OCinBuffalo Posted November 3, 2012 Share Posted November 3, 2012 (edited) http://fivethirtyeig...tically-biased/ Read the whole article if you like, but if you've paid attention to me at all, you already know what Nate Silver is saying, and why. Ask yourself: how it is that I was able to predict Silver saying this, so easily? Why is Silver saying this today...and not 3 months ago? Is this any more true today, than it would have been 3 months ago? If we assume Nate Silver is "brilliant", then doesn't it follow that he would see the flaw in his own work, immediately react, and start the disclaimer process, or, fix his model, as soon as he sensed there might be a problem? Why is he reacting...now? I have a theory as to why: http://www.aei-ideas...uld-sweep-ohio/ Hey Nate? Feel your sphincter tightening? Yeah Nate, if you are only using top line #s, and ignoring the internals of the polls, and the obvious data patterns in them, then...yeah...if there are problems with those top line #s, you are screwed. Nate has, in politics, essentially ignored things he would never have if this were baseball. You will not find that in this article...which is essentially a very long disclaimer. His "house effect" adjustment doesn't take into account what is happening THIS year. THIS is the year we are seeing some real nonsense in the polls: again, the internals not matching the top line, the effect of Independents, etc. Thus, he has NOT accounted for bias. He simply thinks/says he has. Had he been truly objective, he would have seen the problem with the internals and adjusted for THIS year's problems. Yes, Nate, this certainly is unlikely to reflect sampling errors alone. Thank you Cpt. Obvious. It is rather, much more likely to reflect the awful WEIGHTING methods in the polls you use, or the Axelrod pushed demographic model. Silver assumes that there's nothing wrong with the methods used to produce these polls, whose weighting produces bizarre internals in other areas(again, Romney leading independents by 10+, Party ID of D+6, etc, Romney better on economy by an average of +6.). Instead, he uses a strawman = sampling error. Nate Silver needs to ask himself the screaming question: why is he relying on polls that have a D+6 or higher turnout bias, when we know that is a fantasy, and when he himself says that people vote by party ID almost exclusively, and, he himself says races depend on Independents? If you want an in depth(and I mean it) explanation of this...look no further than another baseball guy who does politics too, Baseball Crank. http://baseballcrank...012/10/post.php And, you guys think I write long posts. If you don't want to read all of that, just pay attention to this: Nate Silver's much-celebrated model is, like other poll averages, based simply on analyzing the toplines of public polls. This, more than any other factor, is where he and I part company. If you read only the toplines of polls - the single number that says something like "Romney 48, Obama 47" - you would get the impression from a great many polls that this is a very tight race nationally, in which Obama has a steady lead in key swing states. In an ordinary year, the toplines of the polls eventually converge around the final result - but this year, there seems to be some stubborn splits among the poll toplines that reflect the pollsters' struggles to come to agreement on who is going to vote. Look for excuses #2 and #3...coming soon to Nate Silver's blog. Again, none of this says Obama = lose. What it does say? We can read Silver like a book, and, Nate himself has now exposed a significant crack in his model. Edited November 3, 2012 by OCinBuffalo Link to comment Share on other sites More sharing options...
3rdnlng Posted November 3, 2012 Share Posted November 3, 2012 I don't even agree that you understand what Nate Silver is saying. So how can I agree that you predicted anything that he said? Actually your agreement is probably the last thing he wants. Link to comment Share on other sites More sharing options...
OCinBuffalo Posted November 3, 2012 Author Share Posted November 3, 2012 (edited) I don't even agree that you understand what Nate Silver is saying. So how can I agree that you predicted anything that he said? Yeah...and that's actually, as 3rdnlng says, encouraging. Show me 1(one) analysis that says things are going well for Obama in early voting in general, or in any battleground state. Oh, wait...this is conner, so we need a definition of the word "well", otherwise he will try to say that Obama going from 6% in FL to 2% is "doing well, because he is still winning". Obama doing "well" = winning by the same large margins in EARLY VOTING that wiped out his ELECTION DAY losses in 2008. If this thing comes down to election day voting, with only minor leads for Obama in every battleground? Look out. (Hint: I know the answer to the above...but I'm not going to tell you conner. Do your own work. And, that state is so weird, in terms of Inds, crushing their early voting #s from 2008, that nobody really knows what's going on there) I know what Nate Silver is saying: Silver is simply a bookie. Nothing more. However, a real bookie would adjust his odds the second he gets any new info: like these early voting #s, because he wants to win. Silver hasn't, because doing so puts his model in question, and that's not how he "wins". The model is more important to Silver than the outcome, as Excuse #3(soon to come) will show. No. Silver would much rather keep his model untouched, because his model is the juice here. If the outcome is bad? Silver will blame the polls, something he has "nothing to do wtih" . He just "baked in" that excuse, right after the early voting #s come in ...and you think...that's a coincidence? Edited November 3, 2012 by OCinBuffalo Link to comment Share on other sites More sharing options...
John Adams Posted November 4, 2012 Share Posted November 4, 2012 (edited) Who's going to win OC? What's your electoral prediction? I just did that 270 to win site and predict 311 to 227. Gave Obama Florida and Ohio. Edited November 4, 2012 by John Adams Link to comment Share on other sites More sharing options...
/dev/null Posted November 4, 2012 Share Posted November 4, 2012 I've run the 270towin every day for awhile now and I always end up with the same map. Obama 281-257 http://www.270towin.com/2012_election_predictions.php?mapid=bgUQ Unless Ohio flips then it's Romney 275-263 http://www.270towin.com/2012_election_predictions.php?mapid=bgUT Also possible, but insignificant to either outcome, that Maine's funky single vote goes to Romney Link to comment Share on other sites More sharing options...
IDBillzFan Posted November 4, 2012 Share Posted November 4, 2012 Who's going to win OC? What's your electoral prediction? I just did that 270 to win site and predict 311 to 227. Gave Obama Florida and Ohio. Florida is going to Romney. Even Obama has stopped wasting time and money there. And Ohio? Independents are trending to Mitt like crazy. Still a jump ball, but it's absolutely in play. Link to comment Share on other sites More sharing options...
OCinBuffalo Posted November 4, 2012 Author Share Posted November 4, 2012 (edited) Who's going to win OC? What's your electoral prediction? I just did that 270 to win site and predict 311 to 227. Gave Obama Florida and Ohio. That's the part of this that is so frustrating. If I had any confidence in the data I've seen, I'd make a prediction that I thought was serious. I understand what you are asking, and more importantly: why. Right now, I'd rather give a range: anywhere from O 281-R 257(upper limit if Axelrod's demo model is right)...to O 207-R 331(upper limit if Gallup's demographic model, and their 9000+ sample, likely voter, party ID poll is right) is reasonable, and can be supported, given the data we have. Obama ain't winning FL. You want an example of late game desperation...Obama wins FL, or NC, is it. The difference is: Gallup's model is supported by public data we can see. We don't know the rationale behind Axelrod's model, because it is based on internal data we can't see. It doesn't mean Axelrod can't be right. All it means is: most of the stuff we have, that is in the public domain, says he's wrong. There's simply too many inconsistencies. How can any serious person say that independents don't matter , and that this election is D+6 or higher? However, if their demographic model is right, D turnout may, for the first time in a VERY long time, surpass the Is and Rs, and this may be a close Obama win. Put it this way: if this is work? I'd advise the client that their data is messed up, and to not make any major decisions with it until it can be fixed. At this point, I fear any serious prediction I might make would be more about wishful thinking, than data. I'm not a liberal, I don't try to pass off my wishful thinking as fact. My problem is: with the data we have, it looks like that's exactly what the liberals are doing, again. Edited November 4, 2012 by OCinBuffalo Link to comment Share on other sites More sharing options...
B-Large Posted November 4, 2012 Share Posted November 4, 2012 http://fivethirtyeig...tically-biased/ Read the whole article if you like, but if you've paid attention to me at all, you already know what Nate Silver is saying, and why. Ask yourself: how it is that I was able to predict Silver saying this, so easily? Why is Silver saying this today...and not 3 months ago? Is this any more true today, than it would have been 3 months ago? If we assume Nate Silver is "brilliant", then doesn't it follow that he would see the flaw in his own work, immediately react, and start the disclaimer process, or, fix his model, as soon as he sensed there might be a problem? Why is he reacting...now? I have a theory as to why: http://www.aei-ideas...uld-sweep-ohio/ Hey Nate? Feel your sphincter tightening? Yeah Nate, if you are only using top line #s, and ignoring the internals of the polls, and the obvious data patterns in them, then...yeah...if there are problems with those top line #s, you are screwed. Nate has, in politics, essentially ignored things he would never have if this were baseball. You will not find that in this article...which is essentially a very long disclaimer. His "house effect" adjustment doesn't take into account what is happening THIS year. THIS is the year we are seeing some real nonsense in the polls: again, the internals not matching the top line, the effect of Independents, etc. Thus, he has NOT accounted for bias. He simply thinks/says he has. Had he been truly objective, he would have seen the problem with the internals and adjusted for THIS year's problems. Yes, Nate, this certainly is unlikely to reflect sampling errors alone. Thank you Cpt. Obvious. It is rather, much more likely to reflect the awful WEIGHTING methods in the polls you use, or the Axelrod pushed demographic model. Silver assumes that there's nothing wrong with the methods used to produce these polls, whose weighting produces bizarre internals in other areas(again, Romney leading independents by 10+, Party ID of D+6, etc, Romney better on economy by an average of +6.). Instead, he uses a strawman = sampling error. Nate Silver needs to ask himself the screaming question: why is he relying on polls that have a D+6 or higher turnout bias, when we know that is a fantasy, and when he himself says that people vote by party ID almost exclusively, and, he himself says races depend on Independents? If you want an in depth(and I mean it) explanation of this...look no further than another baseball guy who does politics too, Baseball Crank. http://baseballcrank...012/10/post.php And, you guys think I write long posts. If you don't want to read all of that, just pay attention to this: Look for excuses #2 and #3...coming soon to Nate Silver's blog. Again, none of this says Obama = lose. What it does say? We can read Silver like a book, and, Nate himself has now exposed a significant crack in his model. No model is prefect, but until proven otherwise in results, Silvers predictions are pretty solid.... Link to comment Share on other sites More sharing options...
OCinBuffalo Posted November 4, 2012 Author Share Posted November 4, 2012 No model is prefect, but until proven otherwise in results, Silvers predictions are pretty solid.... See, I'm not really saying that. Silver's model, and certainly much of this thinking, is appropriate. There are rational approaches. But, there are problems. And, the biggest one, that Nate himself has now identified, is that he is 100% reliant on the top line of polls. It's a dependency. Period. Nate's model is only as accurate as the polls. Those polls are only as accurate as the demo model they use, and the weighting methods that are applied to the raw data that is collected, to map the raw data to that demo model. It's just a long string of dependencies, that no, in fact, Silver has NOT corrected for with his "house affect" adjustment. That adjustment doesn't account for Axelrod's demo assumptions being off to the extent that the white voter electorate is 75%...and not 72% like he is saying, and MOST of the pollsters using Axelrod's assumptions. The "house affect" only deals with the individual pollsters, it does nothing if many, if not all of them are using a broken demo model. IF this is what has happened here, and Gallup is right instead? Then Silver is completely screwed. The thing is: in that scenario it actually isn't Silver's fault. How was he supposed to factor in everyone using a broken model? If anything his 80% prediction, and a Romney win...would be the single biggest confirmation of just how broken Axelrod's demo model was. It will be clear as day. Don't think Silver won't remind you of that on Nov. 7th. Link to comment Share on other sites More sharing options...
Recommended Posts