Posted on 11/05/2012 5:45:49 AM PST by kevkrom
Okay... election day is tomorrow, and it's time for all us to make final predictions on the record... you know, for bragging rights.
This is specifically for the Presidential race, but if you want to lock in overall and/or specific Senate and House races for "bonus points", by all means go ahead!
Now you are being silly. RCP does not include partisan sponsored polls, period. They included PPP polls - but not ones which were paid for by partisan groups like Kos. RCP similarly did not include polls paid for by GOP campaigns, superpacs, campaign internal polls, etc. The decision to exclude was provided uniformly.
As far as setting party ID in a poll - ever look at the polls from 1996, which had Clinton leading Dole by 25 points? Zogby made a name for himself because he weighted to party id, and nailed the final result, while others were way off. The thing is, you better be right if you're going to do it.
You are also wrong when you claim Gallup weights by party id... It never has. The reason they were off is because they projected the electorate to be 76% white.
Finally, what do you think would have been accomplished by having everyone shout "the polls are right! We are going to lose!" The net result would have probably depressed turnout enough to cost the House.
RCP is wrong to do that. They throw out polls that are PAID FOR by partisan groups, not performed by partisan groups. If the pollsters get it wrong, those partisan groups don’t call them back in the future. The incentive structure is obvious that they want to get it right...for money, and should not be discounted.
But then RCP goes and includes Rasmussen and Gallup who were the outliers since the summer and just chucked em on in there without questioning methodology, weighting, demographics, nothing. That’s why although RCP did well with its final average, the averages and trends from the summer through October for states weren’t as good as 538, princeton, etc.
On the other end of the spectrum, you have HuffingtonPost who did the exact opposite of RCP but got the same results. They included ALL the polls, again like RCP, without questioning a thing and got the same results as RCP, good but mediocre compared to the mathematicians that had models.
The model builders used almost all the polls but was sure to weigh the polls in their equations based on lots of factors and that is where the magic lies.
What kind of a scientist are you and what are your credentials?
I have a niece who is a scientist but doesn't even know how to change the oil in her car......
The type that can’t show my work until I get a decent p-value or other stat tests to deal with the null hypothesis. We have a statistics dept to which we are supposed to submit our work for official validation but I do the preliminary stuff. I know enough to be dangerous and it crosses over nicely when dealing with polls and methodology. Makes for a nice hobby.
Nebulous response I know but internet and all, I’m just another current events and politics junkie like everyone else here. Actually, since about 2005 I’m more of a policy junkie. Actually reading the bills and understanding how they are implemented and affect lives is more fun for me now.
Temperature in the mid to high 50's with an 80% chance of rain beginning late in the morning....
I read this thread and it is horrible. I see the same predictions this year. Lord I hope they turn out better then what FREEPERS thought in 2012. Scary as heck.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.