Investigating the World of Mathematics to Uncover how Language Proficiency Influences English Language Learners Performance on High Stakes Tests
Location
Exhibition Hall, DeVos Center
Description
PURPOSE: English Language Learners (ELLs) continue to be among the lowest-scoring groups on standardized tests and the gap between ELLs and their proficient English-speaking peers persists. The purpose of this study was to investigate the extent to which language proficiency influences performance on high stakes mathematics tests. SUBJECTS: The study included item performance data, by group, for seventh- and eighth-grade students (n = 24,693) who took the 2007 and/or the 2008 Michigan Education Assessment Program (MEAP), and a questionnaire completed by seventh- and eighth-grade participants (n=16) for triangulation. METHODS AND MATERIALS: Released items from the 2007 and 2008 mathematics MEAP were coded for data analysis and a questionnaire developed by the researcher was given to participants by teachers. ANALYSES: The item performance data set was analyzed using a logistic regression model to determine the interaction effects between ELLs and non-ELLs based on item type, language and strand. Cross tabulation, content, descriptive, and frequency analyses were conducted on questionnaire responses. RESULTS: Findings from the logistic regression reveal that the odds ratio in passing an item is affected by both whether that item was a computation or word problem, and also if a non-linguistic feature was present (p < .0001). The difference in passing rate was not affected by strand. Most of the 19 words identified as confusing or unfamiliar on the questionnaires were context-specific or technical mathematics language features. CONCLUSIONS: Results from this study have important implications for teacher preparation, classroom instruction, test design and score interpretation.
Investigating the World of Mathematics to Uncover how Language Proficiency Influences English Language Learners Performance on High Stakes Tests
Exhibition Hall, DeVos Center
PURPOSE: English Language Learners (ELLs) continue to be among the lowest-scoring groups on standardized tests and the gap between ELLs and their proficient English-speaking peers persists. The purpose of this study was to investigate the extent to which language proficiency influences performance on high stakes mathematics tests. SUBJECTS: The study included item performance data, by group, for seventh- and eighth-grade students (n = 24,693) who took the 2007 and/or the 2008 Michigan Education Assessment Program (MEAP), and a questionnaire completed by seventh- and eighth-grade participants (n=16) for triangulation. METHODS AND MATERIALS: Released items from the 2007 and 2008 mathematics MEAP were coded for data analysis and a questionnaire developed by the researcher was given to participants by teachers. ANALYSES: The item performance data set was analyzed using a logistic regression model to determine the interaction effects between ELLs and non-ELLs based on item type, language and strand. Cross tabulation, content, descriptive, and frequency analyses were conducted on questionnaire responses. RESULTS: Findings from the logistic regression reveal that the odds ratio in passing an item is affected by both whether that item was a computation or word problem, and also if a non-linguistic feature was present (p < .0001). The difference in passing rate was not affected by strand. Most of the 19 words identified as confusing or unfamiliar on the questionnaires were context-specific or technical mathematics language features. CONCLUSIONS: Results from this study have important implications for teacher preparation, classroom instruction, test design and score interpretation.