Is Virginia "in play?" Just count the polls and taxes

With the 2004 Presidential election less than 80 days away, presidential polls will become an increasingly large part of everyday conversation. Each morning, online or in the newspaper, you can also check any of a half-dozen national presidential polls. There are also numerous state polls, purporting to give a preview of the likely Electoral College vote for president.

In some Virginia polls, the race between President George W. Bush and Sen. John Kerry looks surprisingly close, given the certainty with which Republicans have been able to count on Virginia in recent years. These polls have led some experts to declare Virginia “in play,” although I would note that no poll of any kind puts Kerry ahead in Virginia.

Are polls of any use this far out from Election Day, when the Republicans have not yet held their convention, and when there are still four nationally televised debates to go? How should we regard polls that put Kerry within striking distance in Virginia, when other polls put President Bush only a few points behind in Maine, as reliable a Democratic state as Virginia is a Republican state?

Polls are likely to be less useful during this election season than in recent years. To be of any use, polls have to have three characteristics. First, polls should not be taken over weekends. For reasons no one has quite figured out, Democrats tend to be at home Friday and Saturday evenings, whereas Republicans are out. Thus, polls taken on the weekend tend to over count Democrats.

Second, polls should have “likely voters” as their sample. Polls of registered voters, or simply of all adults, tell absolutely nothing about those who will actually take the time to cast a ballot on Election Day. Almost without exception, polls of likely voters put President Bush in a better position than polls of registered voters. John Kerry does best among those who will not commit themselves to actually voting, even to an anonymous pollster.

Third, the number of people polled makes a large difference. (Put differently, polls are one area of post-Bill Clinton politics where size does matter.) Any poll with a sample of fewer than 1,000 people is probably not a representative sampling of the national or state population. And this is becoming a real problem.

Modern technology is hampering pollsters’ efforts. An increasingly large percentage of homes have caller-ID on the telephones. An increasingly large percentage of people with caller-ID pick up the phone only when they recognize the caller. To get their sample of 1,000 likely voters, with the proper numbers of men, women, young, old, minorities and veterans, pollsters may have to speak to more than 5,000 people. To get 5,000 people to actually talk to them, they may have to place 50,000 or more phone calls.

This makes polling labor-intensive, and thus expensive. The caller-ID problem tempts pollsters to make do with a smaller sample, or to settle for registered voters, or to cheat on the representative makeup of the sample, or, finally, to make calls on the weekend, when they know at least the Democrats will be at home. Any one of these short-cuts will make a poll less reliable.

Differences in motivation levels also complicate matters. Four years ago at this time, George W. Bush was 15 or more points ahead of Vice President Al Gore. As we know from the 2000 election results, this is not because there were that many more Bush voters; it was because, four years ago, Bush voters were much more highly motivated than Gore voters. Today, the most highly energized voters are those who hate President Bush. They can’t wait for Election Day.

Their motivation makes Democrats more eager to talk to pollsters (or indeed, to talk to anyone who will listen to them to two minutes.) But the Bush haters can’t vote more than once (in theory), and polls do not reliably tell us how many Bush haters there are.

And this brings us back to Virginia. If it is difficult to find 1,000 representative likely voters on the national level, it is even more difficult to do so in a single state, especially when state polling agencies are less well known than their national counterparts, and thus more likely to have voters with caller-ID ignore their calls. Thus, it is reasonable to conclude that polls showing Kerry doing well in Virginia indicate that Virginia Democrats are more energized than Virginia Republicans.

No one who takes part in Virginia politics needs a pollster to understand this reality. Virginia Democrats are still basking in their win over the taxpayers this year, while Republicans are dealing with internal strife and a disappointing loss to the governor. But things are about to change. Two weeks from tomorrow, Mark Warner’s sales tax increase takes effect. In October, many Virginians receive their auto property tax bills, which in many cases will also go up.

We’ll see who’s more energized then. And we won’t need a poll to see it.