graphic areas but that these estimates have not yet been sufficiently evaluated to serve as the sole basis for allocating funds under Title I.

The panel strongly endorses a model-based approach for county-level estimates of school-age children in poverty and commends the Census Bureau for working to develop a specific model for this purpose. In comparison with the continued use of 1990 census estimates, the use of the Census Bureau's model-based estimates has a clear advantage of employing more up-to-date information. In selecting a model, however, it is important to question the assumptions it uses to see that they are reasonable, to examine predictions to see that they contain no identifiable systematic errors, and to compare the selected model to alternative models. The Census Bureau's model has not yet been sufficiently evaluated in these respects. Many aspects of the model's performance need to be more extensively tested before the panel can recommend basing the Title I allocations solely on estimates from it.

Yet it is not desirable to continue to base Title I allocations solely on estimates of poverty in 1989 from the 1990 census data. Although those estimates have the advantage of being based on a much larger sample than the CPS data used in the model, they are missing the major changes in the distribution of poverty that occurred between 1989 and 1993. For the immediate purpose of Title I allocations, the panel has had to balance its concerns about using the Census Bureau's model-based estimates against its concerns about using estimates based on the 1990 census data. The panel concludes that a solution that takes advantage of the Census Bureau's work on model-based estimates but reduces the impact of possible limitations in those estimates is the most appropriate approach at this time. Therefore, the panel's recommendation uses estimates of the number of children aged 5–17 in families in poverty in each county from both the Census Bureau model and the 1990 census data.

The panel recommends to the Secretaries of Commerce and Education that funds under Title I for fiscal 1997 be allocated on the basis of estimates that are obtained by averaging two poverty rates and then applying the average rate to the 1994 population estimate. First, calculate an average poverty rate for a county as a simple average of (1) the rate based on the number of related school-age children in the 1990 census who were in poverty in 1989 and (2) the rate based on the Census Bureau's model-based estimates of the number of related school-age children in 1994 who were in poverty in 1993. Then, obtain the number of related school-age children in poverty by multiplying this average rate by the Census Bureau's estimate of the number of related school-age children (ages 5–17) in the county in 1994.

The panel's recommendation takes some account of the changing number and geographic distribution of children in poverty by using the model-based estimates, but it also uses the decennial census estimates to moderate the results from a model that has not yet been fully evaluated. This solution will smooth the transition to model-based estimates for subsequent allocations, after further re-