Association for Public Policy Analysis & Management

Objectives: Early colleges are new models of high school designed to increase the number of students who graduate from high school and who continue on to and succeed in college. This paper presents results from a longitudinal experimental study designed to look at the extent to which the early college model, as implemented in North Carolina in the United States, achieves these goals. This paper will look at the impact of the model on high school and postsecondary outcomes including: successful college preparatory course-taking, college credits earned while in high school, high school graduation rates, and students’ enrollment in postsecondary education. The paper will also examine variation in impacts by sub-groups.

 Methodology: Located in North Carolina (which has the largest number of early colleges in the country), this study is based on a multi-site randomized field trial that capitalizes on the fact that some schools used lotteries to select students. From a pool of eligible students, early colleges enrolled students based on random assignment, and the study compares the students who were assigned to the treatment group (early college) with students who were assigned to the control group (generally the traditional high school in the district  or “business as usual”). The sample analyzed for this presentation includes a total of 1,651 students who applied to 12 different early colleges and enrolled in 9^thgrade in the 2005-2006, 2006-2007, 2007-2008, and 2008-2009 school years. These 12 schools enrolled a total of 18 cohorts of students, with five schools enrolling multiple cohorts.  

 The impacts of early colleges on these outcomes are estimated within an Intent-to-Treat framework, supplemented by an estimate for the Local Average Treatment Effect (LATE) (Gennetian, Morris, Bos, & Bloom, 2005). For each outcome, we calculated unadjusted means for the treatment and control groups. We then calculated adjusted impact estimates using multivariate linear regression models that include site indicators (or site fixed effects), interaction of the treatment indicator with the site indicators (site level impact estimates), and baseline student characteristics.

 Results: Analysis of these data show that early college students were more likely to have successfully completed the high school courses required for entrance into college (a 10 percentage point impact). They also attained an average of 22 college credits by the end of 12^thgrade compared to less than 3 college credits for the control group. Early college students graduated from high school at a rate that was 4 percentage points higher than the control group and had enrolled in postsecondary education at a rate that was 15 percentage points higher.  The final paper will include results by sub-groups.  

 Gennetian, L. A., Morris, P. A., Bos, J. S., & Bloom, H. S. (2005). Constructing instrumental variables from experimental data to explore how treatments produce effects Learning more from social experiments: Evolving analytic approaches (pp. 75-114). New York, NY: Russell Sage Foundation.