Suppose a population contains 40000 people. All else being equal, a study based on a population sample that includes which of the following numbers of respondents would be the most reliable?