Hello,

I am completing a data request to our State Education Agency for student level data from 2008-2017 and to our state higher education agency . The data at the public school level will contain by year, the annual testing record which includes variables such as sex, race, wealth, special education, annual test results from grade 6-8 (Reading, math, science) and high school English algebra, and biology and the ACT scores and the school last attended. I am merging this data with data from the state university system which will include application to public universities in the state, data will include variable on: application and acceptance, wealth, parent education, freshman GPA, high school GPA, school graduated from plus others.

My research questions focus on the effect of the state public schools mandating ACT testing beginning in 2011-12 and application and acceptance rates, persistence, and freshman grades (GPA) across sex, race, wealth. The data from the university system will have about 50-75 thousand students in it for each year in question.

I am considering using Regression Discontinuity design since I will have at least 4 years of data pre- and post the policy change and enough covariates to deal with any variation on samples. I also thought a Difference in Differences could work, but wanted to hear a few opinions on this before I finalized my request since I need to provide a basic description of the proposed data analysis.

Thanks for your time,

Ted