Hacker News Reader

Comment by throwawayjava | Hacker News Reader

Their familial controls don't make much sense to me.

An actuary's kid will have an easier time becoming an actuary. Same for lawyers and doctors. That doesn't mean that college is a non causal factor in those students lifetime earnings... Quite the opposite. You're not becoming any of those things without college.

The student went to college precisely because they understand viscerally and precisely how college sets them up for a life of higher than average earnings. And what they need to do at college to tap into that potential.

Similarly, I'm not sure I believe the claim that there's no obvious causal link between education and financial acumen.

The big difference seems to be "do you know WHY you are at college?" And in cases where the answer is yes, it's still a good choice.

View on HN

> Their familial controls don't make much sense to me.

There's a lot of recent evidence that, all else being equal (controlling for educational attainment, test scores, and grades) parental income/social class has a huge impact on income.

That is, if you come from a lower social class, you can expect a lower income increase from college than someone coming from a higher social class.

There's significant policy implications here-- our long-standing goal of getting lower income kids to take college loans and go to college may not have a positive expected return for them.

no, that’s completely wrong. You are confusing averages and conditional probabilities. On average, those from higher classes may do better - that doesn’t mean that a kid from s lower social class should take that as guidance. It means we should try to erase or lessen differences of opportunities between classes. Some of these differences are probably innate. A brain surgeon is likely smart and his kids are probably also smart. Some, but most are due to informational deficiencies and networking effects.

> no, that’s completely wrong. You are confusing averages and conditional probabilities. On average, those from higher classes may do better - that doesn’t mean that a kid from s lower social class should take that as guidance.

??? You're saying that someone should not consider the outcome distribution of people similarly situated to themselves when making a choice? Isn't that completely rejecting empiricism?

If, from a public policy standpoint, our effort has been to get to the poor to college to end poverty; and the educated poor do do better with education, but not nearly as well as their middle-class cohort; and the present-value of education is smaller than the present-value of their lifetime wage increases-- one has to question whether the policy is well informed.

If you're asserting that the picture is unduly distorted by those at the very highest spread of the income distribution-- you're right. But the effect is still when we look at median outcomes.

> Some of these differences are probably innate. A brain surgeon is likely smart and his kids are probably also smart.

Presumably we'd be able to measure this and control for this.

> Some, but most are due to informational deficiencies and networking effects.

Yes, and various kinds of cultural cues about class of upbringing that persist after education and act to limit opportunity.

“??? You're saying that someone should not consider the outcome distribution of people similarly situated to themselves when making a choice? Isn't that completely rejecting empiricism?“

No, what I’m saying is presence of a statistical pattern in a whole population is not evidence that we shouldn’t make policy to change it. Statistically, those who live in Miami are more tan than those who live in Wisconsin, on average. But if you expose a Wisconsin resident to the same amount of sun, on average they will be just as tan.

??? We're comparing the income difference (and ratio) between (poor background + college education) and (poor background + no college education) to the income difference between (non-poor background + college education) and (non-poor background + no college education).

In your example, you'd expect the Wisconsin guy to increase in tan more when you put him in the Bahamas and give him a dose of sun than the guy who started out in Miami and was given an equivalent further dose of sun.

But we've measured the opposite-- the guy in Wisconsin (poor background) does get more tan, but doesn't increase in tan (increase in income) nearly as much as the guy who started in Miami (non-poor backgrounds).

No, because we haven’t done nearly enough for the poor guy. It’s not the case that we’ve done everything possible for the poor and they still didn’t improve much. That’s false.

We're measuring the effect of one thing-- getting a poor person to college-- composed with all the other interventions that are spread through the population.

We measure the value of this-- getting to college-- as very small and even possibly negative.

We also measure interventions and their values / impacts on earnings much higher.

Going "la la la but college would be much better and itself a win if we did 50 other supportive things" is all great, but it's fundamentally handwaving in the absence of evidence.

No, we don’t measure this. The study in question tries to make college wage premium go away using questionable control variables. Sure, some skilled trades can make more money than liberal arts grads. But very few people want to be plumbers or work on an oil rig and for good reason.

Reply on news.ycombinator.com