How do institutional factors shape PhD completion rates? An analysis of long-term changes in a European doctoral program

ABSTRACT

Our paper adds to a growing literature of doctoral training by studying factors that drive time-to-completion based on a new and unique data set from an international European graduate school. While previous research focused on individual factors, we inspect the role of institutional factors and the organization of PhD programs for PhD completion. Based on a theoretical model, we elaborate hypotheses on three sets of institutional factors for thesis submission rates. We use event history analysis to analyze register data on 30 entry cohorts of PhD students. Our findings demonstrate the relevance of the factors related to the institutional environment like the density of supervision and support. Furthermore, our results underscore the importance of program structuration through clear deadlines and sufficient length of funding. We conclude with a discussion on effective measures that can be taken by program designers and grant authorities aiming to improve completion rates in PhD schools.

KEYWORDS:

• PhD completion
• graduate schools
• institutional factors
• event history analysis
• record data

View correction statement:

Correction

Acknowledgements

We are thankful to the European University Institute for providing us with the data and Ken Hulley for precious guidance on the information contained in the dataset. We also thank Koen Geven who supported initial versions of the manuscript and contributed to data preparation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The European University Institute in Florence, Italy.

2 There has been some uncertainty in the grants on a yearly basis because of political shocks, changes in currency exchange, and inflation in Italy over the observation period. Consequently, the purchase power of grants may vary for students from the same country, depending on the year of enrolment.

3 We decided for including contextual variables in logarithmic functional form in our regression. Alternative functional specifications did not change any conclusion.

4 The discussed arguments mainly draw on selectivity arguments. Moreover, substantial ideas about the underlying individual process of successful thesis writing might suggest similar risk trajectories.