Why the Polls Are Skewed Towards Obama
By Imv on June 21, 2008 at 4:15 PM in Barack Obama, Delegates, Democratic National Convention, Electability, Electoral College
Imv is my username, but you should know that I am a delegate for Sen. Hillary Clinton to August’s Democratic National Convention in Denver. I reside in “Red State America.”
The algorithms used by pollsters are beyond the vast majority of voters. Most people take polls at face value. WRONG.
Pollsters massage their numbers based on what they believe about the electorate. Are pollsters taking into effect PUMAs? I don’t believe they are for two reasons: 1) their state results skew unrealistically towards BHO and are out of sync with the national polls 2) almost all analysts believe PUMAs will return to the fold.
The only polls I trust right now are Rasmussen’s. He mentioned changing this methodology after BHO self-nominated.
[Ed. Note: See TexasDarlin's "Rasmussen: McCain Takes Florida" on June 20.]
Here’s a simple example.
Suppose the pollster believes the country is split as follows, based on voting patterns over the last few years:
D – 40, R – 30, I – 30
But, we know between a third and half of Hillary’s voters won’t support BHO. Many, like me, have left the Dem Party. If polled, I would call myself an independent. And, the Dems left will overwhelmingly support BHO because so many Hillary voters aren’t Dems anymore. See the problem? The models over-represent Dems and undercount Independents.
Let’s say, to make are example simple, that 10 out of the 40 Ds become Is. And, let’s assume, for simplicity, that all Ds are voting for BHO, all Rs are voting for JM, and the Is are split evenly. The electorate would look like this:
D – 30, R – 30, I – 40
That would give each candidate 50% each. But, the pollsters’ model has 40% Ds. So, they will poll and/or crunch their numbers accordingly. Let’s see what happens when the pollsters take our results and adjust for their bias.
O– 40, Mc – 30, I – 30 (split evenly) = O – 55, Mc – 45
The poll will show Obama with a slam-dunk (which is GREAT PR) while really the state is tied. I’ll bet even Obama’s internals are skewed like this, since they seem to be blithely assuming PUMAs will come home.
How many pollsters are using pre-primary demographics? Remember, two of the most reliable voting blocs are women and older voters = PUMAs.
I think the question is worth asking.






















