Odd Generic Ballot Polling Numbers

by Stephan Tawney on September 21, 2010

Michael Van Poppel reports that the latest Reuters/Ipsos poll has Democrats leading on the generic ballot by one point. Gallup has similar numbers, which is apparently a change from several weeks ago.

And yet the races themselves don’t appear to be changing for Democrats. Sharron Angle in Nevada still leads by one point. Marco Rubio continues to gain ground in Florida. Joe Miller holds a significant lead in Alaska.  And now suddenly the United States Senate race in West Virginia is competitive, whereas the Democrat held a seemingly insurmountable lead in recent weeks.

Maybe things are changing on the House side, but from what I can tell Democrats are still running away from their leaders while analysts continue to project a bloodbath for the party of higher taxes in November.

So while the generic ballot numbers have changed for the better for Democrats, the individual races have seen little movement. In fact, some previously safe seats for Democrats have become competitive. So what’s happening? I’ve not a clue. Maybe someone can clue me in.

An explanation? Gabriel Malor writes:

We’ve been watching Gallup’s notoriously unreliable polling on the generic ballot for the same reason. One week they’ve got the Republican at +10. The next week, it’s tied. Then the generic Republican +5. Now, again, it’s statistically tied. So, how does Gallup spin this: “Republicans losing strength in September compared with August.” The point being that Republicans are losing momentum. Oh, the horror, they peaked too soon.

Again, BULLHOCKEY. A series of unreliable polls does not indicate momentum or lack thereof. In fact, this series of polls doesn’t indicate anything at all except that Gallup has waited too long compared to its competitors to start polling likely voters. Gallup’s poll is based on surveys of registered voters, which in a mid-term election year are only slightly more accurate than surveys of the squirrels in my yard.

So you can have as much support from registered voters as you want, but only certain voters — likely voters — actually show up in November. And we don’t count the votes of people who don’t show up. PPP, which did the recent West Virginia poll, surveys likely voters. Gallup surveys registered. Different demographics.



Leave a Reply