A Propensity matched comparison of early literacy outcomes for elementary-age ELs in different language programs

• In Event: Do dual language experiences benefit the academic and literacy outcomes of Spanish speaking ELs?

Abstract

In the United States, English learners (ELs) are exposed to a wide variety of programs, including dual language immersion (DLI), sheltered instruction, immersion, heritage language support, and English as a Second Language (ESL). Evidence that these programs support the achievement of ELs, and specifically Spanish-speaking ELs, is mixed at best (National Academies of Science, Engineering, and Medicine, 2017). This study leverages Utah state education data to examine the differences in early literacy among the three dominant language instruction formats employed with Spanish-speaking ELs available across the state. Utah is one of the first states to systematically offer DLI instruction, as well as an array of other instructional formats to support elementary-age ELs. In this study we employ quasi-experimental methods [i.e., propensity score matching (PSM)], to examine the following research question: among Spanish-speaking Latinx children in the state of Utah, what is the association between English language program participation in first and second grade (i.e., DLI, sheltered instruction, ESL) and third grade basic early literacy?

This study relies on educational data collected and maintained by the Utah State Board of Education. Students (N = 4574) were included if they started first grade in 2014-2015, were identified as “Hispanic” and “Spanish-speaking,” and were placed in of the three program types for English learners: DLI, sheltered instruction, ESL (see Table 1 for sample descriptors). DLI refers to programs that include both Spanish and English instruction, including two-way bilingual and dual-immersion programs, with the goal of maximizing bilingualism; sheltered instruction refers to programs in which EL students receive modified content instruction in English in their general classroom with the goal of maximizing English proficiency; and ESL refers to pull-out programs in which children are removed from their general classroom designed for targeted English-only instruction.

PSM, based on child and family factors (See Table 1) and only ELs that received the same program in first and second grade (i.e., those with two full years of program exposure), was conducted to create three matched samples: DLI and ESL (n = 380); DLI and Sheltered (n = 448); Sheltered and ESL (n = 550). Separate regression models tested for difference in Dibels scores at the start of third grade for each matched sample. For each matched sample, we ran three regression models: no covariates, doubly robust plus first grade WIDA, and doubly robust, first grade WIDA, and 3rd grade income and special education status (see Table 2 for more details on each of these three models).

Although there were no statistically significant differences between program types (Table 2 for results), it is also important to consider the effect sizes yielded from the analyses. For example, compared to students in ESL, students in DLI performed .10 to .11 standard deviations better. Although these differences are small, they are still worth noting. Another interesting pattern worth noting is the decrease in group differences when 3rd grade special education status is included in the model. We discuss the policy implications of the findings for EL education in Utah and beyond.

Authors

• Natalia Palacios, University of Virginia - Charlottesville
  Presenting Author

• Natalie Lynn Bohlmann, Interactions Matter

• Bethany A. Bell, University of Virginia

• Yitong Yue, University of Virginia