Today's not a proud day on TPM Polltracker, and I'm simply annoyed that Talking Points Memo, which I otherwise love, has not yet done anything about the House of Ras manipulating their poll tracking numbers by flooding their data with suspicious data.
Here's today's US Cong. Generic Ballot, from TPM's poll averaging:
And here it is without Rasmussen:
Now I can understand including Rasmussen in your polling, that makes sense, but when you include every poll they do -- which they do precisely to manipulate data like this -- then it's pretty clear you're letting them own you.
For some reason, the Obama approval graph isn't showing the trend lines or the numbers, but without them, it becomes clear that Rasmussen is just on another planet.
With House of Ras:
And without:
This isn't as extreme, but it's still very, very bad.
I don't think nearly enough scrutiny has been brought to bear on the House of Ras -- even though a lot of scrutiny has, in fact, been brought to bear. For the life of me, I can't think of a good reason why an unbiased pollster would wind up with disapproval numbers so consistently higher than everyone else that they make their own freaking line on the graph. Look at the first graph -- there are separate trend lines for the disapproval numbers, one for Ras, and one for everyone else. In the second graph, it's just the lower line. Please explain to me how this does not scream cooked numbers.
At the very least -- the very least! -- the suspiciousness of these numbers should clue TPM in to the fact that by counting every Ras poll they are allowing a biased pollster to distort the numbers. They're simply false, and if you knowingly distribute numbers you know to be false, then you're part of the problem.
I'd simply like to see TPM limit the number of polls they include from any particular organization.