Hi, my name is Pollwatcher and I'm a Pollaholic. Ok, I admit I love reading polls, and I get a high when we're ahead in the polls and I get downright depressed when we're behind. But like most addictions, there's the imaginary world when you're using, and then there's the real world when you're sober. That roller coaster ride of watching the polls swing back and forth from day to day can give your emotions a bad case of whiplash.
So let's step out of the polling amusement park and take a look at the very nature of polling human beings.
First, we have the statistical error of the poll. You'll see a margin of error of anywhere from 2.5-5%. So with a margin of error of +- 4%, a candidate who is behind by 3%, could actually be winning by 1% or be getting blown away by 7%. But here's the big trick, that margin of error is dependent on the pollster getting a random sample of the voters who will actually show up on election day. We won't even go into cell phones, tired workers, and people who just won't respond to polls.
And then of course there are the undecideds and those who change their minds. Here's a 2012 article from the Washington Post that discusses these people.
In 2004, 11 percent of people said they made their mind up about which candidate to vote for in the final week before the election while in 2008 10 percent said the same.
Who is this "12 percent"? The NBC/WSJ breakdown suggests they tend to be younger (four in ten are 34 years old or younger), moderate (54 percent identified themselves that way), very pessimistic about the direction of the country (64 percent wrong track)...
Got that?! About 10% of the voters won't make up their minds for a couple of weeks yet. So if we've got a 4% margin of error based on questionable "voter" samples, and we've still got 10% of voters who have yet to make up their minds or who may switch, and we've got a $60 million dollar GOTV effort going, we can see that unless you're way ahead or way behind in the polls, you can change the outcome up until election day!
So as we get close to election day, should the polls nail the results down to a fraction of a percent? It just so happens that Sam Wang over at Princeton Election Consortium has done some digging and came up with some interesting numbers.
Here's how far a median of the pollsters final polls were off in some previous elections.
Notice how except for 2002, since 1996 the pollsters have underestimated the performance of the Democrats by a pretty good chunk, in several elections. My personal wild guess for the cause of this is that we're getting more and more polls that aren't so unbiased.
But the above is the median of ALL the polls in the database over ALL the Senate races. But what about the polls for individual states, in just the close races? Mr. Wang has compiled this information for the last couple of elections.
Well, see a bit of a pattern in these numbers? Look at 2010 closely. The Democrats OVER performed in 7 out of 8 Senate races. And in HALF the Senate races, they over performed by more than 4.5%!!! And these are the final polls.
So anyone who thinks these races are over because we're a few points behind in the polls, doesn't understand how fickle the voters are, and how far off the polls can be. We can swing the current polls 5-10% if we work hard enough over the next 2 1/2 weeks.
So lets jump off the pollster roller coaster and get back to the reality of getting our supporters to vote. Voter registration drives are still going on, early voting with absentee ballots has already started in several states and early voting in voting booths is starting up. There's lots of work to be done, so contact your favorite candidate's office or the local Democratic party office and GOTV! GOTV! GOTV!