|Title:||Science Learning Difficulties: Patterns and Predictions in a Nationally Representative Cohort|
|Principal Investigator:||Morgan, Paul||Awardee:||Pennsylvania State University|
|Program:||Science, Technology, Engineering, and Mathematics [Program Details]|
|Award Period:||2 years (7/1/15–6/30/17)||Award Amount:||$700,000|
Co-Principal Investigators: George Farkas (University of California, Irvine) and Marianne Hillemeier (Pennsylvania State University)
Purpose: The purpose of this project is to analyze nationally representative and longitudinal data to identify factors associated with or predictive of elementary and middle school students experiencing difficulties in learning science. Some groups of elementary and middle school students are far more likely to experience low science achievement, although the factors leading to this greater risk are poorly understood. Students with disabilities, low-income students, and English language learners often display very low levels of science achievement. This research project will focus on two types of student profiles: (1) students with repeated science learning difficulties (SLD) who persistently experience low science achievement and slow achievement growth over time; and (2) students with resolved SLD who initially experience low science achievement but then demonstrate average achievement growth over time. Relevant predictors will include socio-demographic variables (e.g., family socioeconomic status); student characteristics (e.g., behavioral self-regulation, prior reading and mathematics achievement); instructional, teacher, and school characteristics (e.g., class time spent on science instruction, teacher's educational background); and interactions between these variables.
Project Activities: Researchers will analyze the Early Childhood Longitudinal Study—Kindergarten Cohort (ECLS-K) dataset to identify factors associated with or predictive of science learning difficulties, examine inter-relations between these factors, and determine which factors may be most educationally relevant for addressing science learning difficulties in the United States.
Products: The products include an understanding of the relative stability of SLD over the elementary- and middle-school years and potential factors that may help prevent or reduce at-risk students in the United States from experiencing SLD. The team will disseminate findings through published articles in peer-reviewed journals and presentations at national conferences.
Setting: This study will analyze data from the Early Childhood Longitudinal Study—Kindergarten Cohort (ECLS-K), a nationally representative cohort of young school-aged children. Children were assessed in the fall and spring of kindergarten, and the spring of Grades 1, 3, 5, and 8.
Population: The ECLS-K includes data on students entering U.S. kindergarten classrooms in 1998–1999, including at-risk subpopulations (e.g., students with disabilities, students from low-income families). The sample includes 7,395 students who have more than one test score on the dependent variables of interest in this study.
Intervention: There is no intervention.
Research Design and Methods: This is a secondary analysis of the restricted version of the extant ECLS-K dataset.
Control Condition: Due to the nature of this study, there is no control condition.
Key Measures: The dependent variables of interest were measured using the General Knowledge Test and Science Test from the ECLS-K. Predictor variables of interest were measured using a combination of survey and questionnaire data, school records, and parent reports. Other measures included the ECLS-K Reading and Mathematics Tests, and a modified version of the Social Skills Rating System. All measures were created and/or collected through the U.S. Department of Education's National Center for Education Statistics, which maintains the ECLS-K.
Data Analytic Strategy: The project will use multilevel growth trajectory models and propensity score matching, combined with growth mixture modeling, to estimate latent classes of science achievement growth trajectories. The models will also adjust for the clustering of the original sample within kindergarten classes.