From a random sample of 500 registered voters in LA, 400 indicated that they would vote in favour of a proposed policy in the upcoming election. How many people should be sampled so that the sampling error is at most 0.04 with a 95% confidence interval?