Nowadays, polls are so common that a telephone poll was done recently to estimate how often individuals are surveyed (the answer was about once per year).
Written by Columbia University statistician Andrew Gelman.
Also read Lance Fortnow, a Northwestern University professor of theoretical computer science, complaining about this issue and the related matter of self-selected sample bias.
Meta-curiosities aside, Gelman’s statement is part of a very interesting blog post (and follow-up) about whether or not it is rational to vote given the low probability of one vote being decisive.
Gelman and his colleagues blog at Statistical Modeling, Causal Inference, and Social Science, one of the most consistently-educational and useful academic discipline blogs. Although some of the posts discuss very specific issues in applied statistics and are not intended for lay audiences (I don’t know R, although I hope to learn one day, so I have to skip the posts about statistics coding), Gelman often links to and explains his own research. The papers analyzing voting, districting, party affiliation, and other political issues are especially interesting (and timely, considering that we’re approaching a potentially historic presidential and general election season in the US), and he also throws in posts about methodology/philosophy in statistics and social sciences for variety. A new book summarizing his applied research on US elections and voting behavior, Red State, Blue State, Rich State, Poor State (see here also), is due to be published this fall.
P.S.: Thanks to Eli for pointing out in an unrelated conversation a few months ago that meta is Greek for “after”.