I received the following this morning from a friend of mine who teaches Political-Science and specializes in Presidential Elections:
Are the polls telling us it’s all over or just getting started?
The presidential race is still quite competitive. Bush is a little ahead because Kerry lost some ground with the Swift Boat ads and Bush got a “bounce” from his convention. (A “bounce” in the polls after the party convention is normal, and it tends to dissipate after a few weeks.) Kerry's new aggressive focus on Iraq and on the failures of Bush's war on terrorism should focus people’s attention on the major issues of this race. The debates will give Kerry the opportunity to make his case in the major forum of the campaign. (Of the eight elections in which TV debates were held, the incumbent gained in one ('88), two were essentially a wash ('84 and '96, both were obvious victories for the popular incumbent), and four benefited the challenger ('76, '80, '92, '00).
The national polls mostly show a close race and there are several reasons to doubt the “predictive power” of the polls, in general and this year in particular. Polls are a rough estimate of voters’ preferences: “If the election were held today, would you vote for Bush or Kerry?” But the election isn’t going to be held today. Much can, and likely will, change in the next six weeks. Even good scientific polls are often off by 5% or more. In 2000, five days before the election, all but of one of the national polls had Bush winning the popular vote by 2-9%, but the election was essentially a tie, with Bush losing the popular vote by half a percent. (Polls often shift substantially. In 1988, Dukakis was 17% ahead of Bush, Sr. but ended up losing by 8%. In 1992, Clinton was running third in the polls, behind Bush, Sr. and Ross Perot. In 1980, a week before the election, the national polls showed Reagan and Carter tied; a week later Reagan won by 10%.)
Trial heats, among registered voters, including “leaners”
Survey End Date % Bush % Kerry Margin
- Time 9/23 48 44 4
- FOX 9/22 45 43 2
- CBS 9/22 49 41 8
- GQRR 9/21 49 49 0
- NBC/Wall Street Journal 9/19 49 46 3
- Zogby 9/19 47 44 3
- IBD/CSM/TIPP 9/18 44 43 1
- Gallup 9/15 52 44 8
- ICR 9/12 49 45 4
At this point perhaps 20% are undecided, but polling organizations “push” voters to choose a candidate. They do this by asking follow-up questions about how voters are “leaning”. They do this because news organizations want to be able to report who’s winning; reporting that it’s still uncertain is more accurate but less interesting.
This year there are some questions about turnout (the proportion of people who actually vote). Both parties and several independent groups are working hard to register new voters, especially among young people and minorities. Polls are not as good at measuring potential turnout as they are at measuring preferences. For example, recent political science research shows that the Gallup poll method of estimating “likely voters” is flawed. (Some researchers also question whether phone surveys that don't include cellphones are accurate.) Finally, this is not a typical election: the stakes are so high, the issues so unusual and so important, and the candidates differ so substantially, that some of the standard rules of polling and predicting may not apply. (I’m not saying that the polls are wrong; I’m just saying that they may not be reliable predictors of what will happen in November.)
This year, Bush is 20-40% ahead in some states; but the “swing states” (battleground states) are much closer. For example, the recent state polls are within the margin of error in the three most critical states: Florida, Pennsylvania and Ohio. In the Electoral College it doesn’t matter if a candidate wins by ten percent or one tenth of a percent. So racking up big margins in Texas or Utah or Mississippi doesn’t have an impact on who wins the Electoral College. As of now, the Electoral College looks pretty competitive. Kerry is clearly ahead in 11 states (including California, Illinois and New York) with 164 electoral votes; Bush is clearly ahead in 22 states (Texas and a bunch of smaller states) with 183 electoral votes. The remaining 17 states (with 191 electoral votes) could go either way. 270 are needed to win. (In 2000, Bush had 271 electoral votes, Gore had 266.)
There is much potential for change in the next six weeks. Kerry’s new focus on Iraq may energize voters who oppose the occupation. The debates may cause the public and the media to focus on terrorism, Iraq, jobs, the deficit and health care, rather than on what the candidates did in Viet Nam. Events in Iraq or elsewhere could influence voters’ decisions. Finally, there are always some folks who don’t decide till the last few days. Usually, the race gets tighter as Election Day approaches, and more often than not, “late deciders” tend to favor the challenger. As Yogi Berra used to say “It ain’t over till it’s over.”
Watch the debates! The first one will be this Thursday (9/30), 9:00-10:30 pm (EDT). And here’s my advice. As soon as the debate is over, turn the TV off! Form your own opinions; don’t let the commentators fill your head with drivel.