(2) the heterogeneity of need across schools within the school district, and (3) school district size (as measured by enrollment).

Need: For a school district with very high need, the savings associated with eliminating the application and verification processes and the process of distinguishing free, reduced-, and full-price meals in school cafeteria lines is likely to equal or exceed the additional costs of providing free meals to all students who take such meals.¹ For such a school district, Provision 4 is likely to be attractive, assuming that the estimates of claiming percentages from the ACS and other sources satisfy other criteria, such as timeliness and accuracy. In contrast, a school district with very low need is not likely to be interested in Provision 4, regardless of the quality of the ACS-based estimates, because the savings in administrative costs are likely to fall far short of the added meal costs. “In between” school districts face less clear-cut decisions.

Heterogeneity of need: In addition to the aggregate level of need within a school district, the heterogeneity of need across schools could affect a district’s decision regarding Provision 4. A school district might have an “in between” level of need because it has some schools with high levels of need and other schools with low levels of need. Such a district might want to adopt Provision 4 in the first group of schools, but not the second. To assess the attractiveness of Provision 4 for this type of school district would require estimates for groups of schools within the district. In contrast, a district-wide estimate would be adequate to assess the attractiveness of Provision 4 for a homogeneous school district.

Enrollment size: The size of a school district will substantially affect the reliability of the estimate(s) on which to evaluate the attractiveness of Provision 4. For a large school district, the methods we set forth in Chapter 5 might yield reliable estimates for schools or groups of schools. For a small school district, however, it might not be possible to derive estimates with acceptable reliability below the school district level, even using statistical modeling. If that is the case, the attractiveness of Provision 4 would have to be evaluated on the basis of an estimate for the entire school district, although that estimate might not be reliable.