Time to the doctorate: Multilevel discrete-time hazard analysis

Secondary data on 1,028 graduate students nested within 24 programs and admitted into either a Ph. D. or Ed. D. program between 1990 and 2006 at an American public university were used to illustrate the benefits of employing multilevel discrete-time hazard analysis in understanding the timing of doctorate completion in Education and the factors related to this timing. While no single factor was found that explains conclusively the timing of doctorate completion, this analytic technique, which takes into account the clustering of students within programs and includes information about students who do not graduate by the end of the observation period (censored cases), revealed that the median time-to-doctorate was 5.8 years, with the fifth and seventh years as periods students were most likely to complete the doctorate. A student's master's GPA at admission, the proportion of female students in the program, and the mean GRE quantitative score in the program were each positively associated with the odds of doctorate award whereas the size of the department housing the program had a negative association. Implications for research and practice are discussed.