Who Is Like Whom? Reclassification and Performance Patterns for Different Groupings of English Learners

• In Event: Validity Issues in Assessing English Learners

Abstract

Approximately 10% of the US K-12 public school population consists of ELs, who are learning English in addition to academic content in areas like English language arts (ELA) and mathematics. The goal for ELs is to learn English as quickly as possible so they can be reclassified as former ELs, and participate in academic settings where English is used without special support.

ELs are a diverse group of students who may resemble or differ from one another in many ways, including their level of English language proficiency (ELP), age at initial EL identification, type of English instruction, and home language. Although there is general consensus that ELs should not be treated as a homogenous subgroup (see, e.g., Francis & Rivera, 2007; Wolf & Leon, 2009), it can be difficult to determine how students should be grouped most meaningfully for analysis, and to support appropriate decision-making.

The purpose of this study is to explore ELs’ trajectories towards reclassification as a function of various time variables and individual covariates. Working with a longitudinal cohort of ~28,000 ELs in grades 3 through 8 from one state, I explore the likelihood of reclassification over time as a function of (a) grade-level, and (b) number of years a student has been eligible for the state’s EL services. I use discrete-time survival analysis to develop baseline models showing the probability that a student will be reclassified as a function of each timing variable (e.g., reclassified in grade 3, 4, or 5, or reclassified after 1, 2, or 3 years of service), and explore interactions between these variables using post-hoc ANOVA on the probability estimates (e.g., does the probability of being reclassified in Grade 5 differ significantly if grade 5 is a student’s third year of EL service, versus his sixth). Each model is further explored for whether and how reclassification probability changes over time (e.g., probability increases constantly over time; it accelerates and decelerates in a cubic fashion, etc.). In addition, I estimate the effects of covariates such as type of EL instruction, and baseline year ELP on the probability of reclassification.

As a vulnerable population who often struggle to meet academic standards, ELs should be supported by sound, evidence-based policies designed to ensure fairness and maximize their chance for success. Information about which students are likely to be reclassified, when, and why, is critical to support effective policymaking, program design, and resource allocation. While numerous studies have evaluated reclassification rates and relationships (see, e.g. Cook, Linquanti, Chinen, & Jung, 2012; Slama, 2014; Thompson, 2012; Umansky & Reardon, 2014), few have systematically compared the outcomes or interpretations associated with different ways of grouping ELs or attempted to bring multiple timing factors together in one model. The current study will make both methodological and substantive contributions with its design and findings. All data for this study have been collected and are currently being analyzed and all analyses will be completed by November 2015.

Author

• Molly Marian Faulkner-Bond, University of Massachusetts Amherst