Pollster.com aggregates a multitude of polls and averages them out to show a composite view of the election.
Their composite average of the match-up between Obama and Romney currently puts Obama up by 2.9%.
This is the graph if we include all pollster's results (including Rasmussen):
This is the graph if we exclude Rasmussen:
This is the graph if we exclude all of PPP's polling (and the DKos results):
It's clear that Rasmussen's polling reduces Obama's numbers by nearly a point (0.9%), and artificially inflates Romney's numbers by the same amount.
PPP's polling, however, inflates Obama by a mere 0.1% and actually inflates Romney by 0.2%!
Yet, Republicans (and the pundits who routinely shill for them) dismiss PPP as some kind of radical, left-leaning pollster who deliberately suppresses Republican support in their results. This simple comparison shows that PPP favours the Republican front-runner, however slightly.
Let's do this same comparison for polls since March.
First, all pollsters:
Now, let's take Rasmussen out of the mix for the same period:
Next, let's put Rasmussen back in and take out all PPP (and the DKos) results from the mix:
Again, removing Rasmussen adds nearly a full point (0.9%) to Obama's support, while reducing Romney's support by nearly a full point (0.9%). This is a net bias of nearly two points in the Republicans' favour produced by Rasmussen's skewed polling.
And again, removing PPP hardly changes anything, because PPP isn't an outlier. Including PPP in polling averages gives Romney a false 0.2% edge, while giving Obama a mere 0.1% false edge. PPP actually gives a net 0.1% bias in the Republicans' favour.
Consider also that Rasmussen automatically gives Republicans a 3.2% boost in his figures due to his "likely voter" screen. Do we see a 3.2% difference between "likely voters" and registered voters in these results? No, we don't.
-------------------------
Pollster ratings from fivethirtyeight.com, based on 2010 elections:
No comments:
Post a Comment