�鶹������ý

As I’ve pointed out in recent posts, there are considerable limits to what education research can do on its own—because of and . Of course, that doesn’t mean that we should stop researching education, or that we should ignore existing research findings. It just means that we should: 1) be mindful of the limits of what research can do for politics and policy, and 2) even the best research usually has limited prescriptions for policy reforms.

That said, this second element isn’t just a cautionary limit—it’s also a call for more refined research. If we know, to use an example from , that access to a particular high-quality public pre-K program is particularly beneficial for dual language learners, the next step is to figure out the specific characteristics that make it work.

Cue from one of the regional labs in the Department of Education’s Institute of Education Sciences: “The Correlates of Academic Performance for English Language Learner Students in a New England District.” While it’s not directed specifically at early education, the study does offer analysis of how well various language support programs support ELLs’ academic performance.

The Northeast may not be traditionally thought of as a region with a large population of English Language Learners, but the report opens by noting that their numbers rose 7.6 percent from 2001 to 2010, even as the overall student population decreased 3.5 percent. As demographer Dowell Myers noted at , these trends suggest a host of problems for employers and public officials. Fewer children overall mean fewer workers (read: taxpayers) in the future. With an imminent surge in retirees on the horizon, Americans need to invest in all children as critical resources. This isn’t a choice between competing generational priorities, Myers said, “Seniors need this to happen.”

The report explores how “a large, urban Connecticut district is educating ELL students—and checks for correlation between particular language support models (such as English as a Second Language, transitional bilingual programs, and others) and student achievement. Because of the structure of the study, they made no causal claims, but found a variety of interesting correlations.

For instance, the various language support models in use had effects that varied by grade span. The researchers first explored how these various models might be linked to performance on Connecticut’s English proficiency assessment (the exam used to measure whether ELLs are ready to exit language services). Dual language bilingual education programs were “associated with higher English proficiency scores than the average” for ELLs in K–1st grade and 6th–8th grade. Meanwhile, ELLs in transitional bilingual education or English as a Second Language programs underperformed the average English proficiency scores to varying degrees at nearly all grade spans (K–8th grade). ELLs whose parents refused language support services performed above the average for English proficiency scores at all grade levels except 6th–8th grade.

They also considered other metrics—ELLs’ performance on Connecticut’s English Language Arts and Math assessments. Unfortunately, they found very little by way of statistically-significant patterns. ELL students’ English proficiency scores were the best predictor of their performance on these content assessments.

At one level, this is a relatively unsurprising finding: we’ve long known that ELLs’ English proficiency levels can impede their abilities to fully demonstrate what they know on content assessments. That core insight is why states (and the federal government) have developed testing accommodations for these students.

But when it comes to the question of determining which programs best support these students’ academic success, this research offers little guidance. Which is challenging, yes, but also a useful support for the research caution I’ve been recommending in previous posts. There are that do speak directly to that question, but we’re a long way from firm answers.

However, this shouldn’t be cause for paralysis. We should absolutely invest more heavily to support ELLs’ success, given that we currently allocate fewer educational resources to these students (and it’s far from clear that that original baseline was sufficient). That’s a matter of basic justice that doesn’t require research certainty.