Death-penalty deterrence cannot be tested through randomized experiments

Evidence

Resolving this dispute empirically is not as straightforward as it may appear. Society is not at liberty to conduct randomized experiments in which half of the condemned prisoners in one region are executed and the other half are assigned to life imprisonment in a matched region. Researchers must therefore try to tease out the unique influence of the death penalty from among many naturally co-occurring influences that can affect homicide rates. — Albert Bandura, Moral Disengagement, 1944

Categorization

• XREF: Connects to the natural-experiment literature and the methodological limits of quasi-experimental designs discussed elsewhere in social science reading.