Small area student enrollment projections based on a modifiable spatial filter

Abstract This paper describes a new method for making short-term (4–8 yr) student enrollment projections by grade and ethnic group for small geographic areas. The method links an information system that contains student characteristics and home addresses with a commonly available digital geographic database (TIGER) to create a geographic accounting table of student residences by census blocks. Progression rates of students from one grade to the next are estimated for each small area by aggregating student residences, by grade, for a 3 yr period over a larger area (the modifiable spatial filter) centered on the small area. The size of the filter area depends on the geographical distribution of students, a user-specified student threshold value, and a maximum distance constraint. The progression rates are applied to the student population of each census block in a grade-cohort component projection model. Projections of the number of students enrolled in each grade for any defined geographical area are made by aggregating the projections for the census blocks in the area. We show that results using this method are consistent with those from the same model applied to school attendance area enrollment data, whereas results using the small-area data without the filter are not. This result supports our conclusion that the method is reliable for making grade-specific enrollment projections, for which past enrollment data do not exist, such as those for new schools or for revised attendance areas. We describe the techniques used to link the student information system to the digital map, compute the spatial filter statistics efficiently, and make projections for new school attendance areas. We have used the modifiable spatial filter method for projecting student populations and modifying attendance area boundaries in two Iowa school districts.