Saturday, October 11, 2008

Why McCain Is Still Able To Win

Thanks to everyone who has written in, I appreciate being able to make a difference. But I still hear a lot of worry and doubt from republicans about whether McCain can really win the election. This article is one reminder about why this race is still a very close race, and why you not only must not give up, but why you can make a big difference in this home stretch.

I wrote Friday about the fact that in seven of the last eighteen presidential elections – 38.9% - the polls were wrong by a big margin on the race, and so even a big lead for one candidate does not guarantee diddly. I have discussed the principles of statistics to show why the present polls have a critical flaw somewhere in their methodology, since the results are not in line with the confidence criteria. But there’s another way to take apart a poll to see whether it is useful as a barometer for opinion.

The widest lead enjoyed by Obama in the major polls Friday, was the Gallup Daily Tracking Poll. Gallup has been in the business longer than anyone else still doing polls, so it appears reasonable to consider Gallup’s reports to have solid credibility. The problem, of course, begins with the fact that so many other polls show the race being closer. But let’s look at that Gallup poll to see if we can learn from it.

Friday’s tracking poll had Obama up by ten points. But that lead is actually the result of two separate tracks, one for Obama and one for McCain. 51 to 41, so they say.

OK, that’s their starting number, the one they put in the headline. What drives those numbers, I wonder? I always want to see the internals, but Gallup has been getting sneaky, they release that data later, generally a week afterwards. The most recent detailed support by party I have is from the week ending October 5, when Gallup had Obama leading 50-42. I also note for reference that Gallup showed Obama leading 50-42 on September 28, 48-44 on September 21, and McCain leading 47-45 on September 14. The party affiliation for those dates should help us see where the changes came from.

Let’s start with Gallup’s base support for the race, conservative republicans for McCain and liberal democrats for Obama. Here’s how that looked:

Sep 14: McCain 95%, Obama 94%
Sep 21: McCain 93%, Obama 95%
Sep 28: McCain 93%, Obama 95%
Oct 05: McCain 94%, Obama 95%

Pretty comparable, noting to show a reason for changes. Next up, cross-party support, conservative democrats for McCain, liberal/moderate republicans for Obama:

Sep 14: McCain 17%, Obama 16%
Sep 21: McCain 19%, Obama 15%
Sep 28: McCain 19%, Obama 15%
Oct 05: McCain 16%, Obama 19%

Well, Obama got a boost going into October, but this is a pretty small group, hard to see it swinging the overall vote by the way we’ve seen. That leaves the independents:

Sep 14: McCain 38%, Obama 24%
Sep 21: McCain 31%, Obama 22%
Sep 28: McCain 31%, Obama 22%
Oct 05: McCain 32%, Obama 23%

McCain has a good advantage here, so that cannot explain the deficit. As I explained before, the only way this can be happening, is that Gallup has weighted the democrats more heavily, assuming that they will be a greater portion of the voter population this year than in past elections.

I have read from a lot of people who say this means that the polls are trying to lie to us, but that’s not valid, at least not in the case of professional polling groups who need to protect a reputation for accuracy and integrity. To understand how Gallup would honestly believe more democrats would be likely to vote, we have to infer that from evidence we are not shown.

Polls always release results from a respondent pool, but they do not tell you – although the NCPP says they always should – what proportion of people agreed to answer the poll. If a lot of folks refuse to take part in the poll, that’s an important piece of information. If Gallup is getting pushback from people they are trying to contact, they may be interpreting that in a way that may skew their political weighting.

Now one thing I do not buy, is the idea that Obama has 51% support. Not that he could not get that much in the vote, but if that is true it would mean that his support has risen six points in three weeks. What would cause that to happen? It’s not going to come from the democrats, he already had their support. It’s not that republicans would swing over and support Obama, the numbers tell us that and in any case, the parties are very partisan this year. And the independents have not changed in Obama’s favor, in fact he’s a point weaker with them than he was September 14. So what’s driving the bigger numbers? The only explanation is that more democrats in the poll are being counted, which means that Obama’s growth is artificial. In real terms, it means that republicans have become more passive, and the balance reflects a loss of McCain support on both tracks, his own support and Obama’s.

Why would McCain’s support drop? He’s come out weak in most conservative’s minds, letting Obama slide on outrageous lies. Republicans will still vote for him, but they will be less happy about it, and some may not vote at all. Also, while I think all three debates have been effective for McCain and Palin, it will take time for them to show any result, and republicans have to reinforce the messages of competence, intelligence, and sound judgment.

Why is this good news? I mean, this helps explain why Obama is ahead, but how does this help McCain win? Basically, it comes down to three key facts. First, if republicans get back into the fight, we can bring things back in line. It really is that simple, that is republicans match democrats at the poll, John McCain wins and Barack Obama goes back to being a first-term Senator. Second, it’s something we can control. Convincing republicans to get more active in supporting McCain is not all that easy this year, but it’s easier than convincing independents who are varied and stubborn, and it’s very good news that Obama has still not sealed the deal with a lot of democrats. Third, it must be said that in the present financial crisis, John McCain not only missed an opportunity, but fumbled at least twice in his answers. But Obama failed to capitalize on it, his answers have been no better and for many people only emphasized his inexperience and na├»ve understanding of the economy. If John McCain gets out and presses a clear, sound economic policy, he can turn this into a great advantage.

We are in the final weeks of a long, long battle. We are tired, bitter at the cost and some of our losses, but we can come out with a clear victory. It will be rough, these last 24 days, with all the dirty tricks from ACORN and the ACLU, all the lies in the media and hard-left interviews. But for all his advantages and biased allies, Obama has failed to close the deal. That door is open, and we can get there if we stick it out.


j said...

I've read on other blogs that Obama underperformed in the actual vote count in state after state during the primaries; New Hampshire being one example. The polls had him winning the state by approx. 6 or 7 percent and he lost by 2 percent. It would be helpful if you could write about this in some detail for us. Also, here in Pa. I'm seeing a lot of McCain signs and few Obama ones. In 2004, Kerry signs were far more numerous and still he only won by about 140,000 votes. I'm also hearing anectodal evidence that a lot of Dems are voting for McCain, which I'm reading about from others on the blogs. One coworker told me that his wife (a Kennedy Democrat) will be voting for McCain.

Lea said...

Thanks again for this analysis. Very helpful indeed and helps to give me hope. I always thought that McCain would win because I still believe we are a right of center nation. I will be very disheartened if it turns out otherwise.

Did you see the Fox News poll that came out on Friday? I was hoping you might comment on that one too.

jaywhite said...

I am currently living in Northern New York state. Yesterday, while watching TV I saw at least 10 Obama commercials. According to all the polls published, especially recently,in New York state Obama has close to a 20 points lead. Why would the Obama campaign waste money on buying commercials if he has such tremendous lead less than 4 weeks before the election? After reading your last few posts,I wondered if, given what you said, the polls which the candidates have shows a much different picture.

Ronsonic said...

My much simpler and less analytical take on the polls. In a situation where almost all broadcast media, most print journalism, labor unions, academia outside the hard science geeks and public sector employees all heavily favor one candidate, what makes us think that polls would be that different. That they have a financial interest in being seen as objective? No more so than Hollywood or the press and broadcast networks.

In 2004 we say activists gimmick exit polls by showing up to be interviewed. I don't doubt they have new and better means of manipulation now.

Don't trust the polls.

Anonymous said...

You say that the pollsters aren't deliberately skewing their weightings because "they need to protect their reputations".

Why? I mean, seriously: as long as they are reasonably close by election day, no one can definitively prove they were "wrong" in early October. "Wow! what a surge by [insert Republican nominee] during October!"

So, if they are in the tank -- who is gonna know? So what if the partisan breakdown of every poll shifts in late October from massive D oversampling to somewhat close to historical voting norms Every. Single. Year.

Those R's are just really good closers. What a great ground game and GOTV operation, huh? Pay no attention to the pollster behind the curtain...