Suppose a population contains 20,000 people. All else being equal, a study based on population sample that includes which of the following numbers of respondents would be the most reliable??