Experiments in Political Science Research

Experiments

Research methods increasingly used in political science.

Randomized Experiments

Address endogeneity by ensuring exogeneity.

Endogeneity Problem

Issue where cause and effect influence each other.

Exogeneity

Condition where variables are independent from each other.

Average Treatment Effect (ATE)

Effect of treatment calculated across the full sample.

OLS Model

Statistical model used for estimating relationships.

Treatment Variable (D)

Indicates whether a unit is treated or control.

Attrition

Dropout of units from an experiment.

Balance

Equal means of covariates across treatment groups.

Compliance

Whether subjects follow treatment assignment.

Intention-to-Treat Model

Analysis comparing groups by assignment, not treatment.

Covariates

Control variables that may affect outcomes.

Blocking

Designing groups to ensure balance in covariates.

Natural Experiments

Observational studies using naturally occurring events.

Social Norms

Shared beliefs influencing voter turnout behavior.

Electoral Turnout

Percentage of eligible voters who cast ballots.

Control Group

Group not receiving treatment for comparison.

H0 Hypothesis

Assumes no effect of social norms on turnout.

HA Hypothesis

Assumes social norms influence electoral turnout.

Treatment Conditions

Different experimental groups receiving varied interventions.

CIVIC DUTY Treatment

Baseline measure common to all treatment groups.

HAWTHORNE Treatment

Informs subjects they are being studied.

SELF Treatment

Includes individual voting records in communication.

NEIGHBORS Treatment

Informs subjects about neighbors' voting records.

Statistical Significance

Likelihood results are not due to chance.

Generalizability

Extent to which findings apply to other contexts.

External Validity

Degree results hold true in different settings.

Natural Experiment Example

Lottery winnings studied to assess political attitudes.