Jim Saliba, University of Minnesota
The United States Department of Education collects administrative data from all public schools annually and has made this data available publicly from 1987 through the present as the Common Core of Data (CCD). The CCD constitutes a valuable resource for investigating patterns of stability and change across time and of variation and consistency across space in the basic structures of United States public schools, including racial composition, geography, school size, student-teacher ratio, and governance structure. Using the CCD, especially as a longitudinal dataset capable of tracking school-level change across time, comes with many challenges, including how to link schools, improve data quality, and enhance comparability across time and space. First, I describe the data available in the CCD, including spatial and temporal variations in availability. Second, I outline issues in linking schools across years and propose methods to optimize those linkages depending on specific definitions of when a school is the same from year to year and when it is different. For example, the CCD generally treats a school as the same if it occupies the same building, even if it has changed from an elementary school to a high school. For many research purposes, such a change would more appropriately be considered a closure of one school and an opening of a new school. Third, I review common data quality issues – such as extreme shifts in racial composition or changes in grade-level range that only last for a year – and propose methods of addressing them. Finally, I describe changes in the content of variables over time and propose methods to increase their comparability.
No extended abstract or paper available
Presented in Session 9. Data in Education: Assessment, Measurement, and Accounting