The Gallup Generic Tracking Poll – Let’s Just Flip a Coin
The last Gallup Generic Tracking Poll showed Republicans up by a historic 10 points. I meant, the last Poll prior to the one just released, which a week later shows a tie.
Any poll that swings 10 points over a week when there was no news is unreliable.
Let’s just flip coins when it comes to generics, and focus on race-by-race analysis.
My Shocking Theory On The Delaware Primary
Dems Losing The Yutes
CNN Distorts 60 Min./Vanity Fair Poll About Sarah Palin
Follow me on Twitter, Facebook, and YouTube
Donations tax deductible
to the full extent allowed by law.
Statistical noise, Professor. That's why Gallup, like other pollsters, tells you this about this poll:
"For results based on the total sample of registered voters, one can say with 95% confidence that the maximum margin of sampling error is ±4 percentage points."
So the 10-point spread could have been a 2-point spread, 95% of the time and who knows what the other 5%.
We're now seeing one generic poll after another with a GOP lead of from 6 to 13 points over the past 10 days. RCP puts the average, including Gallup's latest, at plus-6.7%. Keep your eye on that. Gallup is one of the best — but anyone can catch an outlier or two when they poll every week.
Gallup had the Dems leading by 6 points in mid-July. If you believe that, I want some of what you're smoking (or some of what Gallup's pollsters are smoking assuming they believe their own numbers).
They've been alternating between 'likely voter' and 'registered voter' this summer, which explains the jumps. They hide this fact in the fine print. You'd think they'd acknowledge it prominently as it is a huge factor swaying their results.
I see it as two equivalent outliers. The GOP is up by 5+/-1%, which on a registered voter poll is still epic. We should never be delusional enough to believe we're up by 10%. It might happen someday, but not today.
Last week: take that, libruls!!
This week: polls are stupid!