Media Polls Are Warping RealClearPolitics Poll Average
Polls are more art than science. Countless variables can shape the composition of any individual poll. To smooth out the noise from individual polls it is best to look at an average of them. RealClearPolitics has the most cited average of recent polls and forms the basis of most pundits and reporters assessment of the presidential race. But, as the old saying goes, "Trash in, trash out." The near-epidemic of skewed media polls is warping the RCP average in a way that is very favorable to Obama.
Today's RCP average shows Obama with a 4.4 point lead nationally. The average is based on nine recently released polls. Five of these, four of which show outsized leads for Obama, have serious partisan skews favoring Democrats. The Pew "poll," which shows Obama up 10, has a D+19 skew. FoxNews Poll is D+9, NBC is D+9 and CBS is D+7. The CNN poll, which I discussed today, seems to have a D+10 skew. None of these partisan breakdowns is an accurate prediction of what the electorate will look like in November, yet they do impact the current RCP Average.
The four polls with more reasonable assumptions of partisan breakdown show a much closer race. Still, only two of these poll "likely voters"--Rasmussen and Democracy Corps. Registered voter polls inherently give Democrats a 2-3 point edge in results. Factor all these variables in and the race is probably a wash, with the probability of a slight Romney edge.
And yet, most pundits and analysts are taking the current average and declaring that Obama is clearly in command of the race. That may in fact be the case, but when the average is impacted by a number of highly skewed polls, its not a conclusion on which I would wager a lot of money.
One of my colleagues has floated the idea that the media are using such highly skewed polls to intentionally game the RCP average. But, if you look at their business models or ratings, they don't really seem cable of lots of long-term thinking. It probably just reflects their built-in partisan bias more than anything else.
Now, I love the RCP average and it is a very effective tool for assessing the current state of the campaign. But, when the average is built on a significant amount of crap being fed into it...well, I would just take the outcome with a mountain of salt.
Follow me on twitter