According to Drudge – The Race is Even!!

Citing the latest AP/GFK Poll, Matt Drudge has put at the top of his site the race is even.  Now, I am as much of a worrier as the next Democrat, but the poll is wrong and AP and Drudge’s representation is out of date.  Here’s the facts about the AP poll.

1) It is a sample of 1001 Registered voters and 800 Likely voters.  This is on the smaller sample size of the daily trackers.  Some of the sample sizes, by the bigger pollsters are 3000.  Those 800 participants were, according to the AP surveyed over 5 days or 160 to 200 calls a day.  Most of the surveys are over 3 days – so between 300-1000 calls a day (depending on sample size)

2) The poll includes some of Obama’s worst polling days in weeks – last Thursday through Sunday.  The tightening we saw in the other tracking polls late last week and through Monday of this week are only now being reflected in the AP poll.  Additionally, they are including only one day of ‘good’ polling for Obama this week – Monday.  Finally, all the other trackers are through yesterday.  AP’s is through only Monday.

3) The average tracker, prior to adding the AP to it’s total, at Real Clear Politics gave Obamaa 7.4% lead.  Even with this outlier, his lead is still 6.8%

4) Nearly all trackers yesterday provided Obama with a significant boost.  In the same Real Clear Politics trackers, Obama gained about 1.5% from Monday to Tuesday – demonstrating he had very strong polling numbers on Monday.

5) I went back and looked at the polling done Mid/Late October in 2004 to see what type of outliers showed up during the Bush/Kerry race.  Bush won by 2.4%.  The average lead by Bush around the 20th of October – per the Real Clear Politics polls, was 2.7%.  Oddly, or not oddly, the biggest outlier was a +3 for Kerry.  Guess who’s poll?  Yep, AP/IPOS.  They were off the mean by 5.7%.  This year, AP/GFK is off the Mean by 5.8%.  So they are oddly consistent.  What’s also facinating, is the 2004 late October margin matched the ultimate win total for Bush.

The point is to understand what the polls are telling you.  You need to look at a few factors.  The first is partisan weighting.  That is the what makes up biggest difference between polls.  How a pollster views the makeup of the electorate between Democrats and Republicans will drive results.  Second, look at dates, sample size, and total calls per day.  In a very fluid election, the latest results with the biggest sample size will be the best.  Third, understand the overall bias of the polls.  Fivethirtyeight.com and Pollster.com rank the different pollsters (AP/GFK I could not find).

There are biases each way.  While the +14 for Obama yesterday may have been an outlier, the +1 for the Battleground poll yesterday (which also included last week’s data) or +1 for AP are also outliers.  Look at the median and look at the general direction of the polls to get a better sense.  Obama had a tremendous day yesterday.  Today’s results, relatively speaking, are mixed but still good for Obama.

Finally, ignore Drudge. He has an agenda and he’s willing to push it with faulty data. I think he does have a ‘nose’ for news and is a great place to go during slow politics time.  But as everyone is amped up – his portrayal can really affect Democrats and Republicans feeling on the election.  And it is usually not fully accurate.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: