I wrote about guerrilla photojournalist Zombie's musings about the unreliability of poll data here. Yesterday, he decided to see for himself by volunteering at an Obama campaign phone bank. His reflections are interesting:
I feel, as I surmised in my essay, that the any polling samples generated this way are potentially way off, and exclude most voters who simply refuse to be polled. The real question is: How do those people intend to vote? Because the “unpollable registered voters” demographic is likely to be the largest demographic of all. And we have no idea how they intend to vote, nor why they refuse to be polled, and if there is some correlation between refusing to talk to an Obama campaign volunteer and refusing to vote for Obama. Could the same principle hold true for calls made by professional polling organizations?I suspect it probably does, and that the majority of the discrepancy between polls and election-day reality comes down to the fact that Zombie notes here, that the unpollable are probably the most important demographic in the world of opinion polling, which nobody really seems to acknowledge.