Date Awarded


Document Type


Degree Name

Doctor of Education (Ed.D.)




Most institutions of higher education are interested in enrollment projections because they are closely related to institutional goals and missions and, are, therefore, essential to financial and program planning at every level. This study was undertaken to determine if relevant factors could be identified and used in a statistical forecasting model to project enrollments in a multidimensional urban community college within the accuracy limitations imposed by a state such as Virginia (who requires State institutions of higher education) to project their enrollment within (+OR-) 1 percent.;Two general types of statistical forecasting models, causal and extrapolation models were explored for use in forecasting fall and summer headcount, and total FTE enrollments within the prescribed accuracy limits. The relevant factors for possible inclusion in the models were identified from previous studies and a student flow model for the institution. The relevant factors used in the final models were selected on the basis of simple correlation coefficients, the mean square error, and average error as variables were added and removed from the models.;The optimum fall and summer headcount forecasts were produced by a combination time-series and multiple regression model. The independent variables used in fall and summer headcount forecasts were a seasonal factor, a time-trend factor, and national economic indicators. In the optimum total FTE forecast, produced by a multiple regression model, the relevant factors were full-time enrollment shifted forward three years and national economic indicator shifted forward three years. The basis for acceptance or rejection of the models was made in context with the fiscal system of the Commonwealth of Virginia for the distribution of public funds to the state colleges and universities. The fiscal system was established primarily to provide a basis for financial planning. Forecasting models were produced for 1 year for fall headcount enrollment and for 2 years for summer headcount and total FTE enrollment within (+OR-) 1 percent.;On the basis of this study certain general conclusions were reached: the large variations between national enrollment projections resulted from different assumptions; enrollment projections have been too generalized for institutions with diverse goals and objectives; present data bases are inadequate to produce accurate enrollment projections; and most projections are not sufficiently reliable for planning purposes. More specific conclusions reached were: state data bases are inadequate for multidimensional institutions; removing quarterly seasonal variations permits the identification of relevant factors; traditional projection models such as the cohort survival and Markov are not applicable in multidimensional institutions such as community colleges; models such as time-series and multiple regression can be developed to accurately project enrollments for less than two years; the current limits of accuracy for Virginia multidimensional institutions are unrealistic; verification of the accuracy of prediction models is valuable for evaluating forecasting models; and models for multidimensional institutions must be revised periodically because relevant factors are constantly in flux.



© The Author