With all the controversy around here about phantom polls and the people around here that uncovered the statistical, um, anomalies that existed in the Research 2000 "polls", would you think Rasmussen might be a bit more careful? The answer: apparently not. There are some things about Rasmussen's numbers which give me pause as to whether they're actually running polls or simply releasing numbers that fit a pleasing narrative.
First of all, I direct you to TPM PollTracker's Florida Obama Job Approval rating...
Obama Job Approval (FL) Approve 47.0% Disapprove 54.0%
I'll put that up a second time in case you missed it...
Obama Job Approval (FL) Approve 47.0% Disapprove 54.0%
That's right. 101% of responders had an opinion on Obama's job approval. This by the way is an improvement over last week where the spread in Florida was 46-55.
Basic errors like this are why the Ras-a-palooza tends to be an object of fun around here. (This is Fark's favorite state Florida, so a possibility could be double voting by Republicans.)
Meanwhile, it's the national numbers that Rasmussen puts out every day that gives me pause that something fishy might be afoot. The approval-disapproval almost always add up to 99%, with it occasionally adding up to 98 or 100%. There are no exceptions in the last 99 daily tracking polls, where there are 7 occurrences of 98%, 64 of 99%, and 28 of 100%.
So for a grin, I went back the same amount of time (13 pages of TPM Polltracker for the approval rating) and checked Gallup's daily tracker for the same. In that case, I got 4 occurrences of 90%, 30 of 91%, 32 of 92%, 29 of 93%, 4 of 94%, and 1 of 95%.
This is one of those things that make me go hmmmm. I figure the distribution of undecideds by Gallup is honest, but the Ras distro seems really fishy to me. Any analysts better than me out there (and I know you're there) have any ideas? Are these numbers just too pat? Is this a case of a pollster doing what Research 2000 did (allegedly)? I'm just thinking that a distribution like this makes me want to ignore anything Rasmussen puts out.