Is That a Landslide I Hear Or Just My Stomach?
With each passing day a new set of polls shows up which show the Dems leading in more House seats. The latest sets from RT Strategies, Zogby/Reuters, and Survey USA are indicating a huge surge toward the Dems. These are heady days for the Democrats, with a 35-seat pickup well in range. In fact, a superficial count of the seats with Dems leading in the most recent polls actually exceeds 35.
The fly in the ointment here is that House polls are spectacularly unreliable. They have smaller samples, are more difficult to target properly (area codes don’t necessarily match with congressional districts and respondents don’t always know which district they’re in), and bounce around wildly. As an example, I give you the Ohio 2nd district, home of the vile Jean Schmidt, famous for accusing John Murtha, decorated war hero, of being a coward. Here are the results from two polling organizations:
Survey USA
9/19 497 LV Schmidt 45 Wulsin 42 Schmidt +3
10/16 555 LV Schmidt 48 Wulsin 40 Schmidt +8
10/31 618 LV Schmidt 45 Wulsin 48 Wulsin +3
RT Strategies
10/10 1003 LV Schmidt 45 Wulsin 48 Wulsin +3
10/26 972 LV Schmidt 51 Wulsin 46 Schmidt +5
Here we have two reliable pollsters with large samples which, between the second week of October and the end of October showed either an 11 point swing to Wulsin, or an 8 point swing to Schmidt.
Now we could say that with a 4 point margin of error, the 10/16 poll is just a statistical accident (1 of 20 polls are outside the MoE) and the others all reflect a race where Schmidt is up by 1 point, with the results all falling within (barely) the margin of error. And we might be right about that – the danger is in looking for momentum or shifting tides from any series of polls.
Still and all, the Republicans can take little solace from the data that has been released this week. The only thing they may take away from it is the ability to spin a 20-seat loss by talking about how “everyone was saying we were going to lose 35, 40, even 50 seats.” I can live with that.
The fly in the ointment here is that House polls are spectacularly unreliable. They have smaller samples, are more difficult to target properly (area codes don’t necessarily match with congressional districts and respondents don’t always know which district they’re in), and bounce around wildly. As an example, I give you the Ohio 2nd district, home of the vile Jean Schmidt, famous for accusing John Murtha, decorated war hero, of being a coward. Here are the results from two polling organizations:
Survey USA
9/19 497 LV Schmidt 45 Wulsin 42 Schmidt +3
10/16 555 LV Schmidt 48 Wulsin 40 Schmidt +8
10/31 618 LV Schmidt 45 Wulsin 48 Wulsin +3
RT Strategies
10/10 1003 LV Schmidt 45 Wulsin 48 Wulsin +3
10/26 972 LV Schmidt 51 Wulsin 46 Schmidt +5
Here we have two reliable pollsters with large samples which, between the second week of October and the end of October showed either an 11 point swing to Wulsin, or an 8 point swing to Schmidt.
Now we could say that with a 4 point margin of error, the 10/16 poll is just a statistical accident (1 of 20 polls are outside the MoE) and the others all reflect a race where Schmidt is up by 1 point, with the results all falling within (barely) the margin of error. And we might be right about that – the danger is in looking for momentum or shifting tides from any series of polls.
Still and all, the Republicans can take little solace from the data that has been released this week. The only thing they may take away from it is the ability to spin a 20-seat loss by talking about how “everyone was saying we were going to lose 35, 40, even 50 seats.” I can live with that.
posted by Barry Rubinowitz at 10:28 AM
0 Comments:
Post a Comment
<< Home