Title III, the law that governs federal funding and programs for dual language learners, provides relatively small amounts of money to states. As we explained in part one of our three-part blog series on the financing of dual language learners (DLLs), the formula provides 80 percent of federal dollars on the basis of each state’s share of dual language learners, and 20 percent on each state’s share of immigrant children.
Importantly, though, when lawmakers rewrote the bilingual education provision as Title III of No Child Left Behind in 2001, they didn’t specify which source of data that the Department of Education should use to calculate those numbers. Beyond the first several years of implementation, the statute only asks that the Secretary of Education should use the more accurate of two sources: the American Community Survey conducted by the U.S. Census Bureau, or the number of children for whom the state administers annual English proficiency assessments, as required elsewhere in the law.
Today, the Department uses the ACS data rather than state-reported figures–and has since 2005–apparently because the state data were initially incomplete. But while the distinctions may seem pretty insignificant, some say it makes a big difference in funding. According to a 2006 report from the Government Accountability Office (GAO) that simulated the Title III funding formula for fiscal years 2005 and 2006 in 12 states, some states would see dramatic increases in funding if state-reported data were used as compared with ACS data–and some would see declines in funding as high as 40 percent. [See graphic below.] Moreover, variations in the sample sizes of the ACS from year to year could mean big annual ebbs or surges.
Therein lies the rub. Members of Congress are notoriously bad at revising funding formulas, because it creates winners and losers. Editing federal formulas to improve equity is rarely worth the risk of angering constituents or losing votes to lawmakers.
A big part of the reason for that variation is because the state data are comprised of a straight count of students, while the ACS data require sampling of students. That’s especially problematic for small school districts and states with lower DLL populations, where it may be hard to find a fully representative sample. But there are plenty of other differences between the two methods, too, not all of which are a bad thing:
- ACS includes children from age 5 to 21, while the state decides its own range for school-aged children;
- Public and private school students are included in ACS data, while just public-school students qualify for the state counts;
- States conduct assessments of children to determine English language proficiency that cover literacy as well as speaking proficiency, while the ACS simply asks whether the person speaks another language at home, and how well the person speaks English;
- Because the ACS survey is conducted among adults, the child himself is not assessed for English-language proficiency in that method, while state counts are child-focused; and
- The ACS survey is uniformly used across states, while the state counts are determined on the basis of state or local assessments and definitions.
That’s a shame, because in this case, the changes could mean a significant difference in the services available to children under Title III. A report for the U.S. Department of Education in 2012 found that per-child funding through Title III totaled less than $120 in seven states, but topped $300 in four states. Particularly in schools with few DLL children, those numbers bear diminishing returns. In the states with the lowest per-pupil federal DLL funding, a school with 10 DLL students could bring in about $1,200–not enough to provide much in the way of dedicated staff, additional services, or new resources.
And another report found that the variations in state and ACS estimates were significant: For example, Nevada identified 10.9 percent of its public school students as dual language learners; but the ACS data showed a rate of just 6.9 percent. In New Mexico, the state found a rate triple that of the ACS: 18.8 percent, compared with 6.6 percent. Just one state, West Virginia, had a higher rate of DLL students in the ACS than in its state data–1.0 percent in the ACS, compared with 0.9 percent in the state data; but the range of variation was substantial from state to state.
And although the report didn’t directly address it, states’ DLL populations are not static. The size of each state’s DLL subgroup rises and falls with each year—and throughout the year—depending on their policies around classifying and reclassifying these students. So methods of assessing students for data collection purposes could have varying effects on states, depending on the frequency and timing of sampling. (For more on reclassification policies, check out our recent publication, Chaos for Dual Language Learners: An Examination of State Policies for Exiting Children from Language Services in the PreK-3rd Grades.)
Rational people could disagree on whether the American Community Survey or the state-provided counts are more accurate. States that define English-language proficiency more loosely might be at a distinct advantage for raking in funding from a system that required states to find and count their own DLL students. There are obvious (and concerning) incentives encouraging states to overcount, given that not all districts provide great services to their DLL students. However, it is apparent that certain types of districts are likely at a greater disadvantage because the Department of Education relies on the ACS data. In particular, undersampled states and districts are probably losing out on some funding, and some larger states and districts with more accurate samples included in the ACS may be benefitting as a result.
So how will Congress address these issues? Check back with EdCentral on Tuesday for Part III of our Title III series. Click here for Part I.
[Cross-posted at Ed Central]