POL222H1 Lecture Notes - Lecture 6: Democratic Peace Theory, Green 2000, Clientelism
Document Summary
As in how random assignment is used in lab experiments, random assignment is also used with similar rationale in eld experiments. There are differences because: in eld experiments, accounting for confounding variables is harder since real-world participants are not predictable. With regards to gerber & green 2000, the control group made up 80% of the participants. 2013, the treatment and control groups were equal (50/50) In wantchekon, only 2 villages received exclusively nationalist/clientelist promises (1 nationalist-only, 1 clientelist), while the rest, which had both promises presented to them, were used as control. In actuality however, the true causal impact of personal contact (value of non- compliance) was determined to be 8. 7 percentage points. Such was found via a statistical method; by the independent variable. Like in a lab experiment, the intent of a eld experiment is also obscured for the sake of ethical guidelines: this is important because using a real-world intervention as an experimental stimulus carries ethical implications.