The denominator comes before the percentage
A program reporting 80% completion needs to identify who entered the calculation and how long they were followed. The figure could describe one entering cohort, several combined cohorts, or a subset of students. It could refer to completion within four years, eight years, or another period.
The Council of Graduate Schools’ work on doctoral completion and attrition treats these as longitudinal questions. A count of degrees awarded in one year is not the same as a completion rate for the people who entered together. The numerator and denominator need to belong to a coherent comparison.
A fictional cohort of twenty students
Imagine twenty students entering a program. Six years later, twelve have completed the degree, three have left without it, and five remain enrolled. The six-year completion proportion is 60%. The proportion known to have left without the degree is 15%. The remaining 25% are not yet in either final category.
| Status after six years | Students | Share of original cohort |
|---|---|---|
| Completed | 12 | 60% |
| Left without the degree | 3 | 15% |
| Still enrolled | 5 | 25% |
It would be inaccurate to call all eight noncompleters dropouts. Five are still studying. It would also be premature to assume those five will all finish. The observation period leaves their eventual outcomes unresolved.
Recent cohorts need time to mature
A cohort admitted two years ago will usually have had less opportunity to complete a long degree than one admitted eight years ago. Comparing their raw completion percentages can therefore confuse elapsed time with program performance.
A report can address this by using a common follow-up period or methods appropriate to incomplete follow-up. Whatever the approach, the reader needs to know how students still enrolled are handled. A percentage without that information can look more definitive than the data allow.
Changes in program structure also complicate comparisons across cohorts. A current curriculum, funding arrangement, or supervision process may differ from what an older cohort experienced. Historical outcomes remain relevant but do not automatically describe the present program unchanged.
Time to degree can exclude the people who left
A median completion time of five years may be calculated among graduates only. That describes the duration for those who completed, not the probability that an entering student will graduate in five years. A program could have a relatively short median among graduates and substantial attrition.
The Survey of Earned Doctorates, for example, concerns recipients of research doctorates. Its scope is useful for understanding graduates, but it is not an entering-cohort census of everyone who began a doctoral program. A source’s population determines which question it can answer.
This distinction is easy to miss because “time to degree” sounds like a forecast. It is usually a summary of observed histories under a stated definition.
Leaving is an outcome, not a complete explanation
Attrition data do not explain why someone left. Reasons can involve changed interests, work opportunities, finances, health, family circumstances, program conditions, or combinations of factors. A rate alone cannot assign responsibility or describe the quality of every individual decision.
Similarly, completion is not a full measure of satisfaction or fit. Someone can finish under difficult circumstances, and someone can leave for a direction that suits them better. Program-level outcomes are useful for understanding patterns without reducing people’s experiences to a success-or-failure label.
Small programs can have unstable percentages
In a cohort of five, one person changes the percentage by twenty points. Combining several years can provide a more stable summary, but it can also mix different program conditions. Privacy may limit how much detail an institution publishes for small groups.
These constraints do not make the data useless. They explain why precise-looking rankings based on small differences deserve caution. The underlying counts and uncertainty matter, especially when the sample is small.
What an informative comparison would include
A useful report identifies entering years, degree type, follow-up duration, counts, completion, departure, and continued enrollment. It explains transfers, leaves, and any exclusions. Time-to-degree summaries are labeled by the population used to calculate them.
Those details allow outcomes to inform a program comparison without becoming an individual prediction. They also clarify what remains unknown: why the pattern occurred, whether current conditions differ, and how a particular person’s circumstances would interact with the program.
