Real Clear Politics Unrelable: Old Polls, Skewed Polls

Please note all polls quoted herein are from the morning of Wednesday, October 10th. The Internet has since trolled me yet again with RCP updating their polls to remove some old ones.

Something old, something new, something borrowed, something skewed. With the exception of the borrowed, that's the makeup of the Real Clear Politics average, aka the "poll of polls."

Before I continue, I'd like to say that I use Real Clear Politics (RCP) a lot. As a source for a blog, it's phenomenal. It compiles basically every poll published anywhere in America leading up to the election. Furthermore, I'd like to note that I am not directly accusing RCP of intentional deception. I'm not exonerating them, either.  I will, however, give RCP the benefit of the doubt on this, and simply assume it for the sake of today's article that the problems with these averages is a case of "a chef is only as good as his ingredients."

But there remains a clear skew created by these polls by certain issues. The first is old polls. Let me give you a few examples. Let's start with the general election average:

Do you perchance see the issue with this CNN poll I've highlighted? It's nearly two weeks old, and dates to before the game-changing Presidential Debate on 10/4.The Politico poll isn't quite as old but all data was gathered ahead of the first Presidential Debate.  So here we have two old polls, both with Obama winning (albeit marginally) that keep Mitt Romney's lead smaller. Take those two polls out of the average, you see Mitt Romney not with a 0.8% lead, but a 2.25% lead in the average of the four polls since the debate.

Here's an average from the swing state of Pennsylvania:

So let's compare. We've got two polls from 2-3 weeks ago showing Obama with an average lead of 9.5%.  Then we have two polls from the last week with an average lead of 2.5% for Obama.  Huge swing of 7%. Yet, due to RCP's average, they have Pennsylvania as "leaning Obama." In the last week, the two polls we have show Obama's lead as within the margin of error. It's a tossup, friends. Governor Romney can win Pennsylvania.

One more old before we go into skewed:

Here's another highly important swing-state. It's considered "Leaning Obama" based on the RCP average. However, except for last week's PPP Poll, the other four polls are 2-4 weeks old.  Clearly, according to the Democrat Public Policy Polling, Obama's lead is down to 2% (within the poll's margin of error).

Now let's talk about what we've been discussing for months on Biblical Conservatism: Skewed polls.

Quick reminder: Ohio's voter identification is +1 Republican.  We Ask America's sample was +4% Democrat (5% skew to Democrats). ARG oversampled Democrats by 9% (10% skew to Democrats). CNN/Opinion Research oversampled +2% (3% total skew to Democrats). Survey USA oversampled Democrats by 4% (5% total skew to Democrats).

These are simply snapshots, friends. But the point I am making is important. The RCP Average is packed often with outdated polls and skewed polls, treated as though they were up to date or as accurate as the other polls. It creates a false impression of the reality of the situation. While Real Clear Politics is a good site, please don't get me wrong, but the average is simply not an accurate reflection of the electorate.


To my regular readers: With the Vice Presidential Debate tonight, I will be writing my reaction to that event tomorrow and publishing tomorrow's blog by about 1 pm.