I have data from the Census bureau that has means and margins of error that are defined at 90% confidence. For example, a zip-code population estimate of 17042 with a 90% moe of 278. I'd like to do some inference based off of this data, but I need to reverse engineer to get the standard error of the estimate. Given the mean and this margin of error, how can I come up with a standard error?
Asked
Active
Viewed 27 times
2
-
Would you object to saying "based on" rather than "based off of"? ${}\qquad{}$ – Michael Hardy Dec 21 '15 at 00:22
-
How did they come up with this population estimate? The comment by @AndréNicolas seems to depend on certain assumptions: a random sample from a normally distributed population, in which each member of the sample has an expected value equal to the thing you're trying to estimate. That doesn't appear to be the case. Are they trying to estimate the size of the population? Or some variable that assigns some number to each member of the population? Or what? ${}\qquad{}$ – Michael Hardy Dec 21 '15 at 00:26
-
@MichaelHardy: You are right, there is no reason to think the estimator in this case has a close to normal distribution. – André Nicolas Dec 21 '15 at 00:31
-
@AndréNicolas : That was only a small part of my point. ${}\qquad{}$ – Michael Hardy Dec 21 '15 at 16:39