Skip Navigation
Funding Opportunities | Search Funded Research Grants and Contracts

IES Grant

Title: The Causal Impact of Attending a Career-Technical High School on Student Achievement, High School Graduation, and College Enrollment
Center: NCER Year: 2016
Principal Investigator: Dougherty, Shaun M. Awardee: University of Connecticut
Program: Improving Education Systems      [Program Details]
Award Period: 4 years (7/1/2016-6/30/2020) Award Amount: $694,741
Goal: Efficacy and Replication Award Number: R305A160195

Co-Principal Investigators: Eric J. Brunner, Stephen L. Ross

Purpose: The research team will examine the impact of attending a career technical education (CTE) high school on students' achievement, high school graduation, and college enrollment. In recent years, there has been an increasing focus among educators and policy makers on the possibility of CTE to meet the growing need for career readiness among students. However, there is limited causal evidence to support the promise of alternative secondary education models that focus on CTE. This study will use admissions criteria and longitudinal data to estimate the causal effect of being admitted to and attending one of 16 oversubscribed high schools in the Connecticut Technical High School System (CTHSS), where all students participate in some form of CTE, relative to attending a traditional comprehensive high school with fewer opportunities to access CTE.

Project Activities: Researchers will merge admissions data from CTHSS with the state administrative data from the Connecticut State Department of Education (CSDE) to allow for the development of a longitudinal database following applicants through high school and into college. The researchers will use application data and information on admissions score cutoffs to compare students who were similar on their application score, but who differed in whether they were offered a spot in a CTHSS school based on the limited availability of seats. Using a regression discontinuity design, differences in outcomes between students who had similar scores on admissions criteria but who did or did not enroll can be attributed to the causal effect of attending a CTHSS school. Researchers will also collect primary data on the quality of CTE instruction and the allocation of resources at CTHSS schools to understand how these schools differ from traditional high schools and therefore what might be contributing to student outcomes.

Products: The research team will produce evidence of the effects of attending a CTHSS school on student achievement, high school graduation, and college enrollment, as well as the costs of this type of school relative to traditional high schools in the state. Researchers will also produce peer-reviewed publications of the findings, datasets, and documentation about how to acquire a restricted-use agreement to access the data.

Structured Abstract

Setting: The study will take place in Connecticut and include high schools in urban, suburban, and rural settings.

Sample: The sample includes approximately 54,000 students who applied to the CTHSS for entrance into one of the 16 schools for admission in 9th grade in the academic years spanning 2006-2007 through 2013-2014. These students represent all districts throughout Connecticut from which CTHSS students originated with roughly 40 percent from the state's largest cities.

Intervention: The intervention is attending a CTHSS school. CTHSS is an autonomous public school district of choice comprised of 16 high schools. The schools offer full-day programs to their students and all students take a substantial share of their coursework in a CTE program.

Research Design and Methods: Researchers will use a "fuzzy" regression discontinuity (RD) design using admissions data from the CTHSS and administrative data from the CSDE to estimate the effects of attending a CTHSS school on student outcomes. In addition, researchers will develop a standardized tool to collect primary data on the quality of CTE delivery in all CTHSS schools and a subset of control schools. Public school budget data will also be collected and compiled for both treatment and control schools to provide a cost analysis.

Control Condition: The control group consists of students who just missed the admissions cutoff score for attending a CTHSS school and attended another high school instead.

Key Measures: Researchers will use continuous scores for the CTHSS application criteria, indicators of whether a student was admitted to a CTHSS school and administrative data on student demographics and school characteristics. Student-level outcomes include proximal measures of attendance, discipline, and state test scores, as well as more distal measures of high-school graduation, college enrollment, and degree or certificate completion. Protocols for school observations and interviews regarding CTE delivery (e.g., number of CTE programs, industry credentials, and work-based learning opportunities offered) will be developed.

Data Analytic Strategy: This study uses a "fuzzy" RD approach, operationalized using instrumental variables in a two-stage least squares estimation strategy. In the first stage, researchers will model the relationship between a student's application score and whether they scored above the cutoff used to make an initial offer of admission to predict whether they actually enrolled in a CTHSS school in grade 9. By using the random offer of admission as an instrument for actual enrollment, researchers will isolate the exogenous variation in enrollment that was the result of receiving the random offer of admission. In the second stage, researchers will use the exogenous variation in enrollment to estimate the causal local average treatment effect of enrolling in a CTHSS school on subsequent student outcomes. Researchers will provide estimates for both intent-to-treat and treatment-on-the-treated. Heterogeneity of effects will be explored for students in cities and students from lower income backgrounds. The research team will collect primary implementation data in the CTHSS schools and quantify the data to use as an additional mediator in the quantitative models to estimate program impacts.