Bell Curve The Law Talking Guy Raised by Republicans U.S. West
Well, he's kind of had it in for me ever since I accidentally ran over his dog. Actually, replace "accidentally" with "repeatedly," and replace "dog" with "son."

Tuesday, October 26, 2004

What Can We Expect Next Tuesday?

An anonymous comment to one of the other postings asked "So RbR, a question for you: how do you feel about the likely outcome of this impending election?...What the hell is going on poll-wise? Are we likely to see voter debacles again resulting in the supreme court making the decision? What is the vibe you get, being a man on the street?"

What is the vibe you get, being a man on the street? Well, I should start by saying that I'm a poor choice as a "man on the street." I'm hardly an average voter. I'm a registered Democrat and I have a PhD in political science. Also, I live in Los Angeles - a fairly liberal city. So I have a partisan bias, live mostly with other people with the same bias, and I have far more than the average level of education. However, I have a large extended family with whom I'm in close contact and most of them are Republicans, some of them are Evangelical Conservatives. Finally, being "Piled Higher and Deeper," I'm more than willing to express my opinions about these sorts of things so here goes:

I've heard a lot of Democrats and people further to the left getting really bummed because the polls that get reported on CNN et al keep showing Bush ahead. Such depression is far from warranted.

What the hell is going on poll-wise? Here are my standard lines on poll watching. No single poll tells you much of anything. They are not perfect measures of anything and you have to look at a number of polls over time to find out anything useful. The news channels only report one poll most of the time. CNN is particularly bad because they over report the Gallup Poll which performed terribly in 2000 (predicting Bush to win the popular vote by a wide margin) and has been criticized for over sampling registered Republicans by as must 30%. When you look at lots of polls, you see that most of the polls show a statistical tie (neither candidate ahead by more than the margin of error).

Furthermore, Bush's level of support seems to be "stuck" at about 47% give or take. The conventional wisdom is that polls are pretty accurate about support for the incumbent. BUT, they tend to under estimate support for the challenger. If that holds this time, Bush is in BIG trouble because he's not getting over 50% outside of the Deep South.

Also, most polls are conducted by using land line telephones and then screening "likely voters" based on a criteria that vary from poll to poll. Both of these methods most likely under count young voters and newly registered voters. Specific polls of both of these demographics are showing that they support Kerry by as much as 2:1. That's why polls of "registered voters" show Kerry doing much better.

Finally, undecided voters tend to break heavily for the challenger (most people are saying 2:1 for the challenger). The logic is fairly simple. If you are a voter who is undecided about George Bush after 4 years, its not because you don't know enough about him. It's because you know him and you don't like him.

How do you feel about the likely outcome of this impending election? Personally, I think Kerry will pull off the victory on Tuesday but I'm no 100% confident about that. I think Kerry's support is being under counted for a number of reasons. While both parties are pushing to register new voters, the new voters for Kerry (young people and ethnic minorities) are distributed throughout the country and will likely be a major force in several swing states (Colorado, Florida, Nevada, Ohio). New registrants that Bush is counting on are Karl Rove's famous 4 million untapped Evangelical Conservatives who are disproportionately concentrated in states Bush already has sewn up. Besides the word is that polls of new voters show that as a group they support Kerry over Bush 2:1. Here is a link to a great website that updates its election day prediction after every days' poll.


Raised By Republicans said...

One other thing about the polls. Lately some of the polls have been jumping around a bit. Since the first debate, the polls have been rock steady and now some are fluctuating a bit - but usually within the margin of error or very close to it.

These fluctuations could either reflect voters who are changing their minds back and forth (doesn't seem right given how polarized the country is right now) or it is because of random error in the sampling/measuring methods used in the polls.

The Law Talking Guy said...

There are other poll issues. Two are well known. First, the media have commented on the "cellular only" crowd, which they estimate at maybe 3% of the 18-30 populace. How they estimate this is, incidentally, a bit of a mystery, given that they aren't ever polled. The expectation is that being young, they are Kerryish. Second, the media has noted the "likely voter" conundrum. Not knowing who is a likely voter skews results in favor of a smaller set.

Others are not as well discussed. Pollsters have a hard time reaching swing shift workers. This same applies to single workaholics, single people with big social lives, and anyone rarely home or rarely home during business hours. Two other groups of people rarely answer the phone: those with caller ID and a hankering for privacy, and those in financial trouble (since creditors begin calling every day). The latter is not a small group in America. Also, those with poor English skills evade pollsters. The sum is that pollsters are much more likely to reach families with younger children, those with 9-5 jobs, the elderly, and the unemployed. These X-factors are not entirely unknown. Most of these suggest, however, that the Democratic vote is undercounted (as we saw in 2000, when most polls on election eve showed Bush with a 2-3 point lead, but Gore got the popular vote by half a million).

These are other reasons why most expect that a "tie" in the polls bodes well for the Democrat.