Comment: Statistically Speaking

(See in situ)


Statistically Speaking

1,100 sample size is an OK sample size for a population that is large. That's why the margin of error was about +/-3%. However, the trick is in the sampling process. No sampling process is truly random and not all 312 million Americans are in the pool to be sampled from. For example, they collected the survey through telephone calls, assuming land lines, then we excluded all the households that do not have a land line. And there are many other issues involved in how they collected the samples. Unless they release the location of all 1,100 samples, we won't know if there's any problem coming from that part of the process.

But statistically speaking, a simple random sample of 1,100 adults is good enough to make an inference on the true population proportion.