We consider regression discontinuity (RD) design with two, not one, running/forcing variables for a single treatment. The main identified entity is a 'double-difference-based ratio', where the numerator is a 'local difference in differences (DD)' for the response, and the denominator is a local...

The main attraction of Regression Discontinuity (RD) is local randomization around the cutoff. However, often the running variable (or just ''score'') is manipulated to ruin the local randomization. To detect manipulation, typically, the continuity at the cutoff of the score density function is...