Correlation is not causation

From the Associated Press, via msnbc.com: Which jobs have highest rate of depression?. The article ends:

Just working full-time would appear to be beneficial in preventing depression. The overall rate of depression for full-time workers, 7 percent, compares with the 12.7 percent rate registered by those who are unemployed.

Sure, having something to do with yourself forty hours a week probably helps ward off depression; although I don't believe I've ever suffered from anything that could be diagnosed as depression, I do feel noticeably sadder when I don't have some sort of full-time occupation. But I have a sense that the causation goes the other way here; being depressed causes people to be unemployed. Those people I have known who have suffered from clinical depression have days where they just can't get themselves to go to work; employers don't look kindly upon that sort of thing.

From the people at Strayer University, who sent me mail: in big letters on the front of the envelope, they inform me that

60.72%: The difference in salary betweeen someone with a bachelor's degree and someone with a high school diploma only.

They cite the "Bureau of Labor and Statistics" [sic]; their name is actually the Bureau of Labor Statistics. Again, correlation is not causation. There have been studies showing that people who were able to get into elite universities -- but chose not to attend them -- do just as well in the work force as people who got into those universities and did attend them. Could the same sort of thing be at work between people who have been to college and people who haven't been to college but could have gotten in? I don't know. Of course, it's much easier for a potential employer to see "this person went to college" than to see "this person didn't go to college but is still just as employable as those who did"; a large part of the value of an education, especially in less technical fields where there isn't a huge amount of subject-specific material that will actually be useful on the job, is precisely this signaling effect. I feel like Paul Graham has written about this but I can't find exactly where.

(In the interests of full disclosure: they didn't send me mail. They sent the "Working Professional" who lives at my address mail.)