Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
Nate Silver has an interesting column on robo calling polls like PPP and Rasmussen compared to live interviews. Robo polls are not allowed to call cell phones which excludes almost 30 to 40% of the population, and those tend to be younger more liberal voters. This explains why Rasmussen and the CBS/Times poll might have a 7 point gap..
Excellent article even though some of the conservatives here will not like what it has to say..
Silver is ignoring that many of those ~+6 Obama polls are oversampling Democrats based on 2008 numbers and registered Democrat numbers.
CNN's poll that had Obama +6 had Independents at just 5%, which is pure ridiculous, especially since CNN has consistently had independents favoring Romney by double digits and evidence suggest that independents make up about 20 to 25% of the nation.
Also, why is Silver putting a lot of credence in a Wisconsin poll that is not only +11 to the Democrats (mind you, in a state where the Republicans have done well in 2010 and 2012), over samples the Milwaukee area by about 40%. Does he actually delve into these, or does he just look at the superficial face value for the most part?
P.S. My 91 year old grandpa has a cell phone.
Last edited by michiganmoon; 09-22-2012 at 09:22 PM..
Silver is ignoring that many of those ~+6 Obama polls are oversampling Democrats based on 2008 numbers and registered Democrat numbers.
CNN's poll that had Obama +6 had Independents at just 5%, which is pure ridiculous, especially since CNN has consistently had independents favoring Romney by double digits and evidence suggest that independents make up about 20 to 25% of the nation.
Also, why is Silver putting a lot of credence in a Wisconsin poll that is not only +11 to the Democrats (mind you, in a state where the Republicans have done well in 2010 and 2012), over samples the Milwaukee area by about 40%. Does he actually delve into these, or does he just look at the superficial face value for the most part?
P.S. My 91 year old grandpa has a cell phone.
Here is a decent article explaining some of his methodology.
Nate was within a single electoral vote in 2008 and something like 98% accuracy in 2010. I'd say he knows what he is doing.
Not true.
#1 His final analysis for 2008 was made midday on election day. He got Indiana and part of Nebraska wrong. Is this that impressive considering that only 3 states had polls within the RCP avg. margin of error on election day? Wow, he got all of the states right that had a candidate with a 4 point lead or more on election day and then went 2 for 3 where the polls said it was too close to call...plus he got the split in Nebraska wrong. Wow. How amazing. I am mesmerized.
Nate Silver openly is an Obama fan. In fact, he started analyzing polls, because he thought the polls were being 'reported' unfairly to hide bias to favor Hillary over Obama. There is a certain danger when an analyst goes in with a mission to help a candidate.
If Nate Silver cares about bias, why does he ignore when a Wisconsin poll has a +11 Democrat bias and polls 40% more in the Milwaukee area than they should have? In fact he weights it more than a poll with only a +3 Democrat bias (a realistic margin for Wisconsin).
Nate Silver did not impress me at all in the GOP primary, sure he got the obvious ones right and made a few poor picks as well.
#1 His final analysis for 2008 was made midday on election day. He got Indiana and part of Nebraska wrong. Is this that impressive considering that only 3 states had polls within the RCP avg. margin of error on election day? Wow, he got all of the states right that had a candidate with a 4 point lead or more on election day and then went 2 for 3 where the polls said it was too close to call...plus he got the split in Nebraska wrong. Wow. How amazing. I am mesmerized.
You're right. It was one state and not one EV. My memory failed me. Still, that is an impressive feat in itself, especially when viewed in the context of Congressional election results.
Quote:
Nate Silver openly is an Obama fan. In fact, he started analyzing polls, because he thought the polls were being 'reported' unfairly to hide bias to favor Hillary over Obama. There is a certain danger when an analyst goes in with a mission to help a candidate.
Which has absolutely, positively, zero to do with his analysis.
Quote:
If Nate Silver cares about bias, why does he ignore when a Wisconsin poll has a +11 Democrat bias and polls 40% more in the Milwaukee area than they should have? In fact he weights it more than a poll with only a +3 Democrat bias (a realistic margin for Wisconsin).
I haven't read anything about the polls you are referring to so I cannot comment with any degree of certainty on it. That said, I'm willing to bet that Nate has a rather straightforward explanation for his reasons if you look for it.
Quote:
Nate Silver did not impress me at all in the GOP primary, sure he got the obvious ones right and made a few poor picks as well.
He impressed the NYT enough for them to offer him a very sweet gig, and impressed the Obama campaign so much that they fed him their internal polling data to check their accuracy. And this was after a pretty successful career analyzing baseball metrics.
You're right. It was one state and not one EV. My memory failed me. Still, that is an impressive feat in itself, especially when viewed in the context of Congressional election results.
Which has absolutely, positively, zero to do with his analysis.
I haven't read anything about the polls you are referring to so I cannot comment with any degree of certainty on it. That said, I'm willing to bet that Nate has a rather straightforward explanation for his reasons if you look for it.
He impressed the NYT enough for them to offer him a very sweet gig, and impressed the Obama campaign so much that they fed him their internal polling data to check their accuracy. And this was after a pretty successful career analyzing baseball metrics.
I will repeat the following, since you seem to miss it.
There are 50 states with 2 splits (Maine and Nebraska). It is not amazing to on election day look at the polls and say all of the states that have a candidate up by 4 or more will win. People predicting on election day had only 4 things where the polls didn't have one candidate up by 4 or more points - Indiana, Missouri, North Carolina, and the Nebraska split. Of these...Silver got 2 wrong.
You are literally bragging that Silver had a 50% accuracy in predicting results where a candidate had a lead smaller than 4% in the polls on 2008 election day.
The same with the 2010 House of Reps...only so many weren't going to big wins. Silver had many horrible predictions in the GOP primary.
Nate Silver does not account for the fact that accuracy of the polls change cyclically - in regards to that it could be more accurate this time around.
Nate Silver does not account for the actual polling data. He doesn't care if a poll has +11 to Democrats in a relatively even state. He doesn't care if a poll has only 5% independents when the national average shows about 5 times that. He doesn't care if a poll over samples 2008 Obama supporters so that people in the poll gave Obama a 15% edge over McCain (real life it was 7%) way back in 2008 for the 2012 poll..... In fact he gave these polls more weight since they call cell phones!!! This is a flaw.
P.S. With all of the 2008 magic and Obama love, it is not surprising that one Obama lover - the New York Times - would fall in love with another Obama lover - Nate Silver. And it does speak to the motives of Silver when he literally started analyzing polls to help Obama beat Hillary by trying to show the problems with not using cell phones for collecting data, which is why it is amazing that he doesn't care what the data is in regards to oversampling groups by any means other than cell phones.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
