Which sampling method is most likely to result in undercoverage bias?
A random sample from a phone directory
A stratified random sample of all citizens
A census of a small town
A simple random sample of a student list