Association for Public Policy Analysis & Management

School choice offers an important mechanism for equalizing educational opportunity, given that American public schools differ strongly in terms of achievement and student demographics. This paper estimates the causal impact of school choice in San Diego Unified School District on high school outcomes and, notably, postsecondary educational attainment, by studying lottery applications to a magnet program, an open enrollment program and a program designed to integrate schools. Students are followed up to six years following high school graduation. The sample includes almost 15,000 student applications to the three school choice programs over a seven year period.

Lotteries are conducted fairly in the sense that lottery winners and losers show no significant differences in baseline test scores or demographic characteristics. Importantly, this validates the lottery as a mechanism for creating an experimental control group that is closely matched to those students who won school choice lotteries.

Families on average applied to schools at which students had higher achievement, grades, behavioral outcomes and socioeconomic status than their local alternative schools. This result confirms that there is a "treatment contrast" between those who win lotteries and those who lose lotteries, and therefore a potential causal effect on student outcomes.

The paper examines a number of medium-term and longer-term outcomes. The secondary school outcomes include measures of being on track to graduate in time, actual graduation on time, and outcomes that are best viewed as at least partly behavioral outcomes, such as citizenship grades and attendance rates. The longer term outcomes focus on postsecondary educational attainment, and include number of years of postesecondary enrollment overall and by college type (two- versus four-year college), and degree attainment within four, five and six years of graduation.

The paper estimates intent-to-treat impacts -- the impact of being offered admission to a school choice program, and the impact of treatment on the treated, interpreted as the impact of being offered a slot in a school choice program and actually enrolling in the given school the fall after the lottery occurs.

In most cases, intent-to-treat and treatment-on-the-treated estimates suggest that school choice does not significantly influence students’ high school or postsecondary outcomes.

The paper also tests for systematic variations in program impact by student demographic characteristics, and by school characteristics. Theoretically, it is the difference in the characteristics of the school to which a student applies and those of his or her default local school that should mediate the causal impact of being offered a spot in the program. There are some exceptions but for the most part these tests show no systematic variations in impact by student background or differences in school characteristics.