Thought Theater

Attempting To Decipher The Polls genre: Polispeak

As expected, there has been a barrage of polls in the last few days. While it is impossible to determine what the final outcome will be on Tuesday night, I think it warrants a look at the polls to see if anything can be gleaned. The following is a list of those polls, the dates they were taken, the numbers on the generic congressional question, and the margin of error. Following the list of polls is my attempt to decipher the meaning of these polls and my own offering of some plausible and possible explanations...including commentary on some of the individual Senate contests.

Time Poll:

Conducted: November 1st through 3rd.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 55 percent; Republican candidate: 40 percent

Newsweek Poll:

Conducted: November 2nd through 3rd.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 54 percent; Republican candidate: 38 percent

Pew Research Center Poll:

Conducted: November 1st through 4th.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 48 percent registered, 47 likely; Republican candidate: 40 percent registered, 43 likely

Washington Post - ABC News Poll:

Conducted: November 1st through 4th.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 53 percent registered, 51 likely; Republican candidate: 43 percent registered, 45 likely

USA TODAY / Gallup Poll:

Conducted: November 2nd through 5th.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 51 percent registered, 51 likely; Republican candidate: 40 percent registered, 44 likely

CNN Poll:

Conducted: November 3rd through 5th.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 58 percent likely; Republican candidate: 38 percent likely

FOX News Poll:

Conducted: November 4th through 5th.
Margin of error: 3 percent
Generic Ballot: Democratic candidate: 49 percent likely; Republican candidate: 36 percent

As I review the data, it seems particularly important to look at the dates the polls were conducted given the recent events that may have had an impact on the numbers. The Times, Newsweek, Pew, and Washington Post polls were all completed no later than November 3rd or 4th indicating that they would have recorded some of the impact of the John Kerry remarks and the associated media attention the story received. In these four polls the average on the generic ballot (using likely voters where available) was 51.75 percent for the Democratic candidate and 41.5 percent for the Republican candidate.

The USA TODAY, CNN, and FOX polls were all conducted to include voter data from November 5th and only the USA TODAY poll included data from November 2nd. Note that the USA TODAY poll has the narrowest margin on the generic ballot of likely voters...which one could argue partially reflects the momentary impact of the Kerry story. In these three polls the average on the generic ballot (using likely where available) was 52.67 percent for the Democratic candidate and 39.33 percent for the Republican candidate.

While I'm not an expert on polling, my understanding is that it takes at least a couple days for current issues to be reflected in a poll. If you look at the Kerry incident, it made its way into the media on Tuesday, October 31st and it remained a topic through the Thursday November 2nd cycle. The Ted Haggard scandal first hit the media on Thursday November 2nd and remained a topic through Sunday November 5th. When looking at these dates in relation to the dates of these surveys, there is obviously a degree of overlap and an imperfect division relative to these two media blockbusters. Nonetheless, the polling results seem to reflect the impact of the two issues whereby the first four polls show a larger slippage of Democratic support while the final three polls suggest a greater slippage for the GOP.

As I attempt to make sense of the numbers, my first inclination is to surmise that the two issues are virtually offsetting. The first four polls reflect some skewing towards the GOP and the latter three polls (particularly the final two) suggest a skewing towards the Democrats. So where does that actually leave us with regards to voter sentiment and the final numbers in tomorrows election?

My own speculation is that the race isn't as close as the first four polls indicate but the Democratic advantage isn't as large as the final three polls would suggest. I've always suspected some narrowing of the Democratic lead as unsatisfied Republicans come to the conclusion that they will hold their nose and support the GOP candidate...primarily based upon fears that Democratic control of the Senate might jeopardize future Supreme Court appointments...which better explains the close Senate races in contrast to the growing evidence of Democratic strength in House races.

I can't recall a time when concerns about the make up of the Supreme Court were more prevalent and I cannot find a better explanation for the disparate polling information that seems to show Democrats doing very well in contested House races while the Senate races seem to be tightening...and some of these Senate races actually appear to be trending Republican. Supporting the possibility that the Supreme Court consideration can explain voter differences between the House and Senate races would be the argument that there is clear voter opposition to the war in Iraq and a desire to impose some accountability on the Bush administration...and that would be in the form of a Democratic House.

Historically, in elections where the House switches control from one party to the other, the Senate also follows suit. Current polling in the individual Senate races seems to indicate that this election could defy that historical trend. I would argue that this may well happen based upon voter concerns about Supreme Court appointments which may lead enough swing voters to vote against their GOP House candidate while still supporting their GOP Senate candidate.

After all, it is the Senate that must approve Supreme Court appointments and voters may still favor conservative appointments...especially if one considers the opposition to same-sex marriage and a general belief that the Democratic Party is against any limitations on abortion rights. In other words, if voters want their unhappiness with the war in Iraq to be heard while still endorsing the social issues they seem to favor, the best solution would be to elect a Democratic House and maintain a Republican Senate.

When looking at individual Senate races, there are some factors that could override the above considerations. Lincoln Chaffee is a moderate which could play both ways in a predominantly Democratic state and may take Rhode Island out of the Senate equation I'm proposing.

In Missouri, the stem cell issue could also change the above equation as stem cell research seems to be favored by a majority of voters. The fact that this race has remained a virtual tie for weeks would seem to indicate just how torn voters may be on this particular race and the impact their votes may have on these underlying factors.

In Montana, Conrad Burns has a cloud of corruption hanging over his head in a predominantly Republican state that is likely opposed to late-term abortion and same-sex marriage. The fact that Tester has led in this race would suggest that the corruption issue is hurting Burns. Nonetheless, the recent surge by Burns may be an indication that voters are afraid to elect a Democrat (even though he indicates he is a social conservative) who will likely be pressured to vote with his Party on the big social issues.

Virginia seems to be self explanatory. Webb is a former Republican and Allen has made numerous mistakes. Again, this race is too close to call and one that may fail to fit the model I've proposed. Regarding Maryland, Pennsylvania, Ohio, New Jersey, and Tennessee, it seems to me that local factors and prevailing political affiliations may overwhelm any of the considerations I've defined herein.

With that said, voters will soon have the last word. It will be fascinating to dissect the results.