An investigation of English Language Learners' performance on regular content assessments: A study of Kansas ELLs

Due to the federal No Child Left Behind Act and accountability requirements, English language learners (ELLs) in Kansas are expected to make progress in both content area academic achievement and English language proficiency (ELP), as is measured using the state mandated testing for Title I and Title III. In Kansas this is done using the Kansas English Language Proficiency Assessment (KELPA) and the content assessments created by the Center for Educational Testing and Evaluation (CETE) for Math, Reading, and Science. Using validity theory as the framework, the intention of this study was to analyze the relationship between students' English language proficiency category, as measured by the KELPA, and their scores on the content assessments in Math, Reading, and Science. One goal of the research is to examine the predictive power of English language proficiency on content area assessment scores. Additional demographic variables were added to the analysis to measure their influence on content assessment scores. Multiple regressions and multiple ANOVA analysis were performed on state-wide data for all ELLs in the state of Kansas in 3rd -- 11th grade classified as ELLs, who took the KELPA and at least one content assessment in 2010. The results confirm that English language proficiency category positively corresponds to content area assessment score for all skills examined. This means that the lower the English proficiency, the lower the content assessment score. Like previous research, Reading had the strongest connection. Students with exceptionality codes (gifted or learning disabled), the English language proficiency category, and the Number of Years in the U.S., were all found to have significance, on average, at least 70% of the time Qualifying for Free and Reduced Lunch, Native Language, and Gender were found to be significant between 60% - 70% of the time overall. When two demographic variables were combined and analyzed as a pair, no pair combination was found to be significant more than 70% of the time overall. Total Proficiency Category and Exceptionality Code was the only pair combination that had an overall influence above 60%, with an average of 67% across the skills. Discussion is provided expressing the implications of these findings in regards to validity, as well as specific suggestions for teachers, schools, state education systems, and the federal education system. A final appeal is made to ensure that the assessments used with the ELL population accurately reflect that population's needs, and take into account the issues regarding validity of assessment scores from the ELL population.