Since you didn't get the choices, I can only give you some direction in selecting your answer.
For the best sample, you want to find a sample that is truly random. Calling names from a directory would be a good example of something random. Selecting random student numbers from a list would be a good idea.
You want to avoid samples that aren't random. For example, don't survey everyone in a particular class. You may get a biased result in favor of that type of class.
