Dave Friedman blogs something that I've seen in similar words in many other places:
"All the arguments opposed to Levitt's that I have seen merely argue that Levitt is incorrect. They do not provide an alternative explanation for the reduction in crime. A successful critique of his argument would, I think, propose an alternative explanation."
That's not how science works. For example, if I take some scientific mystery that's not well understood, such as "What Came Before the Big Bang?" and I assert that the Big Bang was caused by commie pinkos fluoridating the water supply, well, the fact that you might not have an alternative hypothesis doesn't mean I win, or even that my idea should be treated respectfully.
The essence of the scientific method is the falsification of hypotheses. The falsifiers may not be as attractive figures as the hypothesizers, but their job is just as necessary.
Anyway, it's not as if we lack alternative hypotheses. Levitt himself lists a number of perfectly reasonable factors that he believes helped bring about the decline in crime -- the end of the crack wars, the vast increase in imprisonment, the addition of policemen, etc. I suspect those are correct and they may just have had an even larger influence than Levitt claims they did. Further, there are plenty of plausible theories he has never investigated. The crack wars permanently snuffed out the criminal careers of tens of thousands of the most dangerous young criminals by getting them murdered by other young criminals. I'd also point to the growth in popularity of smoking marijuana instead of crack among urban youths. Another likely factor is the increased moral traditionalism that emerged in the early 1990s: the abortion rate dropped sharply and the illegitimacy rate plateaued after a long, long increase.
Also, a lot of Levitt's supporters tend to assume that because some observers predicted an increase in crime in the late 1990s, then, because they were wrong, Levitt deserves credit as a seer. But, of course, Levitt didn't predict anything. He wrote in 1999, using crime data through 1997, by which time the big drop in crime was already in the books. It's easy to pick Giacomo to win the Kentucky Derby after the race is run. The funny thing, of course, is that the Levitt Effect doesn't even match up with what happened in the past when the history is looked at at the appropriate level of detail.
Now, lets make a prediction based on the putative Levitt Effect. The abortion rate per 1000 white women aged 15-44 dropped steadily from about 19 in 1991 to about 11 in 1999. (The black abortion rate dropped too, although not by not as large a percentage.) If Levitt believed in the Levitt Effect, then he should be predicting a sizable increase in the white juvenile (17 and under) violent crime rate over the decade or so beginning about 2007. Yet, I haven't heard Levitt raising the alarm about the coming generation of less culled white boys.
Of course, the Levitt Effect signally failed to predict the past, so it's effectiveness at predicting the future is doubtful as well.
Indeed, it's interesting that the decline in the abortion rate occurred during the same period as the decline in crime. Similarly, the illegitimacy rate among blacks started falling about 1995. In Levitt's "unwantedness" model, it's implied (but never stated) that abortion fights illegitimacy. Yet, when the abortion rate was going up in the 1970s, so was the illegitimacy rate. When the abortion rate among blacks was going down in the 1990s, the illegitimacy rate went down.
I suspect that what happened was that the catastrophe of the crack years inspired a rebirth of moral traditionalism that helped bring down, crime, abortion, and illegitimacy.