Site Meter

Tuesday, September 21, 2010

Rasmussen and the Generic Congressional Ballot in September 2010

I have been doing some arithmetic concerning the generic congressional ballot in my head. Caveat lector. I looked at polls entirely in September 2010 and took simple averages.

There were five non-Rasmussen likely voters polls. The Republicans were ahead on average by 3.4%

There were 8 registered voter polls with Republicans ahead by 1.5 on average.

There were two Rasmussen polls with Republicans ahead by 9.5 on average.

The Rasmussen not-Rasmussen difference dwarfs the likely voter registered voter difference.

OK there were two polls from the first two days in September. since then the Democrats are 0.67 ahead out of 6 RV polls. There were also two LVpolls with huge Republican leads which overlapped August and September so the average would be 5.3 %

That is, depending on exactly when I cut, the LV-RV difference can be maybe as high as 4%. This does not correspond to all the talk about the enthusiasm gap does it ?

and the Rasmussen.other gap is huge.

This is the Rasmussen only smoothed graph (you will note it is pretty smooth to begin with as it is weekly averages of a tracking poll which therefore has large samples).

Oh rats, that's missing the curve.

Non Rasmussen LV polls give a smoothed average of Republicans by 3.7 :R 45.4 D 41.7

With high sensitivity (less smoothing) Republicans by 0.9%.

No comments: