Regression Discontinuity Design

摘要: The Regression-Discontinuity (RD) research design is a quasi-experimental that can be used to assess the effects of treatment or intervention. unique characteristic use assignment solely on basis cutoff score pretreatment measure. In its simplest form, has two groups (those scoring above and below cutoff), pretest (the variable) post-test outcome). More complex variations are also possible. major attraction RD designs they estimate treatments given those who most need deserve them, provided merit determined as qualification variable (i.e., cutoff) nothing else. Because does not require some needy individuals being assigned no-treatment group, it ethical advantages over experiments for assessing effects. have lower statistical power than randomized experimental designs. Avoiding model misspecification, especially threat posed by curvilinearity in pre-post relationship, analytic challenge. analysis accomplished using general polynomial regression model. Alternative approaches been offered