Students' 2018 PISA Reading Self-Concept: Identifying Predictors and Examining Model Generalizability for English Learners

• In Event: Novel Approaches to Analyzing Educational and Psychological Data

Abstract

Self-concept has a vital role in all learning settings, especially in language learning context. While a significant amount of research has been conducted on reading self-concept by focusing on English-only (EO), English-learners (ELs) have been relatively left out. We applied machine learning method to identify proximal and distal predictors of students’ PISA 2018 reading self-concept. Among 1482 variables (943, 192, and 347 at the student-, school-, and teacher-levels), 20, 12, and 3 variables were selected using Elastic Net model as the most salient predictors of self-concept. In the next steps—reported in the full paper—we will first estimate a hierarchical linear model on the entire sample and then test its generalizability to EL and EO populations.

Authors

• Onur Ramazan, Chinese University of Hong Kong

• Shenghai Dai, Washington State University - Pullman

• Robert William Danielson, Washington State University - Spokane

• Bruce W Austin, Washington State University - Pullman

• Tao Hao, Washington State University - Pullman

• Yuliya Ardasheva, Washington State University - Tri-Cities