The Causal Impact of High School Course Graduation Requirements

• In Event: Promising Scholarship in Education: Dissertation Fellows and Their Research

Abstract

Purpose of study
In recent years, state lawmakers have been working to increase high school course graduation requirements (CGRs) to improve student college/career readiness and to ensure the opportunity to learn for all. Yet, the existing evidence on the causal effects of CGRs is scarce due to the methodological challenges, particularly isolating the CGRs impact from potential confounding factors with nonexperimental data. This study addresses these issues by employing quasi-experimental designs with a collection of administrative and survey data from Michigan Statewide Longitudinal Data System (MSLDS) and High School Longitudinal Study of 2009 (HSLS:09). Analyzing both MSLDS and HSLS:09 provides the opportunity for a comparison of the CGRs effects estimated with a statewide student population and with a national probability sample. This study seeks to better inform the policy debates on CGRs by identifying and offering improved estimates of CGRs impact in terms of validity and precision.
Building upon the literature on school effects and social stratification, this study develops a set of hypotheses (i.e., educational productivity, educational equality, school structure, academic organization, and social organization) and empirically tests whether and how schools differ in CGRs impact. Both MSLDS and HSLS:09 provide the opportunity differently to enhance the conceptual and empirical understanding of how CGRs effects vary by school characteristics. Ten years administrative data from MSLDS allow for characterizing a series of school contextual features and for estimating the CGRs effects in longitudinal, multilevel settings that can account for time-invariant and time-varying school characteristics. The HSLS:09 data allow for constructing various additional contextual measures from a rich set of survey items and for hypothesis testing with a nationally representative sample. This study will contribute to understanding of how CGRs effects are produced at the school level and what modifications are needed to improve the policy or practice.

Study design and methods
This study analyzes the data from MSLDS and HSLS:09 separately. However, there are several shared features in both analyses. The treatment of CGRs in both analyses is defined at the school level. Both analyses focus on estimating the effects of (a) math CGRs, and (b) overall CGRs (total CGRs of English, math, science, and social studies). Notably, both analyses involve student outcomes measured by (a) 11th grade achievement score, (b) dropout status in the third spring since entering 9th grade, and (c) on-time graduation status.

MSLDS Analysis
MSLDS is an excellent dataset for examining the CGRs effects as it contains statewide information on students, teachers, and schools for multiple years before and after the implementation of a new set of state-mandated CGRs in Michigan in 2007, the Michigan Merit Curriculum (MMC). Under the MMC, CGRs in many Michigan high schools were exogenously increased, providing a “natural experiment” setting for evaluating the causal effects of CGRs. This study employs comparative interrupted time-series (CITS) design (Bloom, 2003; Jacob, 2005; Lee & Reeves, 2012) to identify the causal impact of the CGRs, in this case the MMC. The CITS design estimates and compares post-MMC changes in average educational outcomes relative to pre-MMC trends between the treatment and comparison schools. To account for within- and between-school variations in changes in school performance trends, the CITS design is conducted in a two-level multilevel framework, where time is nested within schools (Raudenbush & Bryk, 2002), separately for all dependent variables.

HSLS:09 Analysis
While the Michigan context and MSLDS data are prominent, it does not provide a generalizable setting for hypothesis testing across all demographic and geographical backgrounds. To improve the external validity, this study uses the restricted data from HSLS:09, the most recent national high school longitudinal study, which is specifically designed to explore policy-relevant issues with respect to school policies and contextual factors that may affect student educational trajectories and outcomes. This study employs the marginal means weighting through stratification (MMW-S) method to estimate the effects of CGRs. The MMW-S is developed for evaluating the effects of multivalued treatments with non-experimental data (Hong, 2010, 2012), which is particularly useful for this analysis because the main independent variables are defined on ordinal scales.

Findings and policy implications
This study will offer causal evidence on both whether and how schools differ in CGRs effects on student achievement and educational attainment. The results on how CGRs effects vary by school structural, academic, and social organizational factors can help to inform state policymakers and school leaders on improving the design and implementation of CGRs, specifically to make school-based adjustments or interventions.

Author

• Guan Saw, Michigan State University