Something odd seems to be happening in US public opinion. Pollsters are shifting from polls of registered voters to polls of likely voters. Given demographics and the enthusiasm gap, I expected the Republican lead in the House generic ballot to increase. Until recently, that seemed to be happening right on schedule.
Since then there have been several polls showing a surprising close race. If I toss Rasmussen polls out (because I don’t like their results but at least I admit it) and set sensitivity to high so new polls count more I get a pollster smoothed average showing a the Democrats ahead ! * This is mostly determined by ten polls in September six of which are polls of likely voters and five of which are polls of registered voters.
I’m pretty sure I’m over interpreting data and seeing what I want to see. I guess it’s conceivable that recent Tea Party success in primaries have scared voters.
*this is a link to a graph which will be updated. For some reason, when I try to dowload the graph I get the smoothed average including Rasmussen.