Are Charter-School Students Harder to Educate
than Those in the Traditional Public Schools?
IN THIS CHAPTER we begin our in-depth empirical investigations of the issues we have presented in the first three chapters. Here, we investigate further the extent to which the families and students in charter schools are different than those in traditional public schools—the point upon which we ended chapter 2. In this chapter, we investigate the extent to which charter-school students may be easier or harder to educate than students who have remained in traditional public schools.
For many involved in the school-choice debate, the ultimate test of a reform is whether or not the academic achievement of students improves. Answering this question is much more complex than it may first appear, as many decisions must first be made about how achievement can be measured and how research should be designed to properly attribute causality to the reform. A consensus is growing in the educational research field that the gold standard for such research involves a design that examines longitudinal data on student test scores subsequent to random assignment to the reform. Even this design, however, has proven fraught with difficulties for the statistical analyst.1 We do not have such data and, thus, we do not enter directly the fray surrounding this issue. However, another point of debate in the recent controversy in the media and among policy analysts over the academic achievement of charter-school students is whether charter students are harder to educate than their counterparts enrolled in traditional public schools.
In this chapter we examine this question using data from the 2002–3 school year in Washington, D.C. We begin by examining a simple binomial model of the proportion of students in key demographic and programmatic categories linked to educability. As explained below, we then turn to the estimation of a more theoretically appropriate mixture model that assumes two latent categories of charter schools. We then present an analysis that moves beyond simple demographic/programmatic factors to consider measures of educability using our individual-level survey