Saturday, November 8, 2014
Brazos (Convention Center)
*Names in bold indicate Presenter
Broward County Public Schools have been recognized by the College Board for their success in expanding opportunities for Advanced Placement course taking, particularly for minority and low-income students (Broad Foundation, 2013; College Board, 2010). Over a ten year period (2001-2011), AP course taking increased by 340%, with the greatest growth occurring in English and Social Studies, though AP course taking across all subjects increased significantly. The share of minority students taking AP courses doubled in most subject areas and there was a substantial increase in students eligible for free or reduced-priced lunch taking AP courses. As the pool of AP course takers expanded, there was a drop of 0.1 to 0.2 standard deviations in the average prior achievement (8th grade standardized test scores) of the students. Thus, by extension classroom composition has changed over time. Given this context of AP expansion and increasing diversity, we estimate the impact of AP classroom peers and AP teachers on AP test takers’ probability of ‘passing’ the associated AP exam. We also explore whether or not the effects, if any, are different based on students’ prior performance. And, we explore if the effects vary over time. This study uses multiple cross sections of AP student course/test data that include information on student demographics, achievement and courses, AP classrooms and AP teachers from Broward County Public Schools over a thirteen-year period – 2000-01 through 2012-13 academic years. We estimate a probit model that pools the multiple cross sections together, taking into account the ‘nested’ nature of the data with AP course/exam takers nested in AP classrooms that in turn are nested in schools by using school fixed effects and the cluster-robust estimator for allowing within classroom correlation. We will explore different estimates that take into account the potential endogeneity of outcomes and class characteristics. Our very preliminary results, using five years of our data, suggest that classroom peers may matter in both positive and negative ways and this depends too on the subject. For example, the more free and reduced priced lunch students in AP classrooms the lower the probability a student will pass the AP exam and the higher average prior test score (8th grade z-scores) the greater the likelihood. AP teachers’ experience and education are positively associated with the probability of passing the AP exam.