|
|
|
|
Baseball Primer Newsblog— The Best News Links from the Baseball Newsstand
Wednesday, October 31, 2012
Come next Tuesday night, we’ll get a resolution (let’s hope) to a great ongoing battle of 2012: not just the Presidential election between Barack Obama and Mitt Romney, but the one between the pundits trying to analyze that race with their guts and a new breed of statistics gurus trying to forecast it with data.
In Election 2012 as seen by the pundits–political journalists on the trail, commentators in cable-news studios–the campaign is a jump ball. There’s a slight lead for Mitt Romney in national polls and slight leads for Barack Obama in swing-state polls, and no good way of predicting next Tuesday’s outcome beyond flipping a coin. ...
Bonus link: Esquire - The Enemies of Nate Silver
|
Bookmarks
You must be logged in to view your Bookmarks.
Hot Topics
Newsblog: Mark Appel signs under slot deal with Astros (104 - 5:52am, Jun 20)Last:  Jack Carter, calling Beleaguered CastleNewsblog: OT: NBA Finals and June thread (1029 - 5:47am, Jun 20)Last:  RollingWaveNewsblog: [OTP-June] Economic Times: Hope politics, sports don’t get mixed up: Manmohan Singh (2288 - 5:44am, Jun 20)Last:  Greg (U)KNewsblog: Matt Harvey challenged Jon Rauch to a fight (83 - 5:10am, Jun 20)Last: Swedish ChefNewsblog: Sports on Earth: Super-Royal (48 - 4:55am, Jun 20)Last: BrianBriansonNewsblog: OT: The Soccer Thread June, 2013 (645 - 4:36am, Jun 20)Last:  Swedish ChefNewsblog: ESPN.com: Yankees Acquire Fartinez (28 - 4:09am, Jun 20)Last: Jack Carter, calling Beleaguered CastleNewsblog: Former New Orleans baseball player Gene Freese dies at age 79 (5 - 3:55am, Jun 20)Last: esseffNewsblog: Neyer: Computing Manny Machado's shot at the record (48 - 3:02am, Jun 20)Last: OCFNewsblog: Perry: Josh Hamilton and the terrible, horrible, no good, very bad night (52 - 3:01am, Jun 20)Last: Walt DavisNewsblog: OT: NHL is finally back thread (1131 - 2:45am, Jun 20)Last:  zackNewsblog: LATimes: Microsoft unveils new Xbox One console (237 - 2:11am, Jun 20)Last:  CrosbyBirdNewsblog: Deadspin: Manny Ramirez is Leaving Taiwan (9 - 1:17am, Jun 20)Last: RollingWaveNewsblog: Murphy: Ruben Amaro Jr. doesn't "do" five-year plans, but the Phillies need a good one (36 - 1:07am, Jun 20)Last: Jack Carter, calling Beleaguered CastleNewsblog: Draft signings (142 - 1:07am, Jun 20)Last:  Der_K
|
|
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
Btw, I give Silver the most credence of any pollster or aggregator, but that's because of his method. I hardly take him on faith, any more than I did advanced fielding stats, or win shares, the black box nature of which left me essentially indifferent to them, despite their imprimatur.
Interesting point, and contrary to Granite State's (the preferred name of the NH polling outfit cited, according to them) claim that Independents are suddenly breaking violently to McCai..., I mean Romney. If they were, Obama's numbers would be plunging, since there weren't nearly enough undecideds to account for the shift. It would have had to have been Independents in NH previously announcing for Obama who were doing a lot of the shifting.
Isn't it terrific? Those 'fill in the map' webpages are handy, but the 512 trees are a cut above.
No offense, Howie, but I couldn't find any info telling us why the Dispatch poll, which is very similar to most other polls, is something we should pay special attention to. Any other leads wrt it?
Agreed. If PA was a genuine coin-toss, Obama would be losing OH, NC, FL, and VA. PA is an elderly state. Obama would be down by 4 in FL if PA was a tie. It's hugely unlikely the poll is correct, though. It's inconsistent with everything else, including every poll taken recently in PA. [edit: as several people have pointed out by now]
***
It's the HuffnPuff Post, so grain of salt, but it's only fifteen hours old, the most recent update I've found.
Nothing new except for the total figure.
***
That's the main problem. It's very doable, but there's a learning curve, one long enough that by the time people safely have it figured power will be back on. You can't cut corners when one end of the pump is in a ten thousand gallon tank of gasoline.
Discounting Roanoke leaves Obama better than +1 if you total LVs across the last half dozen polls. Include the next most recent half dozen, and all 12 get the total right down to Obama +0.1 (though, interestingly, Obama keeps winning the largest polls). It's a dead heat using average going back 16 polls, but after (before) that, it's all Obama.
I keep thinking Romney has an edge in Virginia. He doesn't.
He does if you look at those polls' internals, which show samples of, among others, D+8, D+8, and D+5, despite Virginia only being D+6 in 2008. It's possible Virginia is more Dem in 2012 than it was in 2008, but it doesn't seem likely.
Also, Joe, to keep hammering on a point that others have made: why do you think you know better than a vast majority of polling firms, who have every incentive to get the numbers right, in predicting the partisan makeup of likely voters? Could you really say with a straight face that you'd be doing exactly the same thing if the situation were flipped, and you thought pollsters were vastly overestimating Republican turnout? I'll have to give you a ton of credit if you turn out to be right come Tuesday - but I'd be quite shocked if you were.
Jack Carter is correct that the Virginia polling has swung backtk Obama and I expdct him to win. The state. Florida looks much better than it had. Ohio is gone. Iowa is gone. Joe still hasnt figured out how polling works.
All in all, this election is about over for Governor #######.
While this thread is more like "50 Shades of the Electorate."
Well, religious fervor implies things taken on faith. I can't speak for anyone else, but for me there's little in the way of faith involved.
If anything, I'd say the "religious fervor" is on the other side (what with the confident predictions of victory that fly in the face of the evidence and all, based on what appears to be a whole lot of wishcasting and unsupported assumptions).
Right--I mean, if I wanted to put the most credence in (sorry--express "religious fervor" for) the guy who's most confident my preferred candidate will win, I'd cast my lot with Sam Wang. (Wang does interesting work, but he *only* relies on state polls** , which I understand the logic of (since state by state results are what determines the winner) but strikes me as a little rigid.)
**and who knows--maybe he should. He was certainly accurate in '04 and '08, missing by a total of one electoral vote in those two elections *combined*.
Really? Thought he predicted the electoral vote count exactly that year.
Pundits can be wrong a thousand times with no consequences it seems that stat heads get only one strike.
Lower bar, naturally. Any blowhard with a laptop can be a pundit; creating a statistical model (however accurate it ends up being) takes some work.
Let me translate that:
"This is our last poll before the election. We're going to make sure it shows an Obama lead, and we'll use the excuse of "the storm" as to why we suddenly line up with the rest of the polls right at the end, when we've been one of the outliers for all of October."
Silver did extremely well state-by-state in 2008, and reasonably well in key races in 2010 (though hardly overwhelming). The widespread assumption that this must mean that the 2012 Presidential election will break according to Nate's say-so reaches 'religious' levels, imo, for some. I can't say where you stand on that continuum, obviously.
And if Nate does very well again this year, the 'religious' ones will be utterly locked in for 2016, so much so that dissent - heck, even questions - are liable to be found even more offensive than now. All hail The Wizard of Polls, 2-for-2 in Presidential election state-by-state predictions. And don't dare suggest that anything can happen in the next 4 years that could render a previously accurate model - and the underlying polls - any less accurate. #gottahavefaith
.........
"No offense, Howie, but I couldn't find any info telling us why the Dispatch poll, which is very similar to most other polls, is something we should pay special attention to."
Fair question, of course. I think the vibe was that this poll supposedly had proven uncannily accurate in past elections. Wish I could find the tweet and/or recall who posted it. Wasn't laying any great stake in it so much as thinking it would be in the chatter today. Will post more if I get it (again, I'm not endorsing the poll or claiming it means anything; just passing along a data point that apparently some poll-meisters were eager to see).
I never said the Dems were taking the House. I said they were picking up seats. You argued that because in 2010 Republicans gained that somehow that meant a bunch of voters had turned to the GOP and I am saying that in 2012 it appears that a bunch of those voters have now turned back to the Dems.
So apparently you're getting the backlash in ahead of time.
If Nate's model in 2016 says that candidate X is up by 5 when all the polls say he is down by 3 people will question it. If his model say the candidate is up by 6 when other sites say he is up by 4 people will think Nate is an effective modeler of elections will probably believe him. This isn't a biggy. Very few people on this site have blinders on and the people that do aren't pro-Silver.
You're trying to have it both ways with that qualification that ends your sentence. Some people are a prone to the edges of the bell-curve in their belief. So? This is news? Either too many people are "religious" in the belief of Nate's numbers, in your opinion, or they're not. Pointing out that "some" are means nothing.
There was a big Gary Johnson pitch here this weekend, hadn't really heard much about him locally before now. As of today, I have seen more Johnson signs than I have Obama or Romney ones.
Whether I support legalization of marijuana or not, I suspect the vast majority of the electorate wouldn't consider that anywhere near the most pressing issue for our federal government. Seemed like an odd way to try to win people over.
I don't really pay attention to who posts what, and I can't read people's minds anyway.
But I haven't found much interest here in setting up some sort of baseline on how to weigh Silver's analysis, before we know how he fared.
Vibe feels like "if his numbers are wrong, blame the polls" yet not accompanied with a lot along the lines of "Nate, like a quarterback, will get too much blame if he's inaccurate and too much credit if he's accurate."
Deciding in advance to dismiss a contradictory point to a narrative, while either accepting or even contributing to a surge of belief when a point winds up fitting the narrative - that sounds vaguely familiar to me.
But I can't argue against anyone who sets up here a fair, pre-election baseline of how much to praise/criticize Nate based on the results.
BTW, I hate hearing this "dead heat" crap in the press and the "over" stuff here. I'm very, very nervous.
As you should be. I don't know about religious fervor, but the certainty people have that the polls are accurate surprises me. In any case, it doesn't matter, at all, what people think the state of the race is TODAY. It only matters who gets the votes on Tuesday. If any of a number of assumptions are a tiny bit off, Romney wins...or loses by 100 EV. But I don't think either side should be feeling like they have it won.
Concur, from the other direction.
I think my guy is going to win (based on the polls and what I think turnout will likely actually be, vs. what's modeled) but I'm still far from sure, and very nervous.
Anyone who thinks this race is over, is just fooling themselves to soothe their own fears.
I do think anyone who has confidently declared it, or has been touting the accuracy or inaccuracy of polling -- one way or another -- sort of has to post a prediction for the sake of credibility. Just because for me, a Romney win would be a surprise (not a huge surprise, but roughly, say, a 4-to-1 surprise...), I would like to see which states, specifically, Romney will win, according to those who think Romney will win.
Bottom line: Obama 303, Romney 235. In addition to all the expected wins, Obama wins (west to east) NV, CO, NM, IA, OH, VA, NH. Romney hangs on in FL and NC. Pop vote is ~51.8 (O) to ~48.2 (R); the variation is for 3d parties.
Happy Sunday to all.
Already did yesterday, but I'll repeat.
R 51, O 48. Romney wins FL, NC, VA, OH, NH, CO and one of WI or IA.
Well, even if your head tends to believe that Silver's explanations for his projection make sense, that still leaves a 3 in 20 chance that we'll wake up one morning soon and be faced with Romney / Ryan / a pair of 45 year old Scalias replacing Ginsburg and Kennedy / the regulatory agencies packed with a Club for Growth clones / and the realization that a sizable part of the incoming president's electoral base is certifiably crazy. That's just a partial listing, and that's more than enough to make me a bit nervous, regardless of what my head tells me the odds are of this actually happening.
This is going to come down to enthusiasm and turnout.
Actually the Onion's New York Times link is even funnier, even though I'm in the demographic that's in their bullseye.
Still more than 36 hours til the polls open in Dixville Notch - plenty of time. But isn't there going to be a separate prediction thread? It'd be so much easier to follow and recap.
Dude. PA and MI are not tied.
O 51, R 48. Obama takes NV, IA, WI, OH, PA, VA, NH. Romney NC, CO and FL.
First declaration by right-wing psychos that Romney lost because he's too moderate: about 11.30 PM East Coast time.
EDIT: If the PA poll's the Susquehanna/Pittsburgh Scaife Daily poll discussed above, we've already dealt with that one. It's a rightwing fringe outfit for whom the reported tie represents a *backslide* of 4 points for Romney.
What makes it a bit odd is that when you look at the numbers, it skews EXTREMELY old (only 2.77% polled were 30-years old and younger, instead of 16% for historical turnouts), and VERY white (86.25% white, 8.05% black, while historical numbers are 74.5% white and 17.5% black).
Again, the relevant process of credit/demerit for the model is not in the projection from the last week of polling. Anyone can use the last week of polling to project a winner. Literally all you have to do is look at the Ohio polls.
Silver should get credit for calling this a 60%+ election for Obama months ago. That's where he stood out from the pack.
EDIT: "MyFoxDetroit.com." Say no more.
EDIT: OK, should have refreshed. Carbonated high-fructose corn syrup-containing liquids all around.
"Silver should get credit for calling this a 60%+ election for Obama months ago. That's where he stood out from the pack."
But 'Obama wins' is a binary issue, 100 pct or 0.
I realize you are just giving Nate Magic Odds there and not a prediction pct of vote.
But even if Nate said it would be 54-46 in June (he didn't) and Obama wins 54-46 (he won't) - even then it would be.... a correlation.
Could have been that it really was 50-50 then, but the populace changed its collective minds over time, and wound up getting to the 54-46. Would Nate deserve credit in that scenario?
Let's put it this way: If Romney wins, won't you then tell us that Nate WAS right back then, but the polls screwed up at the end, so it's their fault?
And if not, what DOES Nate have to lose in your view, should Romney win? I almost get the impression that if Obama wins, Nate gets "first to market" magic beans. But if Romney wins, hey, don't look at Nate!
FMWB: The Leader In Communications Technology....
For instance when silver has someone with a 70% chance of winning it doesn't mean he's winning in a landslide it means that the candidates he have up 70% are going to win 70% of the time meaning given 10 such elections he's going to be "wrong" 3 times out of 10 yet people act as if he's saying O is up 70 to 30.
Then there's the whole unskewed nonsense about adjusting the partisan composition of the polls you don't do that you are not supposed to do that you poll for partisan identification you adjust for demographics like race and sex... here most of that crap is limited to our troll JoeK.
Michigan polls since Romney's convention bump:
52-46 - PPP 11/3
46-47 - FMWB 11/2
52-47 - Rasmussen 11/1
48-41 - Grove 11/1
52-46 - PPP 11/1
53-45 - PPP 10/31
48-45 - Grengariff 1029
48-42 - EPIC-MRA 10/29
47-47 - FMWB 10/23
52-43 - Angus Reid 10/20
52-46 - EPIC-MRA 10/17
52-45 - Rasmussen 10/11
O 49.5, R 49.
EC O with 288
Senate Dems 53 to 47
house Reps 229 to 206
This seems to reflect a serious misunderstanding of what Nate's percentages mean. He's not giving voting percentage, so there's no reason to "correlate" his odds with the eventual vote. He's giving the percentage of times Obama wins a simulated election assuming the polls are right. That has nothing whatsoever to do with the percentage of the vote Obama gets.
A coin flip is a binary event too. We can still state the probability.
This. The same thing used to happen constantly -- and still happens sometimes -- with baseball metrics. People flip out when you say, "35-year-old slugger X is unlikely to return to HOF form after having an off season last year," and someone responds with "Are you kidding?!?!?!?!? He could totally hit 40 homers this year!" You both just said the same thing, with different shadings, but people see something like 70% and it seems like an inevitability.
Not everyone gives a popular vote percentage, or senate picks.
Spreadsheet here.
I don't have any final numbers for Rove or Morris, just their Senate numbers.
If anyone else finds any other online predictions, post the link in bold in this thread and I'll try to add them.
And if not, what DOES Nate have to lose in your view, should Romney win? I almost get the impression that if Obama wins, Nate gets "first to market" magic beans.
If Romney wins, then any national polls that showed Romney winning the popular vote (assuming he wins that, too) are the "winners". And any swing state polls that showed him winning (assuming he wins those states) are also the "winners". And Emperor Joe and snapper are the new BTF Royal Couple, and can ask Susquehanna who gets to wear the pants and who has to wear the dress.
And if that's the case, then the losers are the national and polls that missed, and the aggregating models that relied too much on those polls (like Nate, and like every one of them AFAICT). Every last one of them will have to make (or should be making) adjustments to their models the next time around. That's what smart people do---they adjust.
If Obama wins and Nate's made the right calls, then of course he'll have his "genius" status reinforced, but in fairness, credit should also be given to the state polls that formed the biggest part of his model. And I'm sure that Nate would be the first to say this.
Of COURSE Nate's dependent on those state polls (in aggregate) being right---has anyone ever said otherwise? But RCP is dependent on the accuracy of the national polls, and Pollster is dependent on the accuracy of the polls that they aggregate. And so on. I'm not sure exactly what your point is supposed to be.
And that the election would be an acid test for probability modeling, in forecasting voting AND baseball?
And the American nation.
And in this election, there are fifty states in which Obama could win. Don't get hung up on the national race; Nate's performance can be broken down much smaller than that.
It's normally considered polite to wait until someone says something hypocritical before you call them a hypocrite.
Can I turn the question around? What would Nate have to do to be successful in your eyes?
Bed Kate Upton.
In a sane world, people who aggregate polling data should have essentially nothing to either win or lose on election day. Everybody loves to go on about this or that person's secret sauce, but the reality is that they're all essentially just reporting aggregated data. If the aggregators are right, it means that the data was right and not much more or less than that. If the aggregators are wrong, then it means that the data was wrong. In a sane world, the people who would have something to win or lose would be the people who are ignoring or discounting data-based approaches to election analysis. If Obama wins, their claims to having specific insights that we mere mortals lack would be proven baseless, but if Romney wins, they can plausibly claim that they know things that pollsters don't.
Of course, we do not live in a sane world.
For some of his critics, Obama would apparently have to win with 85.1% of the vote.
The irony is that those people ragging on Nate for that 85.1% figure ignore the fact that right under that number he's projecting that Obama will win 50.1% of the popular vote to Romney's 48.3%. You'd think that a reasonably numerate person might mention that, too, but so far I've almost never seen any of his critics make reference to it.
As to meta-Silver thoughts, if he's correctly interpreted the polls but the polls turn out to be drastically wrong, then he's off the hook. If the polls are right but his interpretation was off, then criticize him. If both are right, everybody's happy. This should be an easy enough operation to work through, although my math and patience aren't up to it.
I think you don't really understand the way the model works.
It is really a surprise? The media cover this thing as a horse race, and here's a smart, articulate fellow with the New York Times writing that "No, no, no. You've been calling the leaders and the trailers in the horse race all wrong." The one thing that wasn't clear to me, but is abundantly clear now is that the media covers it as a horse race because that's what we want.
Exactly.
To clarify my 54-46 post: Nate got credit for "winning" almost every state in 2008, even the ones that were squeakers. If his model had McCain winning a state by 0.2 and actually Obama won by 0.2, that's still a pretty darn accurate model imo. But too many people would say he lost that one.
To me, how close his model comes to the actual vote is more important than who won the state. If he had Obama by 4 in Ohio and he only won by 0.5 pct, that doesn't seem impressive to me at all.
I would say that for Nate to be successful, his analysis will come close in nearly all the swing states - or at least come closer than most.
If Romney wins due to winning Ohio and Nate's model does well everywhere else, I would say that supports Nate's model very well even though he had the wrong winner overall. It would just be a matter of a systemic polling problem in that state.
I think there are scenarios where Obama does win the election but Nate's accuracy should be seriously questioned looking ahead. Is that a fair statement?
A significant portion of that is due to the right-wing complaint machine firing up when their narrative is attacked, so if you felt like Obama had a pretty good chance this cycle, it was inevitable that he would be attacked regularly and often enough to raise his profile significantly. In that sense, sure, I thought he was going to become a very high profile target, and as such would garner more celebrity regardless, and should he prove as accurate as last time, some (well deserved) enduring national status.
In other words the storm has halted Romney's ability to lie to and mislead voters, meaning more focus will be on his completely unworkable tax plan and other crappy positions.
ROVE: PAST OCTOBER SURPRISES WERE FAKED
I'll start at 4242.
Obama Campaign: We've Contacted One Out Of Every 2.5 People In The Country
That sounds awful. It's bad enough being bothered by campaigns right around the election, but to be confronted with it for months by people who are trying to fake a personal connection just so they can get you to vote?
If they are coming to your door it's because you let them take you very step of the way. So to complain about it would be hypocritical.
"Prediction Time: 1) The election won't end Tuesday night. 7 states will b decided by 1 pt or less. (VA, OH, CO, IA, PA, WI, NH)"
The problem is that there's no way to grade Nate, whether the number is 15% or 25% or 35%. You know, it's possible that what Nate is claiming he can do do can't be done to any reasonable degree.
If you're talking about saturation contacting via telephone, e-mail and junk mail, which happens even in non-swing states like Maryland, the answer is easy: Screen the calls, delete the e-mails, and throw away the junk mail. And if someone knocks on your door, if you're not interested just tell them you're not interested. Same thing you do to the Girl Scout cookie crew and the roving roof repairmen, and on Wednesday they'll let you alone for another 2 or 4 years.
But if you're one of the truly targeted voters, you may not have voted already. And if you're new to your area you may not even know where the polling place is, or you may not even know about early voting. Not everybody follows the election with equal interest, but everyone still gets one vote, and the more people who vote, the better.
Obviously the problem here is that the electorate has changed, polling methods have changed, the data set is still pretty limited, etc. And as I said, I don't know what's in the black box, so I don't know if Silver has implemented this method well or poorly. But the criticism should address the method that he's articulated. He has a plausible theory for how to estimate the chance of polling error.
www.nytimes.com/2012/11/04/sunday-review/the-vanishing-electoral-battleground.html
www.nytimes.com/2012/11/04/us/politics/in-ohio-2-campaigns-offer-a-study-in-contrasts.html
http://www.nytimes.com/2012/11/04/us/politics/beck-acts-as-a-bridge-between-romney-and-evangelical-christians.html
One other opinion piece from the Nov 1 WSJ:
http://online.wsj.com/article/SB10001424052970203707604578090962267200892.html
I look at what Nate's doing as similar to James's Favorite Toy, or whatever James's system was for projecting careers out (Brock2?). A fun tool for sh^ts and giggles. Not something to base any serious argument on, as people are doing here.
CO: O 48 - R 48
IA: O 49 - R 46
NH: O 49 - R 47
OH: O 49 - R 46
PA: O 50 - R 45
VA: O 48 - R 47
WI: O 50 - R 46
In the aggregate, Fleischer thinks the swing state polls are overestimating Obama's support by 2.5 points. Or, since he thinks that the polling error is concentrated in a set of demographically similar states, he thinks that Rust Belt polls are overestimating Obama's support by 4 points.
Or, he's just a guy making a possibly correct, highly specific, entirely innumerate prediction in the hopes that he hits the jackpot on Tuesday. There's no downside - if I were a pundit and lacked ethics, I'd predict something broadly plausible and highly specific like that.
If you want to "grade" Nate's overall performance you'd have to find a large sampling of his projections that he assigns an 85-15 split to, and see if his "favorite" wins 85% of the time. But obviously within that 85% there'll be cases where the winner overperforms and underperforms his projected percentage of the PV and/or EV, which are the two numbers that don't seem to be often mentioned compared to that gaudy 85% figure.
You mean sort of like "It's over. It's always been over." Let us know the next time you rag on coolstandings.com.
I think you can grade the Senate/House seats pretty well.
In addition, Nate's really good for primaries.
Sounds like a Dodgers-Giants game in Candlestick I went to in 1971, where souvenir pennants were being set on fire by Dodgers fans in front of groups of Giants fans, and vice versa. And yes, the cops had to be called in to break up several spectator brawls.
Yeah, hence my scare quotes around the word "prediction". Still, I figured you could make something of it. Remember, you don't fill out a spread sheet with the predictions you wish you had, but with the predictions you actually have. Or something like that.
I'm just citing a fact. There are two new polls at RCP showing exactly that; actually R+1 in MI, but that's rounding. I'm not saying those polls are right, but they might be.
Nobody knows who is going to show up on Tues., and nobody knows if the people who answered the polls reflect or don't reflect the actual electorate.
With a race this close, any prediction is literally a guess.
1) They're trying to run up the popular vote so they can claim Obama is illegitimate if he wins the EC and loses the popular. Though I agree with many others who think that Republicans will declare Obama illegitimate either way, just as they've been doing for the last four years.
2) They have a lot of money, and they need to spend it. There's only so much air time in swing states. And when Sheldon Adelson has already cut you a huge check, better to spend that money than it is to go back to Adelson and say "sorry we lost, and oh, we also have $3MM of the cash you gave us sitting in the bank account".
3) But mostly they're just grifters.
In the presidential election, Silver is really just following the polls and spitting out results highly similar to everyone else's. He, like everyone else, is working off the broad consensus in quantitative political science about the relative accuracy of aggregated public opinion polls. The real work, if the polling consensus is off, will be among the actual practitioners of polling and quantitative political science to figure out what went wrong.
This means Silver isn't really all that important. He happens to be the best writer among the various poll aggregators and election modelers, and he happens to have the largest platform, but he isn't doing anything on the quantitative side in this election that should make us more confident in his numbers than others. His value is provided in his writing. I like his writing.
If the SuperPAC people really are interested in seeing if they were getting the biggest bang for their Citizens United buck, they'd hire a guy like Nate to investigate the ROI.
This is not true. And the race really isn't all that close. It's only close if you avoid/discredit large amounts of data.
This is not true. And the race really isn't all that close. It's only close if you avoid/discredit large amounts of data.
Correct.
You must be Registered and Logged In to post comments.
<< Back to main