It should be a statistically significant random sample. While randomly testing townspeople might be the way to go, you may also want to include random people in the U.S. (although their lack of knowledge/interest on the law may skew the results). Basically, to prevent bias, use a random...