POL20050 Research Methods in Political Science – TUTORIAL 5 EXERCISES

Flashcards covering key concepts from the Research Methods in Political Science tutorial.

Experimental, causal design

Random treatment assignment should solve almost all selection bias.

Covariate adjustment

Covariate adjustment will always increase precision in two-arm, randomized, experiments.

Cluster-randomized experiments

Cluster-randomized experiments may produce biased estimates.

Sub-group experimental design

The Difference in Conditional Average Treatment Effects (CATE)

Defier

A unit which does not take the treatment when assigned treatment, but does take the treatment when not assigned treatment

Encouragement design

An instrumental variable

Placebo-controlled experiment

The sample average placebo effect (SAPE)

Stepped-wedge design

A stepped-wedge design assigns treatment sequentially over multiple time periods

Randomized saturation experiments

Randomized saturation experiments assign the same saturation level across all clusters

Sample average treatment effect

A two-arm randomized experiment

Local average treatment effect

A regression discontinuity design

Average treatment effect

An encouragement experiment design