Skip Navigation

Literacy

Grantees

- OR -

Investigator

- OR -

Goals

- OR -

FY Awards

- OR -

Using Growth Mixture Modeling to Identify Patterns of Early Reading Development and Teacher and Program Correlates for English Learners

Year: 2006
Name of Institution:
California State University, Los Angeles
Goal: Exploration
Principal Investigator:
Hafner, Anne
Award Amount: $88,179
Award Period: 1 year
Award Number: R305G060108

Description:

Purpose: In this project, the research team proposed to use data from the Early Childhood Longitudinal Study to examine the development of English reading proficiency among English language learners (ELs). In this project, the researchers planned to

  1. use growth mixture modeling to identify distinct reading growth patterns and trajectories in children from kindergarten to fifth grade in two different large-scale data sets
  2.  compare ELs' reading growth with the growth of native English student
  3. examine the associations that exist between membership in reading proficiency classes for ELs, and a variety of student SES and background characteristics, as well as teacher, instructional, class, and programmatic conditions

Project Activities: The researchers looked at the differences in growth curves in reading acquisition for different types of students, for example native English speakers, ELs who were fluent in English, and ELs who were not yet fluent. In addition, they wanted to determine which measures predict probability of student placement in reading proficiency classes.

Structured Abstract

THE FOLLOWING CONTENT DESCRIBES THE PROJECT AT THE TIME OF FUNDING

Setting: The sample is drawn from a national dataset, the Early Childhood Longitudinal Study, and from a large urban district in the Southwest.

Sample: The Early Childhood Longitudinal Study consists of a national probability sample of 22,000 children enrolled in kindergarten in the 1998–99 school year, who are followed through grade 5. In the first wave in 1998, 7 percent of the children selected were excluded from testing because of low levels of oral English proficiency. This included 19 percent of Asian children and 29 percent of Hispanic children. The second database is a district sample in the Southwest of about 10,000 urban students. Students in grades 2, 3, 4, and 5 in 1999 through 2002 will be used in the analysis. The district is made up of a majority of poor to middle income students. Less than 20 percent of the district students are White. A majority of the students are Spanish speakers.

Research Design and Methods: This study uses a longitudinal cross-lagged design. This is a correlational approach using growth mixture modeling, a variation of structural equation modeling. Research hypotheses of the study are formulated in terms of development in English reading proficiency that can be tested across time points using repeated measurements on students in the Early Childhood Longitudinal Study dataset and the district database. There is no treatment in this study. The researchers will look at the differences in growth curves in reading acquisition for different types of English language learners (for example, native English, fluent in English as a second language, and English language learners). In addition, they will ascertain which measures predict probability of student placement in reading proficiency classes. Measures such as teachers' ratings, socio-economic status level, gender, teacher characteristics, teaching strategies, and program variables can be used as predictors over time.

Key Measures: Key measures include standardized student achievement measures in reading, student language proficiency variables, student background variables, class variables, student program information, teacher background variables, and teacher classroom practices.

Data Analytic Strategy: A variety of statistical techniques will be used to evaluate the research questions, including descriptive analyses, growth mixture modeling, and classification analysis. Growth curve modeling will be used to ascertain how long it takes English language learners to become proficient at reading, to compare English language learners' trajectories with those of native English speakers, and to identify students who have a deficit or are lagging behind their peers in terms of their reading skills. Growth mixture modeling (GMM) differs from other approaches in that it does not assume a common growth trajectory for a population and is well suited for analyzing the broad variation in reading proficiency. GMM uses maximum likelihood estimates to identify groups by their developmental trajectories.

Products AND Publications

ERIC Citations: Find available citations in ERIC for this award here.