The Fifth Column

Thanks to Watchblog for highlighting this Zogby text-message poll. One of the growing concerns in this day of increasing communications modes is whether the traditional home-phone polling will continue to give accurate results. A small but growing group are voters like myself: young cell-phone users with no home phone. There's no way a pollster for a normal tracking poll could reach me.

Zogby worked with Rock the Vote to poll 18 to 29-year-olds who rely "primarily if not solely" on their cell phones. The results showed an almost identical pattern compared to traditional polls of the same age group: Kerry led 55% to 40%. Of course, it's an unscientific, self-selective poll, since it's easier to delete a text message than respond. I suspect that Bush does better among politically unengaged voters than does Kerry, and I also suspect that more of these cell-phone-only voters are concentrated in non-swing Blue States, where technology and mobility are key.

This poll has to be reassuring to the Bush campaign, despite the 15 point spread against him. In 1948, phone-less voters confounded the confident pollsters by voting overwhelmingly for Truman and handing him a landslide win where all expected him to lose big. The fact that phone-less (in 1948) and cell-phone-only (in 2004 and beyond) voters are not randomly selected from among the population means that polls can be significantly wrong in a close election. The indication that cell-phone-onlies mirror the regular population is good for Bush.

Relatedly, one mistake made by the media (and hence by voters) is that polls were accurate throughout the campaign. It is of course in the media's interest to lend credence to the polls since they both administer and report them. Thus, a result of a landslide for either candidate will be seen as an anomoly, or a result of better get-out-the-vote drives. This may be true, but it is also possible that the polls failed to account for structural changes in communications or society. This has been seen recently in referenda on gay marriage. People are (wrongly) ashamed of their opposition to gay marriage, and will lie to a pollster about how they intend to vote. Then, on election day, they vote their gut and the gay lobby is confounded by a ballot-count that does not mirror the polls.

So on November 3rd when the media tells you how shocked they were at the result, don't be taken in. They're only shocked because they forgot that the ballot-box, not the telephone, is the true measure of American democracy.