MOBILE USERS: m.isthmus.com
Connect with Isthmus:         Newsletters 
Friday, November 28, 2014 |  Madison, WI: 16.0° F  Light Snow
Collapse Photo Bar

Nate Silver's Flawed Model

Races for the Senate, U.S. House, etc. and other issues of national importance.

Re: Nate Silver's Flawed Model

Postby Steve Vokers » Sat Nov 03, 2012 2:02 pm

Meade wrote:
Steve Vokers wrote:If Romney gets 359 electoral votes, I'll eat my shorts. Hell, if he gets 359 I'll eat Meade's shorts.

Great! I'll be sure to leave a giant dump in them.


Yea though I walk through the valley of the shadow of death, I don't fear that evil because there is not one chance in hell that Mitt is getting anywhere near 359.
Steve Vokers
Forum God/Goddess
 
Posts: 1165
Joined: Tue Oct 07, 2003 11:58 am

Re: Nate Silver's Flawed Model

Postby Bludgeon » Sat Nov 03, 2012 5:45 pm

Huckleby wrote:I saw Karl Rove on FOX yesterday with his little white board, explaining how the shift to Republicans in Ohio early voting meant Obama was doomed in Ohio. Hell, maybe he is right, who really knows?

He made a side comment that caught my ear: Real Clear Politics is democratic leaning because they incorporate a lot of flawed polls that oversample Dems. That leads me to believe that RCP must be on the level.

Rove might just be spewing propoganda that he knows is wrong in both cases. But I tend to think he is a sincere man who values his reputation as an analyst. I didn't say he is right, or personally likable. He is a very smart guy who has introduced a lot of the voter-wrangling techniques that have been widely adopted.


By now, everybody knows the argument: 1) Democratic leaning polls are operating under the assumption that the 2012 election will mirror or increase the Democratic advantage of 2008; but 2) 2008 is a terrible model to predict the 2012 turnout; and 3) the turnout composition of the electorate is the key question about what is going to happen on election day.

I'm sure Huckleby would be the first to admit the following: 2008 was a horrible year for Republicans - even they were sick of Bush. Obama got his share of the vote because 1) Deomocratic enthusiasm was at a 30 year high, 2) Republican enthusaism was at a 30 year low, and independents broke for Obama by 7 points or so.

Like Huckleby, I'm not saying it means Republicans will win but 1) this year Romney has a 7 point or so edge with independents, 2) moderate Democrats equal a measurable decrease in Democratic enthusiasm, and 3) Republicans are more enthusiastic. According to Pew and Gallup there is no Democratic advantage party v. party, which means it's up to the independents. According to almost every poll in the nation, Romney leads with independents.

To offset Romney's partisan tie + independent lead, Obama needs to win by grinding up turnout. Unlike Huckleby, I will go so far as to say that when a campaign faces that proposition, that's not a strategically favorable position to be in. Look at voter's chief concern - unhappiness about the economy, disatisfaction with the president's economic policies - it begs the question of just how it is in this environment we are to reliably expect the incumbent team to win the turnout game on election day? (Much less win it by a margin that offsets the independents).

The "state polls are right" argument rests on this very assumption: D+5-D+9, vs D+0/R+0. Such is the case in Ohio.

I've only searched casually, but I'm not aware of a modern election where the person who wins the independents has not gone on to win the white house. First time for everything?
Bludgeon
Forum God/Goddess
 
Posts: 1291
Joined: Tue Apr 06, 2010 2:27 am

Re: Nate Silver's Flawed Model

Postby talagaster » Sat Nov 03, 2012 6:48 pm

Bludgeon wrote:I've only searched casually, but I'm not aware of a modern election where the person who wins the independents has not gone on to win the white house. First time for everything?


http://www.ropercenter.uconn.edu/electi ... JWr-Yb2ZMg

In the distant past of 2004, the independents went with the challenger from Massachusetts over the politically divisive incumbent. Didn't change much.

http://www.ropercenter.uconn.edu/electi ... JWs1Ib2ZMg

In 2000, independents went with Bush and, while he became President, he lost the popular vote.

Independents are important but turning out the base has mattered more in the last few elections.
talagaster
Forum Addict
 
Posts: 204
Joined: Fri Jun 12, 2009 12:20 pm

Re: Nate Silver's Flawed Model

Postby Huckleby » Sat Nov 03, 2012 8:16 pm

Bludgeon wrote:By now, everybody knows the argument: 1) Democratic leaning polls are operating under the assumption that the 2012 election will mirror or increase the Democratic advantage of 2008

This is what some Republicans keep saying but I doubt it is true.
Some reputable analysts from conservative side say this theory is fantasy.

None of us, including the Karl Rove's of this world, have access to the inside workings of the various polls. Pollsters are in the business of being right, not of biasing expectations, unless you believe a broad conspiracy is at play this time around.

Most right-wing mouthpieces, including Rove, have mostly backed-off of the biased polls theory. Rove made his comment as an aside to create a little FUD (fear uncertainty and doubt) around unpleasant RCP trends. The right is mostly working the enthusiasm angle now. But you also have to look at the infrastructure for getting-vote out, and Dems have big advantage there in Ohio.

I'm out of speculation mode, have boarded-up all my windows with plywood, and am hunkering-down for Tuesday, 10PM when I turn-on TV for results.
Huckleby
Forum God/Goddess
 
Posts: 7043
Joined: Thu Oct 20, 2005 5:12 pm
Location: parents' basement

Re: Nate Silver's Flawed Model

Postby pjbogart » Sun Nov 04, 2012 10:27 am

Silver is currently portraying Obama's win as virtually locked, but he keeps talking about the election like we're simply flipping a coin (and uses that analogy). I think at this point in the campaign, the final few days, it's not a matter of luck, it's a matter of energy. If the polls are tied up in Florida then the winner will be whichever team does a better job of getting their voters to the polls. I've always had the feeling that Republicans hold an edge on this, not because their ground game or door knocking is superior but because their voters are naturally a bit more reliable. The cynic in me says this is because Republicans like to scare up votes with conspiracy theories and doomsday scenarios, but when I show up to the polls I mostly see an older set of voters waiting patiently to cast their ballots, not dewy-eyed 20 year-olds yearning for hope and change.

So it's not a coin flip. It's muscle. It's knocking on doors, volunteering to be a voter taxi, manning a phone bank. It's not two days before the election, it's three. You have to count Tuesday too. Push, pull or drag people to the polls. Don't bother with the reliable voters, go find some unreliable voters. Go to the Labor Temple on South Park and see where you're needed. Put on a comfortable pair of sneakers and make it happen.
pjbogart
Forum God/Goddess
 
Posts: 6186
Joined: Thu Sep 04, 2003 4:57 pm

Re: Nate Silver's Flawed Model

Postby Huckleby » Sun Nov 04, 2012 11:22 am

pjbogart wrote:Silver is currently portraying Obama's win as virtually locked, but he keeps talking about the election like we're simply flipping a coin (and uses that analogy). I think at this point in the campaign, the final few days, it's not a matter of luck, it's a matter of energy.


You're absolutely right, it's a fight, and Nate Silver's stats don't win anything. The election is close and it is yet to be determined which side has more fight in them.

Silver made a point in his Charlie Rose interview that is obvious but worth repeating. Silver is a baseball stats freak, no suprise there, and he emphasizes that 20% chance of winning is pretty good. If a batter had a 20% chance of hitting a home run every time he stepped up to the plate, you'd be on the edge of your chair every time he came up to bat. Silver is not saying that Obama is a lock, Romney may very well hit it out of the park.
Huckleby
Forum God/Goddess
 
Posts: 7043
Joined: Thu Oct 20, 2005 5:12 pm
Location: parents' basement

Re: Nate Silver's Flawed Model

Postby fennel » Sun Nov 04, 2012 11:27 am

Here is an earlier "left-leaning post" by Silver.
fennel
Forum God/Goddess
 
Posts: 3218
Joined: Thu Sep 28, 2006 4:24 pm
Location: Inside the Green Zone, Madison

Re: Nate Silver's Flawed Model

Postby Bludgeon » Sun Nov 04, 2012 5:22 pm

talagaster wrote:
Bludgeon wrote:In the distant past of 2004, the independents went with the challenger from Massachusetts over the politically divisive incumbent. Didn't change much.


Thanks for the google work. 2004: Democrats and Republicans tied at 37%/37% party ID; independents at 26%; Kerry advantage = +2%/independents. This small advantage is offset because Bush won 11% of Democrats' votes in 2004 (vs. Kerry's 6% of Republicans). So Bush won, because Kerry's 2% blip with independents didn't even account for turnout in a year when Republicans and Democrats were tied for turnout but Kerry lost voters from his own party.

The #1 concern for 2004 voters was national security, this is an issue Bush held a significant, consistent lead on in the polls. This year the #1 concern for voters is the economy; an issue Romney has a consistent lead on in the polls, at the same time boasting a strong and constant lead with independents. If Kerry led on 2004's crucial issue and boasted 5% more of the independent vote, he probably would have won.
Bludgeon
Forum God/Goddess
 
Posts: 1291
Joined: Tue Apr 06, 2010 2:27 am

Re: Nate Silver's Flawed Model

Postby Bludgeon » Sun Nov 04, 2012 5:36 pm

Huckleby wrote:Most right-wing mouthpieces, including Rove, have mostly backed-off of the biased polls theory. Rove made his comment as an aside to create a little FUD (fear uncertainty and doubt) around unpleasant RCP trends. The right is mostly working the enthusiasm angle now. But you also have to look at the infrastructure for getting-vote out, and Dems have big advantage there in Ohio.


While there is a bias, I'm not saying (all) the media polls are producing D+5-D+9 turnout out of an intentional wish to misrepresent; though you could argue that the purpose of certain organizations (Time/etc) is to generate push-polls that kept Obama in the black on RCP; university polls in particular.

In the case of PPP (I like James Carville), ABC/Wapo, CNN, I would describe it as a mix of professional optimism and a bit of bewilderment relating to the 2010 Census. It's a moral issue of some kind on the part of media organizations, that the white proportion of the voting population should not increase from 2008.

Nonetheless we must admit that even D+5 is uncommonly generous to the president in polling, as is the idea of a 72%-74% white share of the population; as is (example) Pew Research's Obama +3 poll today that shows Mitt Romney winning less of the white (54%) vote than John McCain did. Give Romney only 54% of the white vote, give whites only 74% of the electorate, and give Democrats a 4% party ID edge (in this poll), and sure Obama can be ahead by 3 points nationally. This set of assumptions is, however, a kingly gift.
Bludgeon
Forum God/Goddess
 
Posts: 1291
Joined: Tue Apr 06, 2010 2:27 am

Re: Nate Silver's Flawed Model

Postby Steve Vokers » Sun Nov 04, 2012 5:53 pm

Bludgeon wrote: The #1 concern for 2004 voters was national security, this is an issue Bush held a significant, consistent lead on in the polls.


Yes, because Bush was very successful at politicizing national security. Remember all those terror alerts? Gee whiz, the world must have gotten safer all the sudden on the day after election, because the security alerts mysteriously stopped.
Steve Vokers
Forum God/Goddess
 
Posts: 1165
Joined: Tue Oct 07, 2003 11:58 am

Re: Nate Silver's Flawed Model

Postby Bludgeon » Sun Nov 04, 2012 5:57 pm

Steve Vokers wrote:
Bludgeon wrote: The #1 concern for 2004 voters was national security, this is an issue Bush held a significant, consistent lead on in the polls.


Yes, because Bush was very successful at politicizing national security. Remember all those terror alerts? Gee whiz, the world must have gotten safer all the sudden on the day after election, because the security alerts mysteriously stopped.


I was up waiting on the states until 5am in 2004 like the rest of you, not a fan of the former president nor the current one.
Bludgeon
Forum God/Goddess
 
Posts: 1291
Joined: Tue Apr 06, 2010 2:27 am

Re: Nate Silver's Flawed Model

Postby Stebben84 » Sun Nov 04, 2012 6:57 pm

Nate Silver started out as a baseball statistician before moving on to politics. His new book, "The Signal and the Noise," describes his theories.

In the 2008 election, Silver called 49 out of 50 states right.


http://www.cbsnews.com/8301-3445_162-57 ... ontentBody

85.1% chance of winning. I'll go with Nate on this one.
Stebben84
Forum God/Goddess
 
Posts: 5118
Joined: Thu Mar 02, 2006 12:59 pm

Re: Nate Silver's Flawed Model

Postby Huckleby » Sun Nov 04, 2012 8:16 pm

Bludgeon wrote: In the case of PPP (I like James Carville), ABC/Wapo, CNN, I would describe it as a mix of professional optimism and a bit of bewilderment relating to the 2010 Census. It's a moral issue of some kind on the part of media organizations, that the white proportion of the voting population should not increase from 2008.


Where do these polls publish exactly how their results were demographically corrected, or if they needed to be corrected for any given announcement?

I've heard a lot of authoritiative-sounding judgements from the right wing analysts. I doubt that Dick Morris and friends actually know the inside facts, they are spinning out of their butts.

The polling organizations are selling their professional reputations, they want to get it right.
Huckleby
Forum God/Goddess
 
Posts: 7043
Joined: Thu Oct 20, 2005 5:12 pm
Location: parents' basement

Re: Nate Silver's Flawed Model

Postby Bludgeon » Sun Nov 04, 2012 9:06 pm

Suffice to say, there are two prevailing constructs detailing the makeup of this year's electorate; one paradigm is going to be right, one paradigm is going to be wrong. As much as the top lines of the polls have differed, the underlying fundamentals of the polling have been remarkably consistent. The point of variance is the partisan and demographic makeup of the electorate. Is there a poll with an even partisan split that shows Obama in the lead? No.

To be fair, best to remember that the reason for such divergence is that nobody knows what the demographic composition of the electorate is going to look like.

You're right, most of these polls don't publish literature describing how they come to their conclusions about the makeup of the electorate. But whichever paradigm you like, if you look into their percentage math it's pretty clear they're findings are all very similar, and the main difference in the top line stems directly out of said poll's assumptions about demographic composition.

I'm really not trying to argue with you here. Even if the polls did publish such literature describing their methodology, I don't see how it would be any different than the literature Nate Silver publishes to describe his own.

Do you really think the Democrats are going to have a high single digit lead in turnout? Or that minority turnout in 2012 will be as high as it was in 2008? Because those combined assumptions are what generate the top lines in any poll this week.

Really those are the only questions left about who wins this election.
Bludgeon
Forum God/Goddess
 
Posts: 1291
Joined: Tue Apr 06, 2010 2:27 am

Re: Nate Silver's Flawed Model

Postby pjbogart » Sun Nov 04, 2012 10:47 pm

Bludgeon wrote:Is there a poll with an even partisan split that shows Obama in the lead? No.


I'm not a pollster, nor do I possess degrees in statistics or political science, but my understanding is that the partisan leanings in polls are fluid. The people they talk to help to dictate the leanings of the electorate. For instance, Wisconsin might be a D+3 at one point in time but after Romney's 47% speech hits the airwaves they see a shift to a D+7. Similarly, Texas might be an R+7 but after Obama promises to ban guns and pick-up trucks they switch to an R+43.

The point is, I don't think those adjustments are a constant. They change over time and according to the data collected.

And polls need to have some safeguards to trust that you're getting accurate results. Imagine that you start up the Bludgeon Polling Company and you want to see how Romney is doing in Wisconsin. You have your people make 1000 phone calls but when looking over the data you see that 54% of respondents were between the ages of 19-35. Given that you know this would be an inaccurate representation of Wisconsin and would likely favor Obama, you need to adjust your numbers so that only an appropriate percentage of those respondents are included in your final calculus.

These people do polls for a living. Bludgeon and pjbogart don't. They have no interest in looking like shitheads when the final tally is wildly different than their poll.
pjbogart
Forum God/Goddess
 
Posts: 6186
Joined: Thu Sep 04, 2003 4:57 pm

PreviousNext

Return to National Politics & Government

Who is online

Users browsing this forum: No registered users and 2 guests

moviesmusiceats
Select a Movie
Select a Theater


commentsViewedForum
  ISTHMUS FLICKR
Created with flickr badge.

cron
Promotions Contact us Privacy Policy Jobs Newsletters RSS
Collapse Photo Bar