This dissertation models K-12 enrollment within an urban school district using two grade progression ratio (gpr)-based and two housing choice methods. The housing choice methods provide, for the first time, a new spatio-demographic model for projecting school enrollments by grade for any flexibly defined set of individual catchment areas. All methods use the geocoded pattern of individual, address-matched, enrollments within the study district but are different in the way they model this data to estimate key parameters. The conventional method projects the intra-urban pattern of enrollment by assuming no change in grade progression ratios (gprs), which are themselves functions of enrollment change. The adaptive kernel ratio estimation (KRE) of local gprs successfully predicts local changes in gprs from three preceding two-year periods of gpr change. The two housing choice methods are based on different mixtures of a generalized linear and a periodic model, each of which use housing counts and characteristics. Results are clearly sensitive to these differences. Using the above predictions of gpr change, the adaptive KRE enrollment projections are 4.1% better than those made using the conventional model. The two housing choice models were 2.0% less accurate than the conventional model for the first three years of the projection but were 5.1% more accurate than this model for the fourth and fifth years of the projection. Limitations are discussed. These findings help close a major gap in the literature of small-area enrollment projections, shed new light on spatial dynamics collected at areas below the scale of the school district, and permit new kinds of investigations of urban/suburban school district demography.

The problem of this study was to make short-term projections of student semester credit hour enrollments, at each of two state universities of comparable size, based on traditional time series analysis. 1. The first purpose of the study was to identify the cyclical component of deseasonalized enrollment data. 2. The second purpose was to determine a cyclical economic indicator having a high correlation with the cyclical component of the enrollment data. The selected economic indicator was used in establishing explanatory equations for projecting enrollment. 3. The third purpose was to compare projected 1979- 1980 academic year enrollment figures obtained from explanatory equations for each institution with actual enrollment figures of each institution for that year. 4. The fourth purpose was to compare the explanatory equations developed for the two institutions and the projections of student semester credit hour enrollments they yielded. 5. The fifth purpose was to discuss enrollment projections for each institution and the uses of enrollment projections in planning.