Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Recommendations for Title ~ Allocations for the 1999-2000 School Year The Department of Education, following the recommendation of the Panel on Estimates of Poverty for Small Geographic Areas (National Research Council, 1998), allocated Title I funds to counties for the 1998-1999 school year by using the estimates from the Census Bureau's revised county model of the numbers and proportions of poor school-age children for 1993. ("County model" is used in the broad sense to include the entire estimation procedure, which comprises a county regression model, a separate state regression model, and county population esti- mates.) The Census Bureau has now used its model to produce county estimates of poor school-age children for 1995. In addition, it has produced school district estimates of poor school-age children for 1995 by applying within-county school district shares of poor school-age children from the 1990 census to the 1995 county estimates. The Census Bureau has also produced school district estimates of total population and total school-age children for 1996 by using a similar synthetic shares method, to accompany the updated poverty estimates. The 1994 Title I legislation mandates that, beginning with the 1999-2000 school year, the Department of Education use the Census Bureau's updated school district estimates to make Title I allocations directly to school districts unless the Secretaries of Education and Commerce determine that "some or all of the data" are "inappropriate or unreliable" for this purpose on the basis of the panel's study. A direct allocation procedure would replace the current two-stage proce- dure, in which the department allocates Title I funds to counties and the states then distribute the county funds to school districts. Under direct allocation, the states would have the option of aggregating the school district Title I allotments for districts with fewer than 20,000 people and redistributing the aggregate 75
76 SMAL L-ARE4 ESTIMATES OF SCHOOL-AGE CHILDREN IN POVERTY amounts to districts on the basis of another method that is approved for this purpose by the Department of Education (e.g., by using counts of participants in the National School Lunch Program). ASSESSMENT OF 1995 COUNTY ESTIMATES The model that was used by the Census Bureau to produce county estimates of school-age children in 1996 who were in poor families in 1995 is essentially the same model that was used to produce the revised county estimates of poor school-age children for 1993. On the basis of the internal and external evalua- tions that were conducted of alternative 1993 county models, the panel supported the use of the revised 1993 county estimates for Title I allocations for the 1998- 1999 school year. Additional internal and external evaluations were conducted of the 1995 county model that focused on the behavior of the county model (and the state model) when estimated for several time periods. Both sets of evaluations for 1993 and 1995 identified areas for further research and development, including several areas of work that should be completed well before the next round of estimates is produced in fall 2000 (see Chapter 5~. Overall, however, the 1995 evaluations confirmed the results of the 1993 evaluations, which showed that the county model is performing well. There remains, however, the question of whether the 1995 school district estimates, produced by applying 1990 census within-county district shares of poor school-age children to the 1995 county estimates, are appropriate for direct Title I allocations. The panel summarizes the pros and cons of using the 1995 school district estimates in the next section and then presents its recommendations for Title I allocations for the 1999-2000 school year. ASSESSMENT OF 1995 SCHOOL DISTRICT ESTIMATES It is not possible to develop very reliable estimates of poor school-age chil- dren for most school districts with the currently available data. The available data are inadequate not only because most school districts are small in population size, but also because the boundaries of many school districts do not coincide with the boundaries for counties or other governmental units, the boundaries can and often do change over time, and some school districts do not serve all elemen- tary and secondary grades. For 1990, the data on school-age poverty (from the ratio-adjusted long-form sample of the census) have a large degree of sampling variability for many districts because of their small population size (see Chapter 3~. Moreover, for years following the census, no data are currently available on a consistent basis for all districts with which to estimate changes in their poverty population (or total population) over time. Food stamp and federal tax return data are not
RECOMMENDATIONS FOR TITLE I ALLOCATIONS 77 currently available for districts. Data on participation in the National School Lunch Program for school districts that are provided annually by the states to the National Center for Education Statistics are not consistent or complete. Also, variations in program participation make it unclear how reliably school lunch data indicate differences in poverty among school districts across the nation. The limitations of available data for school districts constrained the Census Bureau to use a basic synthetic shares approach. In this procedure, 1990 census within-county school district shares of poor school-age children, reflecting 1995- 1996 district boundaries, were applied to the 1995 county estimates of poor school-age children developed from the county model. By definition, this proce- dure reflects only the changes over time in the numbers of poor school-age children for school districts that parallel the changes in the counties in which they are located. Evaluations of the Census Bureau's procedure over the 1980-1990 period, by applying 1980 census within-county school district shares of poor school-age children to the county model estimates for 1989, revealed large differences be- tween the synthetic estimates of poor school-age children for school districts and the ratio-adjusted estimates from the 1990 census. The reasons for the differ- ences include: (1) the sampling variability in the 1980 census estimates of school district shares; (2) within-county changes in school district shares of poor school- age children from 1980 to 1990, which the synthetic shares method cannot cap- ture; (3) errors in the county model (which are not a major factor); and (4) the sampling variability in the 1990 census comparison estimates. Even if the sam- pling variability in the 1990 census estimates were removed, the differences between the synthetic estimates and the census estimates would be large for many districts. However, for districts that are coterminous with counties, the estimates are reasonably precise because they come from the county model. In addition, for districts with 40,000 or more people, the estimates are not markedly worse than the county model estimates. Together, these two groups of districts comprised only 13 percent of all districts in 1990, but they contained 62 percent of poor school-age children. Use of Estimates for Allocations Although the level of error in the Census Bureau's 1995 estimates of poor school-age children is potentially high for many school districts, the panel none- theless concludes that the estimates are not inappropriate or unreliable to use for direct Title I allocations to districts as intended by the 1994 legislation. Central to the panel's conclusion is that it interprets "inappropriate or unreliable" in a relative sense in this context. Some set of estimates must be used to distribute Title I funds to school districts. The question is whether the Census Bureau's 1995 estimates are more appropriate and reliable than those produced by the
78 SMALL-AREA ESTIMATES OF SCHOOL-AGE CHILDREN IN POVERTY current procedures. Moreover, the clear congressional intent to move to direct allocations argues for the use of the Census Bureau's 1995 estimates even if they are not better than the estimates that the states are currently using to distribute the county allocations to school districts, so long as they are not appreciably worse than those estimates. Three main findings support the panel's conclusion that the Census Bureau's 1995 school district estimates are not "inappropriate or unreliable" in a relative sense for Title I allocations. First, many states currently use a method for distrib- uting county Title I funds to school districts that is similar to the Census Bureau's 1990 census-based synthetic shares method. The Census Bureau's estimates of within-county district shares for these states are likely to be better than the esti- mates that the states have been producing because the Census Bureau has addi- tional geographic information. The Census Bureau has access to the 1990 census block data for retabulating numbers of poor school-age children according to updated school district boundaries. Also, the Census Bureau's estimates of within-county district shares of poor school-age children are ratio adjusted, which somewhat reduces their sampling variability. Second, limited evaluation of school lunch data for one state suggests that the within-county school district shares of poor school-age children produced from such data are not appreciably better than the 1990 census shares, even though the school lunch data used in the evaluation pertained to the same year as the standard of comparison. (Further evaluation of school lunch data for other states would be desirable.) Third, direct allocation using the Census Bureau's 1995 estimates addresses the inequities that result with the current two-stage allocation process for concen- tration grants whereby some eligible school districts do not receive these grants because they are in counties that are not eligible. According to 1990 census data, about 30 percent of school districts, containing about 60 percent of poor school- age children, are eligible for concentration grants under the current two-stage allocation process. An additional 9 percent of school districts, containing about 14 percent of poor school-age children, would be eligible for concentration grants under direct allocations. The Census Bureau's synthetic shares procedure, based on 1980 census school district shares applied to 1989 county model estimates, identified similar percentages of school districts and poor school-age children that would be eligible for concentration grants under direct allocations but would not be eligible under the two-stage process. Only about half of the school dis- tricts that were classified by one source the 1990 census or the synthetic esti- mates as being eligible for concentration grants under direct allocations but not under the two-stage process were so classified by the other source. However, the degree of agreement was much higher when expressed in terms of numbers of poor school-age children (see Tables 3-8 and 3-9~. On balance, these results indicate that direct allocations with the Census Bureau's 1995 school district
RECOMMENDATIONS FOR TITLE I ALLOCATIONS estimates could be expected to help redress the lion process for concentration grants. 79 lil inequities of the two-stage alloca Reallocations for Small School Districts If the Census Bureau's 1995 school district estimates of poor school-age children are used for direct allocations to districts, the 1994 Title I legislation allows states to aggregate the Department of Education's allocations for districts with fewer than 20,000 people and redistribute the aggregate amounts by using some other method that is approved by the department. The department has determined that states can use an alternative approved method for redistributing the department's allocations for districts under 20,000 population both for basic grants and concentration grants, including using the alternative method to rede- termine which districts are eligible for concentration grants. The 1990 census estimated that over 80 percent of the nation's school districts had less than 20,000 population, although these districts included only 27 percent of total school-age children. The panel believes that it is important to take account of the 1995 county estimates from the Census Bureau's county model in any reallocation that states choose to do because these estimates are the only updated small-area estimates of school-age children in poverty that have been shown to be reasonably accurate on the basis of a thorough evaluation. Thus, reallocations for school districts under 20,000 population will likely be more accurate to the extent that they reflect the updated county estimates. The Department of Education can ensure that any reallocations reflect (approximately) the county estimates by requiring that, to the extent possible, plans aggregate the grant amounts for districts under 20,000 population into county subtotals and reallocate the county subtotals, rather than reallocating the total for districts with under 20,000 population without regard to counties. The panel believes aggregation to county subtotals should be per- formed separately for basic and concentration grants. RECOMMENDATIONS The 1994 Title I legislation expressed the congressional intent to move to a system of direct allocations of Title I funds to school districts provided that the Census Bureau's estimates are adequate for this purpose. The panel concludes that the Bureau's 1995 school district estimates of poor school-age children are not inappropriate or unreliable for this purpose in a relative sense. While subject to large errors for many districts, they are at least as good as, if not better than, the estimates that are currently being used for many district allocations. Moreover, direct allocation with the Census Bureau's estimates partly addresses known inequities in the two-stage allocation process for concentration grants.
80 SMAL L-ARE4 ESTIMATES OF SCHOOL-AGE CHILDREN IN POVERTY (1) The panel recommends to the Secretaries of Education and Com- merce that the Census Bureau's 1995 school district estimates of poor school-age children be used to make direct Title I allocations to school districts for the 1999-2000 school year. The panel believes that it is important to reflect the Census Bureau's updated estimates of poor school-age children for counties in the direct allocations to school districts. The county estimates are the only postcensal small-area esti- mates of poverty that have been thoroughly evaluated. For states that accept the department's direct allocations for all districts, the county estimates will be re- flected (approximately) in the allocations because of the use of the county esti- mates in the synthetic shares estimation procedure for districts. For states that choose to reallocate the amounts for school districts with less than 20,000 popu- lation, the Department of Education should require, in general, that their plans reallocate the amounts for school districts on a county-by-county basis. (2) The panel recommends that any state plan approved by the Depart- ment of Education for redistributing the sum of the department's allo- cations for school districts with under 20,000 population maintain the county total amounts for such districts to the extent possible. The panel recognizes that developing county aggregate allocations from school district allocations is not always straightforward. In particular, states must determine the individual county components for school districts that cross county lines. However, states already must determine the individual county components for such districts under the current two-stage allocation process. Moreover, the Department of Education currently allows several states those in which school district boundaries bear little relationship to county boundaries to ignore the county allocations in distributing the total allocation for the state among school districts, and the department could grant similar exceptions for state plans to reallocate amounts for districts under 20,000 population. The panel also recognizes that the development of school district estimates of poor school-age children is a complex process for the Census Bureau and that the use of these estimates for direct allocations imposes burdens on the Depart- ment of Education to obtain the additional data that are needed to implement the formulas (e.g., counts of the other categories of formula-eligible children and the dollar amounts of Title I allocations that school districts received in the previous year). Because direct allocation of Title I funds to school districts is a new proce- dure, it will be important to assess how it is implemented and its effects in comparison with the current two-stage allocation procedure. The Department of Education should plan to conduct a thorough study in this regard, including such aspects as the operation of hold-harmless provisions and the differences in the
RECOMMENDATIONS FOR TITLE I ALLOCATIONS 81 number of districts and poor school-age children that are eligible for concentra- tion grants. LOOKING TO THE FUTURE Marked improvement in school district estimates of poor school-age children will require a substantial research and development effort. However, modest improvements in the estimates can likely be made in the near term with additional work to modify and evaluate the current 1990 census-based synthetic shares method. Specifically, a priority for the near term should be work to further reduce the sampling variability in the estimates of the 1989 shares of poor school- age children within counties that are derived from the census long-form sample. One approach is to use additional short-form data in the estimation of poor school-age children. Another approach is to smooth the 1990 census school district estimates with the 1990 census county estimates (see Chapter 5~. Looking to the longer term, it will be necessary to obtain relevant adminis- trative records data at the school district level if there are to be significant im- provements in the school district estimates of poor school-age children. School district-level administrative data (e.g., federal tax returns with addresses geocoded to districts) could provide the basis for model-based estimates of poor school-age children for school districts that more fully reflect changes in the extent of pov- erty among districts over time. In addition, data from the 2000 census and the planned American Community Survey have the potential to markedly improve the estimates for school districts. Research and development in these areas should begin as soon as possible. SPECIAL CASE: PUERTO RICO The Title I allocations include Puerto Rico, which has been treated as a county equivalent under the two-stage allocation process and will be treated as a single school district coterminous with a county for direct allocations. While the commonwealth's 1990 decennial census provides estimates for 1989, no esti- mates of Puerto Rican school-age children in poverty can be made for 1995 from the Census Bureau's county model because the appropriate federal tax and food stamp data are not available for Puerto Rico. The Census Bureau computed 1995 estimates for Puerto Rico from data collected in the 1996 Puerto Rican Family Income Survey that was conducted in the commonwealth in February-March 1997. Several adjustments had to be made to produce the estimates of school-age children in poverty in 1995. The Census Bureau previously used a similar approach to compute 1993 estimates of poor school-age children with data from an earlier round of the Puerto Rico income survey. The panel concluded in its first interim report that the Bureau's approach for producing updated estimates of poor school-age chil
82 SMALL-AREA ESTIMATES OF SCHOOL-AGE CHILDREN IN POVERTY dren in Puerto Rico seemed a reasonable one given the data available, although there was limited information about the quality of the data (see National Research Council, 1997:App. F). Additional information was obtained from Puerto Rico about the quality of the income survey that, in general, supported the use of the survey data to develop 1995 estimates of the number of poor school-age children for Puerto Rico. Consequently, the panel recommends that the 1995 estimates for Puerto Rico be used in the direct Title I allocations for the upcoming 1999-2000 school year. The Puerto Rico Family Income Survey is expected to be conducted at reg- ular intervals in the future. It will presumably be the basis of updated estimates of poor school-age children in Puerto Rico for 1997 and later years. Through cooperative work with Puerto Rico, the Census Bureau should continue its evalu- ations of the quality of the estimates and their comparability with the model- based estimates for U.S. counties to determine if there are ways in which the data and estimation procedures for Puerto Rico can be improved for use in Title I allocations.
