Correlation, Causation, and College: a guest (!) weekend open post

by regular commenter wj

A recent blog post at the Economist looks at a comparison of the financial impact of obesity. And compares it to the impact of a college education. (The Swedish study that they look at finds them to be of the same magnitude. (And even being just overweight apparently has negatives, just not quite as large.

Which leads me to a question: why does a college education go with higher lifetime earnings? The assumption, especially among parents pushing their children to go to college, seems to be that the relation is causal – getting a college degree allows you to earn more. To some extent, and in some fields, that is doubtless true. After all, you have to have a college education to become a doctor or a lawyer.

But overall, how true is it? My suspicion is that there are certain personality traits which are generally necessary in order to get through college successfully, i.e. to get a degree. And those same traits are valuable, in any job, in getting ahead and earning more money. Which is why some people drop out of college and get rich – they have the traits, and grab the opportunity before finishing off the degree. And why some people who never attended college still do well. They have the traits, but either didn’t have the academic inclination, or just didn’t have the economic resources which would let them do college.

To wrap this back to the economic negatives of obesity, some of them may be due to discrimination (just as the lower incomes of women do). But some of them may be due to the fact that the traits needed to earn more are the same ones needed to avoid obesity. So, causation, or merely correlation?

And if the latter, how do we convince parents to stop stressing out their kids by pushing the wrong things?

Comments

[Sniffnoy]

Don't forget the signalling hypothesis: Getting a college degree allows you earn more because it acts as a demonstration to others that you had the traits needed to get through college successfully.

[wj]

It would seem, however, that the utility of a college degreee as a signal has been significantly reduced as the portion of the population getting such a degree has skyrocketed. When 5% of the population has one, it says something if you have one. When more like 25% do, it says rather less.

[Nombrilisme Vide]

That conception does still exist, though. It may be less of a singular bar, but it's still a bar. E.g., all other qualifications to the side, the Army imposes a non-waiverable requirement for would-be officers to hold a Bachelor's degree. It can be engineering, it ca be liberal arts, it can be underwater basket weaving - as long as it's a Bachelor's, it meets the standard. There's definitely some class issues tied into that particular situation, though.

[wj]

I know if you want to get a commission (via the acadamies, as well as via ROTC) you have to have/get a college degree. But is it possible to go mustang (i.e. become an officer after being an NCO) without formally getting one?

It definitely is a class thing. The class markers have changed in the civilian world, but the military has not paid attention. I wonder if that is because the are getting less class-conscious, or just because they feel free to ignore the rest of society.

[liberal japonicus]

Thanks for this wj, and in another example of great minds and synchronicity, Paul Campos over at LGM wrote about this.

With my daughter starting to think about uni, this is an interesting topic for me. Currently in Japan, we have a massive oversupply of university places because of the current demographics, which nudges this from the 'interesting topic' category to the 'I'm screwed'.

[wj]

Thanks for that link, LJ.

I do think, however, that Campos has missed something obvious. He contends that more education credentials (high school diplomas as well as college diplomas) do not indicate an increase in human capital. His basis for this is that, while both have increased substantially over the past 40 years, mediam incomes have been essentially stagnate (in current dollar terms).

But he mentions in passing that per capita GDP has rougly tripled in that time. (And, by the way, the cost of an undergraduate education has also tripled.) Which combination of facts could also suggest that human capital has indeed increased substantially . . . but the reward for that increase hasn't gone to those people, but to those who employ them.

[Nombrilisme Vide]

wj, nope, the Army wants the Bachelor's even if you go to Officer Candidate School. Up until 2009 you could squeak by with 90 credit hours completed, but at this point the only way you can get a commission w/o a 4-year degree in hand would be as a warrant officer, which is generally quite a different career path - as well as a substantially lower payscale... the class identity is more blurred, though.

I don't know if the other branches of service let their prior service candidates commission w/o 4-years, but given the drawdown (and weak economy) I'd be surprised if they did right now.

[wj]

Yeah, increased supply (of college degrees) would tend to mean that there isn't any incentive for the military to relax the standard.

[ua]

But overall, how true is it?

Pretty true. What is the R squared?

[ua]

"And those same traits are valuable, in any job, in getting ahead and earning more money."

Perhaps. Successful sales people are highly sought after and paid well. The personality traits for success in that field do not match those required to succeed in an academic setting by any means.

I speak from experience. In the course of my economic career I have experienced both. I was a very successful student, but the worst salesman (100% commission based no less!) you ever met.

[Laura Koerbeer]

And it works the other way around: lacking the traits, a person can get through college, but the degree will not make that person a success outside of the ivory tower.

[liberal japonicus]

Great point, wj. I've always, for various reasons, been interested in this question in other countries where there isn't an oversupply. Australia is (or was) one example, the UK used to be another. 30 years ago, the number of graduates was not that big, because/and so you could get a job that could support a family without a university degree. It seemed to me that one of the reasons that worked out was that Australia had a decent labor union movement that could fight for a living wage.

When viewed from the pov of the individual, it certainly seems like not only a chicken/egg thing, but something that is just the way it is, but when you start looking at other societies, you realize that it doesn't have to be like that.

[wj]

The sad thing is, while the supply/demand ratio has changed and the monetary value with it, the message hasn't gotten thru to a lot of parents. Which means they areputting enormous pressure on their kids to do something, Something that is not going to have the result that they assume, and may even end up being counterproductive.

[Slartibartfast]

The assumption, especially among parents pushing their children to go to college, seems to be that the relation is causal – getting a college degree allows you to earn more. To some extent, and in some fields, that is doubtless true. After all, you have to have a college education to become a doctor or a lawyer.

Yes, but what happens when you have more people with law degrees than the market can support? OK, sure: like that could happen. Still, it's worth considering.

My point is that even if you build a causal link between degrees and outcomes (we will perversely use "income" as a metric for outcomes) based on, say, data from 2000, you cannot necessarily use that correlation/causation to predict outcomes 20 years from now, when the number of degrees in the employment pool is substantially different. As will be the job market.

In long run, though, we're all dead.

[wj]

But there will be a tendency for parents to assume (if only subconsciously) that what was important in the job market when they joined the work force is still what is important for their children 20 years later.

[Slartibartfast]

In the world of systems engineering, you may have some metric of importance that has e.g. two dozen parameters that influence it. Let's say you have one parameter that dominates; then for small variations in the value of that parameter, you get a linear change in the metric. However: once you've moved that parameter's value significantly, the performance metric is now at a different operating point, and the parameter that once dominated may have a lesser effect. Not all systems are linear.

This also holds in random variables: if you have a number of random variables that combine, the resulting variance is the sum of the variances of the random variables being summed. If one variance is a lot bigger than the others, then the variance of the combined variables will be close to the variance of the dominant variable, and you can control the variance of the sum directly. But if you decrease the variance on the variable that was initially dominant too far, it will no longer determine the output variance; other variables will begin to dominate. Thus you can have a linear system that yields a result that doesn't behave linearly.

This is analogy, so bear with me.

[Slartibartfast]

But there will be a tendency for parents to assume (if only subconsciously) that what was important in the job market when they joined the work force is still what is important for their children 20 years later.

This is a lesson a lot of people will have to learn and learn again. People have been failing to appreciate the tendency of markets to bust when there's a glut for hundreds of years. Maybe forever.

[wj]

People generally not only dislike change, they ignore it as much as possible. Even why doing so repeatedly hurts them.

[skeptonomist]

There seems to be a tendency to interpret these result in terms of obesity being the cause but that may have it backwards. Are smart people less likely to become fat? There are several correlations which may be more fundamental than the obesity-income one, especially that between intelligence-test results and obesity.