Nice papers (!)
April 22, 2010 § Leave a comment
The Disgruntled Sociologist read two recent AJS papers last night and actually somewhat enjoyed them. This was after TDS had nearly given up hope upon – what with the Jason Beckfield article in a recent issue that had something like 11 figures, all of them maps of the world with incoherent scribbles all over them. Apparently this was a network diagram of the world system. Impossible to make heads or tails of (or to tell the difference between them). Amusing, though, that the article referred to the figures as if they were in color.
In any case, TDS should focus on the good news.
First, the winner for best title in a very, very, very long time is Rick Grannis, for “Six Degrees of Who Cares?” The whole six degrees of separation / small world stuff has gotten out of hand in recent years. Nay, it has been out of hand since Watts’s 1999 article. Mathematically elegant and all, even beautiful.
But the scope conditions for that stuff are pretty constraining; most importantly, that everyone be connected in one giant component (i.e., everyone is reachable). If in fact everyone is connected in this way, then small world effects are pretty cool. But of course the literature has taken off on the assumption that the world is connected in this way. Everyone forgets that 70% of Milgram’s letters did not reach their destination.
And what Grannis does a nice job of showing is that this is a pretty strong assumption; more importantly, he shows that very small differences or errors in the measurement of ties can have a huge effect on whether or not we can identify a giant component. The examples involving PhD flows between departments are particularly nice. A lone voice in the wilderness advocating conservatism in measurement.
Another nice paper, methodologically at least, is by Christine Schwartz, on how changes in marital patterns may have contributed to increasing income inequality in the United States. TDS liked this paper for the clean and sensible empirics. There’s not much in the way of theory. The story is that an increase in the tendency for married partners to be matched in income levels leads to growing income polarization. The nice methodological part is to point out that studying this through simple correlations is problematic, since there is no guarantee that the changes are the same in different parts of the income distribution. So a more flexible log-linear specification is used.
The remarkable fact, which TDS did not know, is that changes in the association of husbands and wives’ incomes account for 25-30% of the increase in inequality. That seems huge. Maybe even too huge, but TDS doesn’t have a story for why, and it seems in line with other work.
In any case, nice to know that sometimes good stuff does get published.