick and Harry will LOOK at Yo-Yo Ma’s grades and scores as evidence that THEIR precious snowflake has a shot at getting into Harvard with a 3.4 GPA and a 680 math, 700 verbal SAT score. Because everyone thinks their snowflake is Yo-Yo Ma on some dimension or other.

How hard is this to understand?

No study was published in a public journal, and the only peer review appears to be giving a presentation about the research at a conference, so we don’t have enough specific details about the research to understand what assumptions were made. Plenty of published studies exist on this topic, many which come to differing conclusions, depending on how the study was conducted. One of the most well known examples is at http://www.nber.org/papers/w17159 . The abstract states:

Note that the abstract mentions “SAT score of the colleges that students applied to” rather than attended. They found that applying to selective colleges was more correlated with earnings than the college attended , which suggests applying to selective colleges and earnings are both correlated with certain personality traits/background/connections/internal goals/…

I could also link to studies that came to quite different conclusions than the quote above. My point is that the conclusions vary with the conditions and assumptions of the study, so I’d wait until the research discussed in the original post is published, rather than just a brief news article with many missing details about the research methods, before making large conclusions.

dfbdfb, you write: “Also, I think you mischaracterize the causation argument—it’s not that there is something non-measurable causing the gap, it’s the possibility that there is something that was not measured that is causing the gap.”

Everything is possible, but not everything is probable. What you said above can apply to every single scientific study the results of which you don’t like. What is this thing that was not measured? If you can’t say what it is, then it is no better than saying that any conclusion is the wrong one because it is always possible that some important fact was not considered. You can’t disprove that there is no such important fact, as you can’t prove a negative.

Why it is so hard to accept that ceteris paribus an elite education gives you benefits that a non-elite one doesn’t give?

data10, you said: “They found that applying to selective colleges was more correlated with earnings than the college attended”

This is not surprising, as high performing kids apply in higher numbers to selective schools. Which selective school you get into - Harvard or Yale, say, for example - will of course have a very small effect, if any at all.

They found that if a student gets rejected by Harvard/Yale and instead goes to a less selective college, the student’s career earnings are much more similar to Harvard/Yale students than students at less selective college he attended. The selectivity of the college attended added little to the prediction beyond the prediction based on the most selective college applied to (and other control variables) with the exception of certain subgroups, such as URMs with disadvantaged backgrounds. The author writes:

Dale & Krueger have been criticized because they didn’t compare colleges that are far enough apart in selectivity/prestige (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2473238). But when you try to do that, it may be even harder to find a statistically significant body of highly capable students who wind up at far less selective schools. The available data may be too sparse to support the complex comparisons this problem requires.

According to the article cited by the OP,
“… the study finds significant differences in earnings at all colleges, with women earning less, and with degree program playing a major impact. As a result, the researchers are dubious of rankings of colleges by income of graduates – including those based on data from the College Scorecard.” (emphasis added).

So I think the right take-away message is not that there is a financial payoff from attending the most competitive schools (maybe there is) but that it isn’t necessarily so great to be able to recommend, confidently, that it makes sense to pay a large up-front price premium for that benefit alone. Can available social science research predict a reasonable ROI break-even point (e.g. that College X is worth paying up to N% more than College Y for the ROI)? I don’t think so.

@1Wife1Kid, sorry, but that rebuttal doesn’t hold water, both because of the reasons others have posted, and because “Everybody has the same flaws!” isn’t a valid defense of a study. This is part of why we don’t just accept a single study of a phenomenon, but rather researchers conduct replications (to validate the original study) and studies that look at different factors (to test whether previously-unmeasured issues are at play).

Simply accepting one study, especially a study of an extraordinarily complicated social phenomenon, is, in a word, silly.

“Can available social science research predict a reasonable ROI break-even point (e.g. that College X is worth paying up to N% more than College Y for the ROI)? I don’t think so.”

Should future social science research try to predict a reasonable ROI break-even point (e.g. that College X is worth paying up to N% more than College Y for the ROI)? I think so.

dfbdfb, I didn’t say that “Everything has the same flaws.” I said that if you want to find flaws with any research based on hand-waving, you can.

“They found that if a student gets rejected by Harvard/Yale and instead goes to a less selective college, the student’s career earnings are much more similar to Harvard/Yale students than students at less selective college he attended.”

That is also not surprising, as if a sports team goes down a division, their performance would continue to be more similar to the higher division than the lower, as they will be in the highest echelons of the lower division, and pretty close to those in the lowest echelons of the higher division.

The paper uses the phrase “generally indistinguishable from zero”, suggesting that the kid who is rejected from Harvard and goes to Tufts instead has a “generally indistinguishable from zero” difference in career earnings from the similar background/controls kid who goes to Harvard. The Harvard reject attended Tufts kid is actually predicted to have a slightly higher income than the attended Harvard kid among the subgroup of persons from well off families whose parents both have college degrees.

Like I’ve mentioned in earlier posts, you can find studies on this topic that come to a variety of different conclusions, depending on methodology and assumptions. One can find plenty of critiques of the Dale & Krueger publications. I’d list even more issues with the study you linked to. My point was that we should not treat a single study on this topic as gospel, particularly one that has not been published, so we don’t know certain key details about methodology and assumptions, like the one discussed in the original post.

Pointing out that there could be other explanatory factors at play—particularly when some reasonable possible ones have been identified, as has happened upthread (you did read the thread before saying nobody was really discussing it, yes?)—is not “hand-waving”.

I really don’t get why this is being something you’re being so rigid about. I mean, if my social-science-researcher self submits an article for peer review, I expect other possible explanatory factors to be pointed out by the reviewers. I can then conduct a follow-on study to test for a possible effect or come up with a principled reason for not dealing with them, but whatever I do, dismissing such critiques as “hand-waving” would be a marker of being a pretty poor researcher, you know?

dfbdfb, I am not being rigid about it. I am being ecstatic. Too many colleges charge the same tuition without being able to provide their graduates with the same future as elites. I am hopeful that research such as this will shift the tuition curve and make less impactful colleges significantly cheaper. That will be good for everyone in USA.

“The Harvard reject attended Tufts kid is actually predicted to have a slightly higher income than the attended Harvard kid among the subgroup of persons from well off families whose parents both have college degrees.”

The article in the OP seems to indicate that adjusting for all other discernible factors the average kid at Tufts will do worse than the average kid at Harvard in terms of lifetime earnings. That’s a very important finding. If you believe that the lifetime earnings of the average kid at Tufts would go up significantly if they only send in an app to Harvard, then they should do so. It’s a cheap safety net.

However, it doesn’t seem logical.

1wife- do you know anyone who believes that Hofstra provides the same educational opportunities as CalTech? Do you know anyone who believes that going to Adelphi will put his or her kid on equal footing as going to MIT?

These schools charge what they charge because they can- not because there is a lack of research or data on earnings potential and the tuition curve. Hofstra gets away with their tuition because people will pay it- not because anyone is delusional enough to think that the quality of instruction and the employment opportunities are on par with Stanford or Harvard.

Yes, the different studies came to different conclusions because of different methodologies and assumptions, as I have repeatedly said. In the study I linked, applying to selective college was used as a control for various individual student variables. The vast majority of high achieving students do not apply to any Ivy type elite schools. Only a small portion do, who have unique characteristics from the general population. This might include things like different personality traits and different encouragement/pressures from family. The study found that the higher earnings followed the unique traits that were correlated with applying to elite colleges, rather than attending an elite college. Maybe students who are more focused on attending HYPSM… are also more likely to be focused on pursuing careers associated with higher incomes than other high achieving students. Or maybe students who prefer to stay in their immediate area for college are also more likely to want to stay in their immediate area after graduating, which more limits job options than students who are thinking nationally. There are many explanations that “seem logical” to me.

Regarding your comments about choosing how much extra to spend for elite colleges for the increased income, note that HYPSM… type elite colleges cost less than state schools for ~90% of families in the US, and the study I mentioned and others (such as the one at http://aer.sagepub.com/content/early/2014/07/25/0002831214544298.full.pdf+html?ijkey=iAyUJ9v44wd1I&keytype=ref&siteid=spaer ) have found that increased tuition is associated with greater career earnings after controlling for other variables, including selectivity. Possible explanations include things like students being more likely to complete the degree with a greater financial investment, and higher tuition schools being more likely to have better counselling/career/individual resources. I believe the study linked in the OP only looks at students who have graduated with a specific major, so it does not consider things like students having differing chance of graduating or switching out of majors associated with higher incomes at different colleges, missing this type of effect. This is the kind of differing methodology and assumptions I am referring to.

This. Has. Not. Been. Demonstrated.

ETA: Conventional wisdom supports the claim, sure. But CW is not good social science research.