The Early College Initiative, funded by the Bill & Melinda Gates Foundation, seeks to create a new high school model for students underrepresented in college preparatory courses in high school and in postsecondary institutions. This reform model seeks to increase rigor and engagement in high school and smooth the transition to postsecondary institutions by housing small high schools on postsecondary institution campuses and placing students on a track to graduate from high school in five years with college credit. There are currently over 200 early college high schools (ECHS) serving more than 50,000 students in the U.S. North Carolina leads the nation with more than 70 early college high schools. This study uses linked administrative data from the North Carolina Department of Public Instruction, the University of North Carolina General Administration (covering the entire 14 campus public university system), and the North Carolina Community College System to examine the following three research questions:

1. Are ECHS more likely to enroll students underrepresented in college preparatory curriculum and in postsecondary institutions than high schools in the districts they serve?
2. Do ECHS students have better secondary outcomes (End of Course exam scores, high level math and science course taking, college credit / degrees obtained while in high school, absences, suspensions) than comparable high school students from the same districts?
3. Do ECHS students have better postsecondary outcomes (college enrollment, postsecondary remediation) than comparable high school students from the same districts?

This paper will contribute to the growing literature on ECHS by covering both a broader range of ECHS and a wider range of postsecondary outcomes, including postsecondary remediation. Experimental designs of ECHS effects offer enhanced internal validity at the cost of reduced external validity. For example, experimental estimates generally require over-enrolled schools, which may or may not be representative of the population of ECHS in a particular state or in the nation as a whole. This study will use non-experimental methods (propensity score analysis on intact groups with pre-tests and regression analysis with district and sending school fixed effects) to produce estimates. We will benchmark non-experimental estimates against experimental estimates produced for a subset of six ECHS by Edmunds et al (paper to be presented in this panel). By evidence for each of the research questions above, both for the ECHS with experimental estimates and for the ECHS without experimental estimates, we can better gauge the impact of this reform model and any possible heterogeneity in the treatment effects found.