Outcomes and Consequences of One State's High-Stakes Retesting Policy for Language-Minoritized and Racialized Students

• In Event: Language-Minoritized Students and Assessment: Social Consequences of Test Use

Abstract

Objectives
This study examines retest outcomes of a large-scale, high-stakes reading exam for students who are learning English as an additional language (EAL) and who are racially minoritized. I predict outcomes for each retake based on students’ demographics and language proficiency, and I describe prior trajectories for students learning EAL who were unsuccessful.

Theoretical Framework
This study frames high-stakes language assessment with Flores and Rosa’s (2015, 2019) raciolinguistic ideologies, which critique language education and assessments as embodying a normative white, middle-class, heteroglossic gaze. I evaluate the consequences of assessment (Messick 1994, 1998), focusing on the pathologization and oppression of students who use minoritized language practices.

Data Sources & Methods
The reading and language cohort data (State of Texas Assessments of Academic Readiness; STAAR) are for students in Grades 3-8 (2012-2017; N=410,576). Using logistic regression, I analyze the likelihood of exam resistance and success contingent on language proficiency, ethnicity, and economic disadvantage (ED).

Results & Scholarly Significance
In March 2017 367,728 Grade 8 students took STAAR reading; 81,848 rewrote the exam in May due to not meeting expectations, and 34,635 again in June. Table 1 provides descriptives and regression results for the likelihood of multiple attempts. Students classified as EAL were almost 5 times as likely to sit the test multiple times as non-EAL; Black and Hispanic students were 1.21-2.65 times as likely than white students; ED students were about 2.5 times as likely as non-ED. Encouragingly, students in 1st-and 2nd year monitoring after exiting EAL were less likely than non-EAL to retake.
Table 2 provides statistics for passing the exams. Students learning EAL were one fifth as likely to pass as non-EAL students the first time, with odds well under 1 for subsequent administrations; on a positive note, monitored students had better likelihood to pass than non-EAL. Black and ED students were overall less likely than white and non-ED students to pass across all administrations. Most outcomes of students of other ethnicities were not significantly different from white students’.
The 11,586 EAL students who failed the third STAAR administration represent one quarter of Grade 8 EAL students. Table 3 indicates their STAAR reading trajectories. Of these, 17% had been in US schools for two years or less. However, 57% have been in US schools since Grade 1, 56% were EAL since Grade 3, and 96% were in a Grade 8 ESL course. In fact, 53% of all Grade 8 EAL students had been in US schools since Grade 1, and 66% were identified as EAL in Grade 3.
STAAR is not valid for EAL learners as there is no evidence of positive consequences within or across years. Students learning EAL and racialized students are most likely to be subjected to repeated testing and failure both within and across years, which can cause them to disengage. There are critical social consequences, as accountability-focused policies encourage teaching-to-the-test instead of sustained learning. A humanistic, child-centered, culturally and linguistically sustaining education and assessment system is necessary to ameliorate the wrongs these children have experienced.

Author

• Jeanne Sinclair, Memorial University of Newfoundland
  Presenting Author