A lot of people have noticed that there are numerous Iowa polls with conflicting results. However Iowa is not the only place where polls have been all over the place.
HWC, a MyDD commenter has a theory
But, less than half of the WaPost sample for the Democratic Primary consists of registered Democrats 48% versus 52% not registered Democrats.
The last time there were contested primaries in both parties (2000), the split was 65% registered Democrats and 35% not registered Democrats. It was also 62% women, 38% men.
The reason we are seeing polls all over the map in all directions is that pollster are applying different assumptions about the primary electorate. That’s why you can have two polls calling the same random people on the same three days in the same tiny state and get wildly different results. They are applying (or not applying) different weights to the sample.
Do you apply the 2004 weights (when there was, in effect, no Republican primary)? Or the 2000 weights when there were contested primaries in both parties?
This makes a lot of sense to me. It will be interesting to see if more independents will vote in the democratic or republican primary.