As far as I can tell, the data from the new report from the Ph.D. Completion Project shows only four disciplines with 10-year doctoral completion rates below about 45%. One of these, computer science, is undoubtedly depressed by graduate students getting attractive job offers and leaving voluntarily (especially since the data span the mid-to-late 1990s).
The other three are Communications, Sociology,...
and Political Science.
(See Slide 7 in that power point presentation.)
Communications has the lowest rate of completion by year, through year 9. But the gap closes steadily over the years 7-9, and by year 10 Political Science seems to have the lowest rate of completion of the three.
This doesn't tell us what proportion of entering students complete sometime after year 10 (which is worrying in one way) and how many drop out or are failed (both worrying in another way). But notice that the completion rates are consistently lower than in econ, and econ grad students certainly have more attractive job opportunities that become available during their years of study. The completion rates are also lower than in the literary humanities, disciplines whose grad students face notoriously bad job prospects that presumably encourage many of them to drop out.
100% completion rates are implausible and undesirable. But completion rates this low suggest that departments and students are doing a bad job matching expectations at the beginning of programs. The departments can't judge who's likely to succeed in grad school, and students can't tell whether a program is a good one for them. Something seems badly wrong. Some discipline has to come in last in these sorts of measures, but we shouldn't take any solace from that. Either econ or English would have a good reason for a very low completion rate; I can't see that political science has any excuse to be behind both.
Prospective grad students, ask for hard data on time to completion and attrition rates!
See Chronicle coverage here, subscription required. See also Professor Rojas' latest round of advice to grad students-- finally reaching the all-important "The only good dissertation is a complete dissertation."