Nigel Henry
June 20, 2015
Conceptually, it may seem counterintuitive that the views of a small sample (for example less than 1,000) can accurately estimate the views of tens of millions of people.
This leads to poor perception of public opinon polls.
This is how sampling error is usually taught…
Boring!
The user starts with an acutal population, of known distribution
population
DEM REP
499671 500329
He gets to see the actual results of the (few) people that he sampled
poll
DEM REP
305 295
Comparing he sees that with just 600 responses, he has approximated the population within 0.04
poll
DEM REP
0.508 0.492
The app is learner-friendly, so that one can see the effects of