Date of Award


Level of Access Assigned by Author

Campus-Only Dissertation

Degree Name

Doctor of Philosophy (PhD)


Interdisciplinary Program


Theodore Coladarci

Second Committee Member

Phillip Pratt

Third Committee Member

Alan Kezis


Providing support for institutional planning is central to the function of institutional research. Necessary for the planning process are accurate enrollment projections. The purpose of the present study was to develop a short-term enrollment model simple enough to be understood by those who rely on it, yet sufficiently complex to serve varying planning needs: developing and monitoring enrollment goals, budgeting, resource allocation, and course planning.

One research question framed this study: Does increasing the complexity of a short-term projection model improve its accuracy and flexibility? To answer this question, four models representing varying levels of complexity were developed: an institution-level grade progression ratio model, an institution-level Markov model, a discipline-level grade progression ratio model, and a discipline-level Markov model. The models were compared on their accuracy for generating six types of projections for fall 2012, 2011, and 2010: overall enrollment; enrollment by class level, discipline, and course; credit hours; and tuition revenue. This study is unique in that it compared model performance for projecting multiple outcomes, therefore permitting the evaluation of the models on both accuracy and flexibility.

The results revealed that most promising among the four models were the more complex Markov models. However, the more complicated discipline-level model was not necessarily superior to the institution-level model. Adding discipline to the Markov model increased its accuracy in projecting discipline-level enrollment but did not offer an improvement for projecting enrollment at the institution- or class-levels. Discussed are the practical implications of these findings for institutional researchers. Those valuing simplicity may be willing to sacrifice a small amount of accuracy at the finer levels for the overall increase in simplicity and transparency provided by an institution-level model. Others may be willing to accept additional complexity for more accurate discipline-level projections. Although the Markov models were superior for projecting enrollment by discipline and course, the error rates for such projections were considerably higher than those associated with the class-level and institution-level projections. Necessary are studies that more thoroughly address the application of these models to discipline- and course-level projections.


Interdisciplinary in Institutional Research, Higher Education, and Public Policy