Fun With Numbers: Polls vs Google

There are alternatives to polls. If you are a betting person, or curious about them, you can look at sites like InTrade where people bet money on political races. You can go to fivethirtyeight.com and see what Nate Silver thinks. His area is baseball odds, but this season he has applied it to political races. The problem is that some betters do not even live in the US. Others do but may be more logical about betting money and more emotional when they actually vote. This would seem to skew the results in the "good direction" and then we will not be prepared for the "worst scenario." (See also Hominid Views)

Then there is Google. Huffington Post had this cool article that I saved because I love numbers even though I am rotten at math. Put a dollar sign in front of the numbers or make them practical and it's a whole new game.

Google Trends is able to show the popularity of various searches. Last election, it was possible to predict the outcome of several races quite well using Google Search data. The results could be misleading, though, because they reflected passive search data - there was no way to tell whether people were searching for information, donating, or looking for hit pieces, just given the name of a candidate.

Pollsters spend millions to get specific data. That said, Google data mirrored poll data such as from Rasmussen. I show you below an average of major national polls, from Pollster.com and it's a couple of days old. (The race is supposed to be "tightening.")
This is for an entire year.

Now take a look at the Google data from the same time period (candidate search.)

You can see convention "bounces" and the tightening of the race, the small McCain lead briefly and the spread to Obama majority.

The article shows several Google state polls, but I'll show you the one for Pennsylvania, since McCain insists he must win this one to have a chance. The article does several state breakdowns like this for battleground states (for the last month period), with good agreement with poll averages. It's interesting but only so reassuring.

I did not even discuss my addiction to electoral maps, such as at Americans Abroad . I particularly like the ones that allow you to change states from "solid" to "leaning" and to review the results of past elections. The 1968 and 1982 maps are particularly troubling, and I feel slightly better when I look at them, which is perverse.

Then there is still the spectre of voter fraud and voter disenfranchisement. I will not even get started with the Greg Palast/RFK Jr. scary articles on vote fraud, or "Uncounted" or the Mark Crispin Miller book. Even the venerable, conservative-leaning TIME magazine has to come out with an article called "Seven Things That Could Go Wrong on Election Day". I realize that complaints about "mainstream media" ignoring voter fraud are now moot, which is not reassuring because it validates them out of conspiracy terroritory. My uncle sends me an article this morning from the N Y Times called This Year's Butterfly Ballot. It seems you have to know to check off a box at the top and still check off the other boxes below. Then there is this one from today, from CNN, about purging of the rolls in battleground states. It's creepy.

My ballot is all filled out on the counter, I know an official drop box to take it to, and I have a tracking number. It is slightly more than a week that we have to wait, and alot of early voting has already happened. There are also three good articles on the wierd phenomena of the "undecided voter," two by Mark Morford at SF Gate (and this one is even better), the third by David Sedaris. The question is - HOW can someone be an undecided voter, at this point?