Simple and Credible Value-Added Estimation Using Centralized Assignment

Many large urban school dis­tricts match stu­dents to schools using algo­rithms that incor­po­rate an ele­ment of ran­dom assign­ment. We intro­duce two sim­ple empir­i­cal strate­gies to har­ness this ran­dom­iza­tion for mea­sur­ing the causal effects of indi­vid­ual schools. In appli­ca­tions to data from Denver and New York City, we find that our mod­els yield high­ly reli­able school effectiveness […]

Posted in | Comments closed

Breaking Ties: Regression Discontinuity Design Meets Market Design

Centralized school assign­ment algo­rithms must dis­tin­guish between appli­cants with the same pref­er­ences and pri­or­i­ties. This is done with ran­dom­ly assigned lot­tery num­bers, non­lot­tery tie-break­ers like test scores, or both. The New York City pub­lic high school match illus­trates the lat­ter, using test scores, grades, and inter­views to rank appli­cants to screened schools, com­bined with lot­tery tie-break­ing at unscreened schools. We show how to iden­ti­fy causal effects of school atten­dance in such set­tings. Our approach gen­er­al­izes regres­sion dis­con­ti­nu­ity designs to allow for mul­ti­ple treat­ments and mul­ti­ple run­ning vari­ables, some of which are ran­dom­ly assigned. Lotteries gen­er­ate assign­ment risk at screened as well as unscreened schools. Centralized assign­ment also iden­ti­fies screened school effects away from screened school cut­offs. These fea­tures of cen­tral­ized assign­ment are used to assess the pre­dic­tive val­ue of New York City’s school report cards. Grade A schools improve SAT math scores and increase the like­li­hood of grad­u­at­ing, though by less than OLS esti­mates sug­gest. Selection bias in OLS esti­mates is egre­gious for Grade A screened schools.

Posted in | Comments closed

How Well Do Structural Demand Models Work? Counterfactual Predictions in School Choice

Posted in | Comments closed

Minimizing Justified Envy in School Choice: The Design of New Orleans’ OneApp

Posted in | Comments closed

Free to Choose: Can School Choice Reduce Student Achievement?

Posted in | Comments closed

Research Design Meets Market Design: Using Centralized Assignment for Impact Evaluation

Posted in | Comments closed
  • Archives