Michael Kremer has expressed justified concern that prospective graduate students might use the data compiled by Andy Carson to guide their choice in graduate programs. Even if the data were perfect, one should be cautious in using them; the data are backward-looking and reflect circumstances that might well have changed significantly by the time one shows up at the school of one's choice. Even when Andy Carson removes all the current (serious) problems with his data, placement data absent attrition rates are of limited use. Moreover, given the incentives involved we should not expect to get fully reliable data easily. Ideally, once Carson has worked out the obvious snafus with his data people can use his data and measures and compare them with old and more recent Gourmet reports and figure out ways to use them in light of each other given their individual needs.
But for now, nobody should let these data play a decisive role in their decision-making.