interim report (National Research Council, 2010), describing its planned approach for assessing the utility of ACS-based estimates for a special provision to expand access to free school meals. This, the panel’s final report, presents the panel’s findings and recommendations and concludes the second phase of the study. The bottom line is a glass half-empty and half-full story. The half-empty part is the panel’s conclusion that there is no immediately obtainable and usable set of estimates from the ACS that would enable USDA to specify a new special provision eliminating periodic base-year applications in all of the schools or entire districts that want to provide free meals to all students. The glass half-full part is that the panel developed a set of procedures through which an assessment by a school district could lead USDA to approve the use of ACS-based claiming percentages for some or all of the district’s schools.
The panel’s initial goal was to identify a universally applicable method for estimating ACS-based claiming percentages that could be used in any school district operating under a new special provision. The panel anticipated that one or more simple adjustments might be needed to account for consistent differences between ACS-based estimates and those from the traditional certification process of the school meals programs.
However, the panel’s comparison of ACS estimates with administrative data for all school districts and for all schools in five case study districts revealed that the ACS generally understates the percentages of students eligible for free meals and overstates the percentages eligible for reduced-price and full-price meals, particularly in schools and districts with high percentages of students eligible for free and reduced-price meals. More important, the systematic differences observed vary substantially across schools and districts. In Norfolk, Virginia, for example, the differences between ACS estimates and administrative data are small, whereas in Pajaro Valley, California, the differences are quite large.
Several major factors appear to contribute to such systematic differences in varying degrees in different places: underreporting of Supplemental Nutrition Assistance Program (formerly Food Stamp Program) participation by ACS respondents, use of annual income in the ACS to determine eligibility rather than monthly income as in the application process, limitations of using ACS data to count migrant and other students who do not live in traditional housing or do not live in the district all year, the presence of charter schools and other school choice opportunities that draw students from their neighborhood schools and the districts in which they reside, and errors in the certification process. The variation in differences between ACS and administrative estimates illustrated by Norfolk and Pajaro Valley demonstrates that a one-size-fits-all approach to correcting for the effects of these and other factors will not work.
Accordingly, the panel suggests a more tailored approach to using