indisgz, dont know where you came up with the 25%. i don't think its too big or too small. it's wrong. usually, the standard error is obtained from the size of the sample size, and the degree of variation in response within the sample size. assuming that responses are 0 or 1, the degree of error decreases as the square root of the sample size if the distribution is normal. i'm not trying to show off my knowledge here, just pointing out the facts. in this case, this point may not even be relevant, because the sample is not randomly taken. for example, if you sampled the top 0.1% wealthies americans and asked them for their wealth and reported that the mean wealth of all americans is equal to the mean from your sample, what would your standard error be? the error margin does not make sense, becaues the sample set is biased.