POL SCI 3 Lecture Notes - Lecture 13: Internal Validity, Quasi
Document Summary
Researcher controls who gets x, what x is, and the setting. A subset of observational designs with high internal validity due to clever choices about. No randomization (1) when x occurs (2) on whom to gather data (3) when to gather data on y. Example: effect of police reform (x) on crime rate (y) Ics: treatment xz, control z, rule out nf, mat and regression, and history. Mits: x occurs more than once, rule out nf, mat and regression, build in replication (rules out fluke) Mics: hard to find control groups with the same zs multiple times, rule out mat, regress, mf, history (zs can repeat or there can be new zs_