Random does not mean perfectly representative in every draw. Chance alone can produce a sample with somewhat too many or too few members of a group, which is why estimates have sampling uncertainty. Larger probability samples usually reduce that chance variation, but size cannot repair a frame that excludes people or a process in which selected respondents systematically refuse.

Watch for common substitutes. A convenience sample recruits the easiest people to reach. A voluntary-response poll lets participants select themselves. Choosing every tenth name can be systematic probability sampling only if the starting point and list support it. On the exam, look for a known selection process rather than the vague claim that respondents were chosen 'at random.'

Video lesson: Science Behind the News: Opinion Polls & Random Sampling