There’s an op-ed in the Times today about fixing the nation’s higher-education system. It’s significantly less whacky than another recent op-ed on a similar problem, perhaps because it’s more focused on producing qualified applicants and encouraging better rankings than with fixing the professoriate. So far so good.. but here’s where I get nervous:
The benefits of an extra year of schooling are beyond question: high school graduates can earn more than dropouts, have better health, more stable lives and a longer life expectancy. College graduates do even better.
Just because high school graduates do better than dropouts, and college grads do even better, that doesn’t mean that if everyone had more education everyone would earn more. For example, jobs that used to require no degree or only bachelor’s degrees (librarians, say, or certain administrative jobs) now require specialized master’s degrees. There are serious issues with credential creep, especially with the sorts of credentials that aren’t associated directly with acquiring skills. So the implications derived from comparing individuals don’t scale. A high school graduate earns more in part because a high school dropout earns less. Maybe there are some gains to overall productivity, but it seems like it’d be very difficult to tease these out. And this logic (people with more X have higher Y so we really need to make sure everyone gets more X) seems sound because the causality is relatively clear on an individual level – more education does lead to higher wages. And yet, the macro-level policy implication doesn’t necessarily hold up (and certainly not in a straightforward way). Am I missing something? If not, what’s a good name for this problem?
A (Budding) Sociologist's Commonplace Book 
Thomas
/ June 9, 2009The schooling system in Ontario isn’t that much different, and not that long ago, grade 13 was eliminated in favour of a high-school system that ended with 12th grade.
It might be different to examine graduates’ economic success in two successive Canadian education regimes, holding constant changes in the political economy of ontario. However, it would certainly me more empirical than the speculation in the above article.
As for the name of your problem? I think David Hume already named it!
Dan Hirschman
/ June 9, 2009Thomas, I was looking for something a little more specific than “the problem of induction”. The problem here isn’t induction per se – if an individual student does better through some intervention and gets a higher level of education, that student will likely earn more, etc. It’s more like a fallacy of composition, but not quite.
And yes, it’ll be very interesting to see what happens a few years down the road when there is good time series data on the Canadian experiment.
Thomas
/ June 12, 2009Perhaps I was quick to jump the gun on that one. My appologies to Dr. Hume.
Without any claim to quantitative mastery: my intention was simply to suggest that taking existing data on education, desvising a general law from it, and then applying that law in a wildly differing hypothetical scenario is hardly the rigorous analysis that it is taken to be in the opinion piece–which I think was the whole point of your post!
This, however, may be the problem encountered in using the Ontario example.
Elizabeth
/ June 9, 2009I think “SUTVA violation” is what you’re looking for, but it’s such an awful name you might want to keep on looking.
By the way, Stanley Lieberson talks about this same problem (inferring from the way that education is used to sort people in this society, with its educational distribution, that a society with equalized education would have equalized outcomes) in his book Making It Count. It’s such a clear example of the fallacy that I, too, was struck by seeing it in yesterday’s op-ed.
Dan Hirschman
/ June 9, 2009I really enjoyed the sections of Making It Count I read for a methods/phil of social science class I read two years ago. I’ll have to revisit it!
But yeah, SUTVA doesn’t really roll off the tongue…
Elizabeth
/ June 9, 2009Nor is its meaning the least bit transparent from the acronym. Communicative failure all around.
Carl
/ June 20, 2009Ah, so. You are wise to the ways of quant fallacy, Bodhisutva.
Seriously, you’ve put your finger on a delusion that is so widespread, unquestioned and fervently held that it is best understood as a religious belief.