Polls
Even though a good friend, probably Nate, told me years ago that sample size (after a certain point) does not significantly affect error, I've continued to have a bee in my bonnet about media sources not citing the margin of error when they cite poll results. It seems to me easy to do and therefore a red flag that they were hiding something from me.
Listening to the Bryant Park Project just now, they cited the results of a gallup poll which irritated my old wound and sent me a searchin'. So it seems to be true that the sampling error based on sampling size is fixed, but a greater determinant of the accuracy of a poll turns out to be the response rate, which is rarely reported.
Introduction to Sampling page at UC Davis

Recent Comments