The Influence of Doctoral Program Designs
Expectations for the dissertation are extremely
high—it's treated less as a learning document
than the first draft of a book—one that should rev-
olutionize one's field, no less!—and for me this
has been somewhat paralyzing. It does seem some-
thing that could be remedied to some extent with
some sort of post-proposal workshop or mini-
course. I find it all too easy to lose track of the hori-
zon and purpose of this project.
—Student in English who entered in
1993 and left the program in 2004
THE ANALYSES we presented in the previous two chapters estimated the impact of the Graduate Education Initiative (GEI) on doctoral students' attrition probabilities, completion probabilities, and time-to-degree (TTD). Although we discussed the important influence of financial support methods on these outcomes in Chapter 5, all other aspects of doctoral education in the humanities were treated as a black box. In this chapter, we use the Graduate Education Survey (GES) to go inside that black box to understand how the characteristics of doctoral programs, including the expectations that doctoral students confronted, were influenced by the GEI and how these changes in turn affected doctoral students' attrition and completion probabilities.1 In the process, we demonstrate the general relationship between doctoral program characteristics and doctoral student outcomes, and we suggest that our findings may have important implications beyond the evaluation of the GEI.2
1 The material presented in this chapter draws heavily on Ronald G. Ehrenberg, George
H. Jakubson, Jeffrey A. Groen, Eric So, and Joseph Price, “Inside the Black Box of Doctoral
Education: What Program Characteristics Influence Doctoral Students' Attrition and Grad-
uation Probabilities?” Educational Evaluation and Policy Analysis 29 (June 2007): 134–50. We
are grateful to our co-authors for permitting us to include this chapter in the book. Read-
ers interested in the technical details of our analyses should refer to that paper.
2 The 44 treatment and 41 control departments whose data are included in our analy-
ses in this chapter are the ones listed in Table 4.1.