Both the Trib and a poll Rich will lead with in the morning show Judy at real low levels of voter support. What catches my eye is the effect of Rasmussen on the graph that Pollster does
Rasmussen not only shows a tighter race, but a race in which Judy is over 40%.
The difference between Rasmussen and most other polling firms is that Rasmussen adjusts for voter identification. IOW, it tracks previous voter ID and tries to use the standard breakdown to then sample voters from that sort of breakdown.
Most other polling is done allowing the sample to be random and allow Voter ID to vary.
For several reasons then in a typical election, you’ll tend to see Rasmussen not vary as much and tend to have good results especially towards the end of the race.
Where the system is weak is in a year when the typical isn’t occurring. When things are going so good or so bad that people change their Voter ID. So if normally a person identifies as Democrat, Independent or Republican, they change that self Identification because of the effect of the external events. This doesn’t happen with everyone, but you saw it somewhat in the Keyes election here where some people who were essentially Republicans stopped identifying that way because of the fiasco.
There is a lot of evidence this is happenting this year on a national scale. When it occurs and you have a system like Rasmussen, the pollster then is selecting the most hardcore party members and it tends to prop up the numbers of the party that is tanking–this year that be Republicans, though Democrats could have the same situation and did in 1994. Of course, in 1994 there were far fewer polls and little access to those that did exist.
It’ll ultimately be an empirical question as to what model is correct, but given the tendency of other polls to be agreeing with each other and not Rasmussen, the smart bet is that Rasmussen’s system won’t work as well this year. It doesn’t make them a bad pollster, it just points out that an assumption they are working on may not be true this particular time.