Early voters and election polls

With the rapidly increasing numbers of early voters, the polling community has had to adapt. Already in 2000, and following up in 2004, the National Annenberg Election Study included explicit questions designed for early voters.

The American National Election Study, which has been conducting election surveys since 1948, had to adapt as well. In 2000, they discovered an increasing number of respondents who, when asked about vote intentions, told the interviewer that they’d already voted. The 2004 study adjusted to this fact.

How about 2006 and beyond? Under the current balloting system, there are states where as many as a third of voters have already cast their ballots by the Friday before election day. Surely, these voters will show up in pre-election tracking polls.

Have the pollsters noticed? You bet they have. A poll recently released by the Tester campaign in Montana claims that “Montanans are voting for a change.” The poll shows a 21% lead for Tester among those who report already voting.

We’ve long wondered whether pre-election polling alters the decision making of voters. Do citizens flow to the majority choice? Will citizens stay home if it appears that their candidate is trailing in the public opinion polls? Now voters, journalists, and the candidates themselves can make estimates about actual votes.

The biggest effect, I suspect, will be on campaigns. Now a candidate can not only track his position in the pre-election polls, but among voters themselves.

The poll is reported here