We have weird evidence standards
In order to suggest that a program may or may not be working as designed, we tend to expect our quantitative evidence to carry a high level of statistical significance. We also tend to look for large random samples.
Because we wouldn’t want to inaccurately imply that a program is not quite as effective as everyone would like…
Alternatively, in most US states with the unanimous decision from 12 death-qualified jury members, a person can be condemned to death.
Should we be treating federal programs with similar standards as criminal cases? Or should we be treating criminal cases with similar standards as federal programs?