Several years ago the American Economic Review published an article in which the authors tried to verify the statistical output of ten articles in a previous issue. About five authors simply refused to cooperate, despite AER policies that they must, and only one enthusiastically cooperated and had his results replicated. From this I became convinced that fraud in empirical economic research may be very widespread. I suspect a journalist with enough free time could make quite a splash by looking into it.
The Economics of the Jihad II - The Demand for Jihad.