A contextualized school effect study of mathematics achievement at New York City public high schools (Region #6) using higher order linear modeling

New York City public school system, with its diverse learning population, faces many challenges to ensure that schools maintain Adequate Yearly Progress in accordance with the No Child Left Behind Act of 2001 accountability provision. The primary objective of this study is to find if the variance in mathematic achievement is greater within schools or across schools in New York City Public School System, Region #6. The 2004--2005 cohort mathematics achievement data of 16 high schools were used in the study. It was originally intended to involve all 26 schools in the study but 10 of the schools are new schools with no 2004--2005 cohort data. From similar research studies done with national data using Hierarchical Linear Modeling (HLM) to do school effect studies, it is known that socioeconomic status (SES) is a significant predictor of math achievement and with SES controlled, public school students performed better than private school students. The results from similar research studies using national data also showed that the variance in mathematics achievement was greater within schools than across schools. The results of this study showed that the variance in mathematic achievement was greater within schools than across schools. At the school level, minority status was a significant predictor of schools' mathematics achievement. Student level SES was not a significant predictor of a student's mathematics achievement in this study because schools were unable to accurately report SES data from federal government aided free and reduced lunch application forms---many students did not submit these application forms to schools. SES in this study is a categorical variable. Students who did not submit their school lunch application form were categorized as high SES. Hierarchical Linear Modeling was used to analyze the data because schools have a nested structure. HLM produced HLM estimates and robust estimates of the models tested. Comparison of the HLM estimates with the robust estimates indicates the degree of misspecification of the HLM estimates. The results of this research study have useful information on school effect issues for policy makers in New York City Public school system.