UC Irvine

Latinos comprise a significant portion of the total school enrollment in the U.S., and an increasing share of Latinos are pursuing higher education today. However, Latino students still face many educational barriers, and research has repeatedly demonstrated that Latino children lag their peers in terms of their academic achievement. In addition, reading skills are the foundation of student learning and provide students with the literacy skills necessary for accessing academic material in other subject areas. Thus, in this dissertation, I focus on expanding the evidence base on instructional approaches that target Latino students’ early reading outcomes. This two-study dissertation examines the effectiveness of two different instructional approaches that could be implemented in elementary school classrooms to raise the reading achievement of Latino students, many of whom also come from Spanish speaking homes. The first study is based on data collected from a large, urban school district in Southern California where nearly all students are Latino, and many come from low-income backgrounds. The second study utilizes a nationally representative sample of Latino students in the U.S. with over 70% of students within the sample coming from Spanish speaking homes. Employing a mixed-methodology framework, the first study investigated the impact and effectiveness of a district-implemented Biliterate Instructional Assistant (BIA) program in a predominantly Latino and low-income district. BIAs were paraprofessionals hired by the district for the purpose of providing instructional assistance to first-grade teachers within the lowest performing schools. A quasi-experimental approach was used to assess the impact of the BIA program on students’ reading outcomes. Specifically, I used a short comparative interrupted time series (CITS) design to compare the reading test scores of students within BIA schools (n=6) and those in comparison schools (n=5) that were selected to match these schools. Utilizing data across three years (i.e., 2016-2019), I analyzed the school-level trends of BIA and comparison schools prior to the implementation of the program during academic years 2016-17 and 2017-18, and the reading outcomes of both groups after the implementation of the program (AY 2018-19). This analytic approach provided causal estimates of the BIA program on the average school-level reading achievement of students within BIA schools. To complement the school-level results from the CITS analysis, I also assessed the impact of the program on student-level reading outcomes by comparing the reading test scores of students (n=865) within BIA and comparison schools during the implementation year while also controlling for their prior reading achievement and demographic characteristics (i.e., English Learner status, Free/reduced priced lunch status, parent education, and eligibility for Special Education). Findings from the CITS analysis revealed that the BIA program had a large, positive impact on BIA schools’ average reading performance, though this was a marginally significant finding (ES=.825, p<.10). This p-value is considered acceptable given the small number of school clusters and considering this was a pilot study. Additionally, findings from the student-level analysis revealed that the BIA program had a significantly positive impact on students’ reading outcomes (ES=.195, p<.05), even after controlling for students’ demographic characteristics, lending further support to the positive findings from the CITS analysis. In addition, these quantitative findings were supplemented with qualitative data from BIA classrooms obtained through interviews with first-grade teachers and the BIAs themselves, as well as observational data of their instructional content and delivery and time-log diaries outlining their daily instructional duties and activities. These data were collected to describe and highlight implementation features of the BIA program for the purpose of informing district-leaders of which classroom implementation features appeared most promising in improving students’ reading outcomes. Considering each school had discretion in how to implement the BIA program, the amount of time BIAs spent in each first-grade classroom and the primary type of instructional grouping BIAs provided (i.e., 1-1 or 2-1, small group, or large group) varied between BIA classrooms. Thus, these implementation features were calculated and coded for each classroom by triangulating across the interview, observational, and time-log diary data. Then, I used regression analysis to predict BIA classrooms’ (n=28) average spring scores during the implementation year while controlling for BIA time, instructional grouping type, the classroom’s fall average reading test score, and percent EL. The classroom-level findings revealed that the amount of time BIAs were in first-grade classrooms, measured in minutes, and the instructional grouping type significantly predicted classrooms’ average spring reading performance. For every additional hour BIAs spent in these classrooms, classrooms’ average reading scores increased by .459 of a standard deviation (p<.001). Additionally, those classrooms that primarily provided supplemental BIA instruction 1-1 or 2-1, as opposed to large group instruction (5 or more students), had significantly higher average classroom reading scores (b=.634, p<.05), though there were no significant differences between classrooms providing small group (3-4 students) and large group instruction. Qualitative data further revealed that the range of literacy components covered during instruction, student progress monitoring, communication between teachers and BIAs, and behavior management were implementation features that appeared most salient to classroom reading performance among BIA classrooms. The second study examined the relationship between bilingual or dual language (DL) instruction provided to a nationally representative sample of Latino students and their reading outcomes at fifth grade. Analyzing data from the Early Childhood Longitudinal Study, 2011 cohort, I investigated the typical patterns of treatment (i.e., bilingual or DL instruction) that students received from kindergarten to fifth grade. I also analyzed which student and school characteristics were most predictive of bilingual instruction at each year using logistic regression. Then, using full information maximum likelihood (FIML) to account for missing data and student fixed effects analysis, I assessed the impact and relationship between bilingual instruction and students’ fifth grade reading outcomes by using a series of six dummy variables representing whether a student received bilingual or DL instruction at each year (grades K-5), as well as a continuous variable representing the total number of years students received bilingual instruction from grades K-5. Findings revealed that only 20% of the total sample received any bilingual or DL instruction throughout elementary school with nearly 4% receiving this instruction every year. Another 7% of students in the sample received bilingual instruction beginning at kindergarten with 1-2% exiting these programs each subsequent year. Additionally, results from the logistic regression revealed that students’ linguistic background (i.e., English proficiency scores and Spanish language use in the home), as well as school characteristics (i.e., whether schools received Title I funds, percent Hispanic, and percent EL) were the most significant predictors of bilingual instruction. Finally, FIML results revealed that with each additional year of bilingual or DL instruction, students scored significantly higher on the reading test at the end of fifth grade (ES=.042, p<.01). These findings were supported by student fixed effects analysis which revealed that the total number of years students received bilingual or DL instruction positively predicted their reading scores and this was a marginally significant finding (ES=.04, p<.10). Post-hoc analyses revealed that bilingual instruction was particularly most effective when provided during the early grades (K-1st). Future research and implications are discussed.