This is hugely important, and seems to have been glossed over. "attending college" != "graduating from college". College is, for better or worse, a binary signal. You either graduated or you didn't.

When you apply for an Engineering license, the Engineering board doesn't care that you completed 100% of your Engineering courses but never received a degree because you forgot to take History 101.

Because of this, it's even worse in that it's a sunk cost. Better to go ahead and drop out earlier rather than later, because not only will you not have a degree, you'll have wasted that many years for no measurable improvement in hiring potential.

This should be the top comment. This article seems to be playing games with the numbers. Why do they choose to focus on the fraction of the unemployed? This trend could just be explained by greater numbers of people attending some amount of college relative to the past. They should really show the fraction of college educated unemployed, vs fraction without college educated unemployed, but that wouldn't fit their narrative.

That's what caught my eye as well. I don't think a semester counts as "attended college". I'd like to see a more detailed breakdown.

Another demographic variable mentioned in the article is that the older end of the unemployed cohort had much lower rates of "attended college" and rises are expected as they age leave the workforce. At first my concern seeing that chart was "OMG why didn't anyone see this trend over the last 20 years and why hasn't it been part of the public discussion", but given at least that confound is not factored into the data display, my feeling is now that it's more of a meaningless fished graphic/statistic for effect/pageviews.