Projecting grade-specific enrollments for school catchment areas: An evaluation of two methods

Historical enrollment data for the catchment areas of individual schools are rarely useful for projecting future enrollments because their boundaries are frequently adjusted to balance growth and decline in the enrollments of individual schools. The modifiable spatial-filter method (MSF) was developed to meet this problem (Rushton, Armstrong, and Lolonis, 1995). In this article, a modifiable catchment area method (MCA) is developed, and the performance of the two methods is compared in one school district. Both methods use individual, address-matched, student records, but are different in the way they spatially aggregate these data to estimate key model parameters. Results are clearly sensitive to these differences. We apply both methods retroactively to the address-matched student records, from 1989 through 1992, of 16 elementary schools in one school district in Iowa, and we project the 1993–1996 enrollments for 17 school catchment areas in the district. We show that the average of the results of the two methods produce projections that are more accurate than either method alone. We also show that it is the nonstationarity of model parameters that leads to different levels of performance by the two methods in certain parts of the study region. © 1998 John Wiley & Sons, Inc.