Baseball for the Thinking Fan

Login | Register | Feedback

btf_logo
You are here > Home > Baseball Newsstand > Baseball Primer Newsblog > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Tuesday, October 02, 2012

OTP: October 2012-THE RACE: As Candidates Prep, Attention in DC split between politics and baseball

While President Barack Obama and Mitt Romney bone up in Nevada and Colorado for Wednesday’s opening debate, back in the nation’s capital attention is split between the hard-fought presidential race and baseball playoffs.

The Nationals won the first division baseball championship for a Washington team since 1933 by clinching the National League East race Monday night.

Washington, D.C., has the only ballpark where so many Cabinet members, politicians and other luminaries routinely gather and where fans now are openly rooting for a particular president — one who served more than a century ago, Theodore Roosevelt.

“Let Teddy Win” banners and buttons are everywhere. Fans like 2008 GOP presidential nominee Sen. John McCain of Arizona say it’s time for Roosevelt’s 500-plus losing streak to end.

[...]

“Teddy, you are the victim of a vast left-wing conspiracy by the commie pinko libs in this town,” McCain said in a video played in the stadium Monday night. “But you can overcome that.”

The October 2012 “OT: Politics” thread starts ... now.

Joe Kehoskie Posted: October 02, 2012 at 02:14 PM | 6119 comment(s) Login to Bookmark
  Tags: nationals, politics

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.

Page 55 of 62 pages ‹ First  < 50 51 52 53 54 55 56 57 58 59 60 >  Last ›
   5401. DA Baracus Posted: October 28, 2012 at 09:31 PM (#4286476)
Oh BTW, PPP's PDF is the press release.
   5402. Joe Kehoskie Posted: October 28, 2012 at 09:33 PM (#4286480)
Moving the goal posts.

This is bizarre. Your attempted "gotcha" in #5377 talked about the order of PPP's questions, I pointed out that I never said anything about the order of the questions, and now you're accusing me of "moving the goal posts."

I don't see Fox or Rasmussen or CNN doing that.

The Fox link is a news story, not a Fox poll. The Rasmussen link has some of the partisan breakdown at the bottom, but not as much as I'd like to see. The CNN link is also a news story and not the actual poll results; the actual results are here and the report does include the sample breakdown. (The CNN/ORC sample was also very reasonable; there's a big difference between burying an unreasonable sample like PPP did and not making a big deal out of a very reasonable sample like CNN/ORC.)
   5403. The Yankee Clapper Posted: October 28, 2012 at 09:35 PM (#4286491)
PPP did that. They have a full break down in the PDF, which is linked to in the article on their site. So what's the problem?

Most of the news coverage of that poll, like the Washington Post poll I mentioned in #5394, won't mention the partisan tilt of the sample. Having a link to the material is better than nothing, but the info reaches relatively few people. Of course, it would be even better to have samples that were more reflective of the electorate, but with an 8% response rate, that may be difficult to do.
   5404. Joe Kehoskie Posted: October 28, 2012 at 09:39 PM (#4286499)
PPP did that. They have a full break down in the PDF, which is linked to in the article on their site. So what's the problem?

You're the one making a federal case out of my use of the word "buried." But just to play along, if a person wants the latest polls from PPP, they need to follow them on Twitter. Then, after reading a tweet, they need to click over to PPP's site. Then, once at PPP's site, they have to click "continue reading" to get the full article. Then, they need to scroll to the bottom of the article and click "Full report here" to download the PDF. Then, they need to read through to the third page of the PDF, to question #16, to get the sample breakdown.

If a person needs to do all of the above to get the second-most important piece of information — information that would fit in the original tweet or, at most, a second tweet — then, yes, I'm comfortable saying that the info. was "buried." Others are free to disagree.
   5405. Joe Kehoskie Posted: October 28, 2012 at 09:45 PM (#4286509)
Except, of course, that it's not standard practice. Do you criticize Rasmussen for not giving their demographic cross tabs? Do you accuse them of bias for not doing so? No, you don't. Instead, you criticize and accuse PPP, who actually does provide cross tabs.

I believe Rasmussen does provide the full breakdown for its subscribers, but I'm not sure. Either way, there are very, very few national or statewide political polls that don't disclose the sample info.
   5406. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 28, 2012 at 10:05 PM (#4286563)
That Barone guy is going to be right one day,

Michael Barone is one of the founding co-founders and co-authors of The Almanac of American Politics, and probably knows about as much about the minutiae of state and local politics as just about anybody in America. He's also a rather hardcore right wing Republican, roughly the equivalent in both opinions and temperament to Charles Krauthammer, and he's not exactly objective in his observations about current political issues and personalities. I doubt if he'd be wanting to bet that Romney's going to win Pennsylvania in the absence of a natural disaster like Sandy that throws the entire voting process into chaos.
   5407. Lassus Posted: October 28, 2012 at 10:15 PM (#4286592)
I'm going with Scholastic Magazine.
   5408. JE (Jason) Posted: October 28, 2012 at 10:17 PM (#4286594)
Michael Barone is one of the founding co-founders and co-authors of The Almanac of American Politics, and probably knows about as much about the minutiae of state and local politics as just about anybody in America. He's also a rather hardcore right wing Republican, roughly the equivalent in both opinions and temperament to Charles Krauthammer, and he's not exactly objective in his observations about current political issues and personalities. I doubt if he'd be wanting to bet that Romney's going to win Pennsylvania in the absence of a natural disaster like Sandy that throws the entire voting process into chaos.

So who is "exactly objective," Andy? David Gregory? George Stephanopoulous? Gwen Eiffel? E.J. Dionne?

And please try to stay clear of strawman territory: Barone did not even imply that Romney was going to win Pennsylvania.

BTW, you still haven't answered my question regarding the cartoon.
   5409. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 28, 2012 at 10:39 PM (#4286648)
So who is "exactly objective," Andy? David Gregory? George Stephanopoulous? Gwen Eiffel? E.J. Dionne?

All of them and none of them. I'd trust Nate's multi-faceted numbers crunching more than I'd trust the informed hunches of any one person. I'd even trust them more than I'd trust your spelling of one of PBS's main anchors---Gwen's admittedly a big ol' gal, but she's not quite a tower.

And please try to stay clear of strawman territory: Barone did not even imply that Romney was going to win Pennsylvania.

Then what's the big deal about Biden paying a visit to his home town?

BTW, you still haven't answered my question regarding the cartoon.

If I haven't, that's only because I've forgotten what cartoon you mean, and what the question was. I duck in and out of here dozens of times a day and often miss things. But ask me again and I'll answer it.
   5410. Swoboda is freedom Posted: October 28, 2012 at 10:50 PM (#4286675)
The kids have spoken! The winner of the Scholastic Student Vote is President Barack Obama, the Democratic nominee, with 51 percent of the vote. The Republican nominee, former Massachusetts Governor Mitt Romney, received 45 percent of the vote, while 4 percent of kids voted for other people.

Among the other people were Mickey Mouse, Mike Hunt, and Ronald McDonald.
   5411. DA Baracus Posted: October 28, 2012 at 10:51 PM (#4286678)
This is bizarre. Your attempted "gotcha" in #5377 talked about the order of PPP's questions, I pointed out that I never said anything about the order of the questions, and now you're accusing me of "moving the goal posts."


You said PPP buried the D/R/I breakdown in the PDF. Then you said you weren't talking about the PDF but "about the fact that PPP didn't disclose the sample in the text of the blog post, press release, or tweets it posted in hyping the alleged Obama lead." I'll be liberal with you (pun intended) and cede that you were talking about that the whole time. Because you are wrong anyway, as the PDF is the press release.
   5412. The District Attorney Posted: October 28, 2012 at 10:51 PM (#4286680)
Matt Yglesias @mattyglesias
Today’s conservative war on the central limit theorem was an unexpected development.
   5413. DA Baracus Posted: October 28, 2012 at 10:52 PM (#4286682)
The kids have spoken! The winner of the Scholastic Student Vote is President Barack Obama, the Democratic nominee, with 51 percent of the vote. The Republican nominee, former Massachusetts Governor Mitt Romney, received 45 percent of the vote, while 4 percent of kids voted for other people.


4 percent. I feel sorry for Gary Johnson that he'll get a larger share of student votes than actual votes.
   5414. JE (Jason) Posted: October 28, 2012 at 11:03 PM (#4286708)
All of them and none of them. I'd trust Nate's multi-faceted numbers crunching more than I'd trust the informed hunches of any one person. I'd even trust them more than I'd trust your spelling of one of PBS's main anchors---Gwen's admittedly a big ol' gal, but she's not quite a tower.

Even if you have religious faith in his numbers, Nate has opinions just like everyone else.
Then what's the big deal about Biden paying a visit to his home town?

Sorry, Andy, it is pretty doubtful that a candidate would visit his hometown simply to chew the fat with old buddies when there are only 96 hours left in what may be a 2000-style election.
If I haven't, that's only because I've forgotten what cartoon you mean, and what the question was. I duck in and out of here dozens of times a day and often miss things. But ask me again and I'll answer it.

You asked last night about the electoral cartoon. The "1992" caption said that a Democrat had never won without a majority of the Catholic vote. Did JFK not accomplish this feat in '60?
   5415. CrosbyBird Posted: October 28, 2012 at 11:08 PM (#4286730)
As for the positive advantage of a national popular vote: It would force candidates to campaign all over the country; and it would give both parties incentives to maximize their votes in the former swing states. It would almost certainly result in a bigger overall turnout.

I honestly think it would lead to the opposite scenario. Not only would candidates have practically no incentive to campaign in the smallest states, but they would concentrate their efforts primarily in densely populated areas.

More than 32 million people live in the NYC and LA metropolitan areas; that's over 10% of the population. More than 25% of the population lives in the top 6 CSAs. If you could get even 3% more of the NYC metro folks to vote for you, that's over 200,000 people, which would be more than 10% of the population of each of maybe a third of the states.

I think a shift to a popular vote would be very bad for the Republican party. Armed with the knowledge that their votes no longer are meaningless, the population in large urban areas, heavily blue-leaning areas, would be mobilized to vote.
   5416. DA Baracus Posted: October 28, 2012 at 11:08 PM (#4286733)
Sorry, Andy, it is pretty doubtful that a candidate would visit his hometown simply to chew the fat with old buddies when there are only 96 hours left in what may be a 2000-style election.


Obama won PA by 10 points in 2008. Biden went to Scranton days before the election then too.
   5417. JE (Jason) Posted: October 28, 2012 at 11:16 PM (#4286755)
Obama won PA by 10 points in 2008. Biden went to Scranton days before the election then too.

... which is why Biden could have hiked Kilimanjaro in late October '08 and no one would have cared.
   5418. Slivers of Maranville descends into chaos (SdeB) Posted: October 28, 2012 at 11:17 PM (#4286758)

More than 32 million people live in the NYC and LA metropolitan areas; that's over 10% of the population. More than 25% of the population lives in the top 6 CSAs. If you could get even 3% more of the NYC metro folks to vote for you, that's over 200,000 people, which would be more than 10% of the population of each of maybe a third of the states.


Yes, but the advertising cost per person may not be most economical in New York, though I don't know.

I think a big effect of the NPV would be more intensive GOTV efforts in all 50 states, proportional to the number of potential supporters there. Ad buys may concentrate in NY, but that's not the only way a campaign operates.
   5419. DA Baracus Posted: October 28, 2012 at 11:24 PM (#4286783)
... which is why Biden could have hiked Kilimanjaro in late October '08 and no one would have cared.


And you know he'd do it.
   5420. Gold Star - just Gold Star Posted: October 28, 2012 at 11:25 PM (#4286793)
What's happening down-ticket in Scranton?
   5421. Ray (RDP) Posted: October 28, 2012 at 11:27 PM (#4286802)
So, the ideas that tons of people here are saying it's over (some people were before the first debate, but not many really are now) is wrong, as is the idea that Silver is "out on a limb."


Silver most certainly is out on a limb. If Obama loses this election, Silver's battleship sinks.

It seems odd to deny that.
   5422. Monty Posted: October 28, 2012 at 11:31 PM (#4286816)
Pundits rarely face serious image consequences from being wildly wrong. If it happens to Nate Silver, I expect he'll find a way to stick around.
   5423. Ray (RDP) Posted: October 28, 2012 at 11:33 PM (#4286824)
Nate is not a pundit; he's someone who claims to have designed a model that works. If his model doesn't work, he doesn't work.
   5424. bobm Posted: October 28, 2012 at 11:43 PM (#4286868)
Silver can always use the "Garbage In, Garbage Out" defense.
   5425. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 28, 2012 at 11:47 PM (#4286882)
If I haven't, that's only because I've forgotten what cartoon you mean, and what the question was. I duck in and out of here dozens of times a day and often miss things. But ask me again and I'll answer it.

You asked last night about the electoral cartoon. The "1992" caption said that a Democrat had never won without a majority of the Catholic vote. Did JFK not accomplish this feat in '60?


I doubt if JFK lost the Catholic vote.

The wrong answer was the one for 1952. The GOP won the House that year, and in fact also controlled the Senate by the time of the 1954 election.
   5426. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 28, 2012 at 11:53 PM (#4286935)
All of them and none of them. I'd trust Nate's multi-faceted numbers crunching more than I'd trust the informed hunches of any one person. I'd even trust them more than I'd trust your spelling of one of PBS's main anchors---Gwen's admittedly a big ol' gal, but she's not quite a tower.

Even if you have religious faith in his numbers, Nate has opinions just like everyone else.


Sure, but his opinions rely on a broader set of facts than those commentators you mentioned.

And please try to stay clear of strawman territory: Barone did not even imply that Romney was going to win Pennsylvania.

Then what's the big deal about Biden paying a visit to his home town?

Sorry, Andy, it is pretty doubtful that a candidate would visit his hometown simply to chew the fat with old buddies when there are only 96 hours left in what may be a 2000-style election.


So he's worried, although the Dems aren't likely to lose the state. Fine. I guess your ouija board is as good as anyone's.
   5427. bobm Posted: October 28, 2012 at 11:54 PM (#4286942)
JFK won Catholics 78-22

http://www.gallup.com/poll/11911/protestant-catholic-vote.aspx
   5428. JE (Jason) Posted: October 28, 2012 at 11:59 PM (#4286956)
I doubt if JFK lost the Catholic vote.

The wrong answer was the one for 1952.

If so, then 1992 was wrong too.

EDIT: Thanks, Bob!
   5429. Matt Clement of Alexandria Posted: October 28, 2012 at 11:59 PM (#4286959)
Silver most certainly is out on a limb. If Obama loses this election, Silver's battleship sinks.
Silver is basically just taking a weighted average of the state and national polls.

This year, for reasons that have yet to be adequately explained, the state and national polls are diverging to an unprecedented degree. Barring a good explanation for the effect, I think the most rational choice is to take a regressed, weighted average.

It's like a ballplayer who has a "breakout" year or a "decline" year. Barring other information, the best way to handle the data is to weight and average it with all the other data.

Sometimes a ballplayer really will be "cooked", sometimes he will really have "broken out". In those cases, the regressed weighted average will miss. If this year the national polls are right and the state polls are wrong, or vice versa, Silver's model will call the race incorrectly. But I don't think it's going out on a limb to make your projection by using all the available data and taking a regressed, weighted average.
   5430. rr Posted: October 29, 2012 at 12:04 AM (#4286976)
It seems odd to deny that.


Silver is a polling analyst; his numbers tell him, at the moment, that Obama has about a 74% chance to win the election, which means that Romney has about a 26% chance to win it. 26 is a lot greater than 0. If Romney wins, Silver will look at his model and re-evaluate it.

As I said earlier, he does have a lot on the line, in terms of taking crap, rep, etc. largely because, as we have seen on this thread, there seems to be a lot of emotion surrounding what his model is telling him. Lefties want to believe the numbers are on; righties, like yourself (and spare me the "I'm not a righty" line--that was old four years ago) don't, and so are looking for flaws/biases in the model and will make a big deal about Silver's model being wrong if Romney wins.

I have already explained why this is the case: Silver is a Liberal, he works for the NYT, he was right in 2008, he has gotten a lot of attention, and unlike Gallup or RCP or Rasmussen, which are amorphous, largely faceless organizations, Silver is one guy, with a face and a name and a book to plug on Jon Stewart. But his career as a polling analyst will not come grinding to a halt if Romney wins the election. He will take a lot of crap on the net for it, but that is a different thing.

Finally, while Silver is certainly well-known in context, IMO there is a disproportionate emphasis on him here, due to the baseball/sabermetric connections. I know many people who are following the election closely who don't know who he is and do not read, and have not heard of, 538.
   5431. Lassus Posted: October 29, 2012 at 12:05 AM (#4286978)
Silver most certainly is out on a limb. If Obama loses this election, Silver's battleship sinks.
It seems odd to deny that.


This seems dumb. If he predicts a high percentage of an Obama win, and Obama loses, that means he's like you, only not as bad.
   5432. Matt Clement of Alexandria Posted: October 29, 2012 at 12:05 AM (#4286979)
Silver's model isn't a black box. He's been incredibly open about his methodology, and he's explained it well and clearly. His value as an analyst isn't limited to the numbers the model spits out - his ability to discuss the numbers, contextualize them, and so on is what keeps me reading his blog. I am confident I'll still want to hear what he has to say about politics even if the 2012 election turns out several points differently from Silver's model's projection.

The rejection of discussion of methodology in favor of discussion of results should be anathema to an analytic board like this.
   5433. tshipman Posted: October 29, 2012 at 12:08 AM (#4286990)
The rejection of discussion of methodology in favor of discussion of results should be anathema to an analytic board like this.


They're not opposed to discussing methodology! Look at how many times they bring up the exit poll partisan split from 2008--they're quite comfortable talking about polling weights.
   5434. rr Posted: October 29, 2012 at 12:09 AM (#4286994)
If this year the national polls are right and the state polls are wrong, or vice versa, Silver's model will call the race incorrectly.


As I said, I think there is a good chance that Silver's model will call the electoral result right
--Obama will be re-elected, but will miss on the popular vote, which will be dead even (49/49 rather than 50-48) with perhaps Romney actually getting more total pop votes.
   5435. Joe Kehoskie Posted: October 29, 2012 at 12:12 AM (#4287002)
Silver's model isn't a black box. He's been incredibly open about his methodology, and he's explained it well and clearly.

Has he explained what he's projecting for Ohio? I agree with your #5429, but that doesn't seem to be what Nate is doing with Ohio. He has Romney at -4.0 relative to the historical mean for the GOP in Ohio vis-a-vis the national polling.
   5436. DJS and the Infinite Sadness Posted: October 29, 2012 at 12:13 AM (#4287004)
Silver is a polling analyst; his numbers tell him, at the moment, that Obama has about a 74% chance to win the election, which means that Romney has about a 26% chance to win it. 26 is a lot greater than 0. If Romney wins, Silver will look at his model and re-evaluate it.

Quibble. Presumably, he will (and should) improve his model with 2012's data and insights regardless of whether Obama or Romney win, and to the same degree.

The hardest thing for Nate to deal with, in general, isn't whether one guy wins or the other, but the fact that people are really stupid in general and even worse when dealing with probability. After Melky's suspension, I projected the Dodgers as having a 0.3% better shot at winning the division at that point. So naturally, when the Giants win, people tell me how wrong I was and insist that I actually said "the Giants have no chance of winning."
   5437. The Yankee Clapper Posted: October 29, 2012 at 12:14 AM (#4287006)
If his model doesn't work, he doesn't work.

I doubt if it would be that bad for Nate. He'd take a big hit, but the NYT is unlikely to replace him as long as he is bringing in traffic and buzz. Given their plummeting revenue, they might pay him less going forward once he is no longer infallible, but he may not be the only one. A lot might depend on how he frames his final prediction. If Romney maintains or expands his polling lead and Nate says Obama has a 98% chance of winning, that'd be a lot worse than if his model leads him to a more hedged prediction. If Obama wins, he's golden again; but if it's Romney, folks will be comparing him to the Literary Digest.
   5438. McCoy Posted: October 29, 2012 at 12:16 AM (#4287010)
Even if Romney wins I don't really see a reason why Nate should be disregarded in the future. I think he does what he does rather well and I certainly would place more confidence in aggregators and those creating models that work over a single poll or two. I mean what should a person do? Stick with Gallup and ignore all other polls because it picks one election right?

If Romney wins and Nate predicts a win for Obama but Nate nails 98% of the congressional and gubernatorial elections can someone really crow and say they knew it all the time that Nate was wrong on his %? 75-25 or 70-30 isn't a certain thing by any stretch of the imagination. Play craps and you'll discover that quickly.
   5439. rr Posted: October 29, 2012 at 12:16 AM (#4287011)
Quibble. Presumably, he will (and should) improve his model with 2012's data and insights regardless of whether Obama or Romney win, and to the same degree.


That is a good point. At the same time, though, the context of his doing that will obviously be a lot different if Romney wins.

I will be interested to see if Silver talks about the hurricane--what effects it could have, is there historical precedent, etc.
   5440. Matt Clement of Alexandria Posted: October 29, 2012 at 12:18 AM (#4287013)
Silver's model is basically just a weighted average of all the polling done in advance of the election. If the 538 model projects a 98% chance of an Obama win and Romney wins, that will mean there was a massive breakdown in public opinion polling across the industry.

That could certainly happen. Polling has missed pretty badly on a number of modern elections (1996 being the most recent), and it could be that this year is the tipping point for traditional polling methodologies failing to measure a complex contemporary electorate.

But if it happens, it will require a mass re-consideration of the whole polling industry. Silver, as an analyst and aggregator, will have a lot of questions to deal with, but the polls and pollsters themselves will be the bigger issue.
   5441. DJS and the Infinite Sadness Posted: October 29, 2012 at 12:19 AM (#4287015)
That is a good point. At the same time, though, the context of his doing that will obviously be a lot different if Romney wins.

Perhaps, but I suspect that Nate's come to terms with the fact that even solid models will get a lot wrong and to not let it change his best practices. I know I have and I'd be surprised if Nate was all that different in this regard (I was never very close with Nate, so I'm not speaking from any personal insight here, just my experience in making projections).

I know essentially, I'm employed to be a writer and to write stories that people are interested in reading. Now, I have tools and experience using baseball stats so that I can tell different stories, stories I believe to be based more on an intellectual foundation than an emotional one, but on a fundamental level, I'm employed as a writer. I doubt Nate's situation is any different - as long as his models generally don't say crazy stuff, as long as he has insight on the political process from an intellectual, interesting point of view, he'll be fine.
   5442. Monty Posted: October 29, 2012 at 12:20 AM (#4287018)
Has he explained what he's projecting for Ohio? I agree with your #5429, but that doesn't seem to be what Nate is doing with Ohio. He has Romney at -4.0 relative to the historical mean for the GOP in Ohio vis-a-vis the national polling.


I think he's just relying on the polls and not using the relationship Ohio has had to the National voting. It's just a philosophical difference, really. You can either say "If this is the National average, this is how Ohio usually votes" or you can look at what the polls have to say. And obviously whichever approach you use, you can regress it a bit with the other one, but I think averaging the two approaches isn't going to serve either one all that well.

It's a little like the (possibly) skewed polls. If the polling says there are this many Democrats, I think it's a valid approach to say, "This is what the polling tells us." Unskewing them by applying results from previous years is something that just seems at odds with the premise of polling to begin with, which is to go out and try to get actual current data.
   5443. Ray (RDP) Posted: October 29, 2012 at 12:20 AM (#4287019)
Silver is basically just taking a weighted average of the state and national polls.


----

Silver is a polling analyst; his numbers tell him, at the moment, that Obama has about a 74% chance to win the election, which means that Romney has about a 26% chance to win it.


I say that Silver is out on a limb, and people respond by telling me what Silver is doing. That is odd.

He has a methodology. If it misses badly, he misses badly.
   5444. Monty Posted: October 29, 2012 at 12:21 AM (#4287023)
He has a methodology. If it misses badly, he misses badly.


I agree with that. If his methodology is unsound, we'll find out eventually.
   5445. Matt Clement of Alexandria Posted: October 29, 2012 at 12:23 AM (#4287027)
I say that Silver is out on a limb, and people respond by telling me what Silver is doing. That is odd.
To belabor the projection analogy. If a player has hit .300, .330, and .220 in the last three years, and I project him to hit .275 next year, am I out on a limb? I'm projecting a big 50 point jump in batting average, but my methodology is simply to take a regressed, weighted average like I would for any projection. I don't think it's "going out on a limb" to apply a aimple and rational methodology to solving a problem.

EDIT: Belaboring further - in this analogy, the .300 and .330 are the state polls which show Obama up about three nationally, and the .220 is the national polls which show Romney up about one point. Nate's pretty much just averaging them together, weighted for various complicating factors (sample sizes, recency, etc).
   5446. Joe Kehoskie Posted: October 29, 2012 at 12:27 AM (#4287034)
If the 538 model projects a 98% chance of an Obama win and Romney wins, that will mean there was a massive breakdown in public opinion polling across the industry.

Except that there's a divergence between the polling and Nate's model. Nate currently projects Obama as beating Romney by 1.7 points nationally, while Romney is up 2.0 points over Obama in the tracking polls at RCP over the past 7 days. If the polls hold at these numbers and Nate still projects an Obama win, I don't see any way that the pollsters can be blamed, unless the criticism is narrowly tailored to the polls of Ohio.
   5447. McCoy Posted: October 29, 2012 at 12:29 AM (#4287035)
Nate makes predictions on over 500 elections every 2 years. I'd say he is out on a limb every two years and so far he has been very very accurate. If Romney wins that doesn't prove his models don't work.
   5448. DJS and the Infinite Sadness Posted: October 29, 2012 at 12:30 AM (#4287038)
Except that there's a divergence between the polling and Nate's model.

No, there's a divergence between the non state-specific polling and Nate's model.

And you *know* I'm not one of the liberals. I've already voted for Romney in Ohio.
   5449. rr Posted: October 29, 2012 at 12:31 AM (#4287040)
As DJS, a professional in the field of statistical projections, suggests, there is a disconnect of sorts between "what Silver is doing" and the public perception of it and much of the public discussion of it. So, yeah, in terms of "How much shitt he will take from certain BTF righties and on FOX News" and how much "You blew it! You let us down! We trusted you!" backlash there will be from angry Democrats in the event of an Obama defeat, yes, Silver is "out on a limb." But in terms of what he is doing, the approach seems to be reasonable and so does the projection.

As to predictions, my guess it that the numbers won't change much in the next eight days, and Silver will issue a hedged prediction, pointing out that the model says that Romney has a 25-30% chance to win, which means that Obama is not a lock.
   5450. Joe Kehoskie Posted: October 29, 2012 at 12:33 AM (#4287043)
Nate makes predictions on over 500 elections every 2 years. I'd say he is out on a limb every two years and so far he has been very very accurate. If Romney wins that doesn't prove his models don't work.

But how many of those were competitive races? Of the most competitive Senate races in 2010, he went just 2-for-5. We didn't need Nate to tell us what would happen with Barbara Boxer or Daniel Inouye or ~25 other Senate races.
   5451. Ray (RDP) Posted: October 29, 2012 at 12:34 AM (#4287045)
To belabor the projection analogy. If a player has hit .300, .330, and .220 in the last three years, and I project him to hit .275 next year, am I out on a limb?


But Nate doesn't have three Obama-Romney election cycles. He doesn't even have one.

   5452. The Yankee Clapper Posted: October 29, 2012 at 12:34 AM (#4287046)
To be clear, if Romney wins in a polling environment that resembles what we are seeing today, I don't think the problem would be Nate's model so much as the large number of state polls that appear to be oversampling Democrats. I have a gut feeling that the huge number of non-responders doesn't break evenly along partisan lines. Curmudgeonly conservatives, media suspicious Republicans and those with heightened privacy concerns may make up a disproportionate number of the non-responders, tilting the remaining pool toward the Democrats. Folks on the left may be more inclined to give out personal info - remember how many here contributed to the ass wiping thread.

I'd be curious where Nate's model would come out if he adjusted those polls to 2008 and 2004 partisan splits.
   5453. Matt Clement of Alexandria Posted: October 29, 2012 at 12:35 AM (#4287048)
No, there's a divergence between the non state-specific polling and Nate's model.
And, to quibble slightly, there's also a divergence between the state-specific polling and Nate's model.

Because Nate's model works by basically averaging the state-specific polling with the national polling, and because there is a greater volume of state-specific polling, Nate's model diverges more from the national polling than from the state polling.

But, obviously, I'm just belaboring a point that you're perfectly clear on. (And with that last belabor, I need to go to bed to prepare for a big day of working from home while it rains.)
   5454. Joe Kehoskie Posted: October 29, 2012 at 12:36 AM (#4287050)
No, there's a divergence between the non state-specific polling and Nate's model.

And you *know* I'm not one of the liberals. I've already voted for Romney in Ohio.

Right. I was simply responding to MCoA's contention that a Romney win would mean there had been a "massive breakdown in public opinion polling across the industry," when it would really only mean there was a breakdown in polling in as few as one out of 50 states (Ohio).
   5455. Ray (RDP) Posted: October 29, 2012 at 12:38 AM (#4287053)
To be clear, if Romney wins in a polling environment that resembles what we are seeing today, I don't think the problem would be Nate's model so much as the large number of state polls that appear to be oversampling Democrats.


But that _would be_ a problem with Nate's model.

It would be like projecting Curtis Granderson to hit .300 and then when he hits .250 saying, "Oh, the problem is not with my model; it's with the fact that so many of Granderson's PAs result in strikeouts."
   5456. McCoy Posted: October 29, 2012 at 12:41 AM (#4287056)
It would be like projecting Curtis Granderson to hit .300 and then when he hits .250 saying, "Oh, the problem is not with my model; it's with the fact that so many of Granderson's PAs result in strikeouts."

Not really. It would be more like if he projected Granderson to hit .300 and then the official scorer scores his performances so that he has a .300 average all season and then in November the leagues says "oops we made a few mistakes. Curtis actually hit .295."
   5457. DJS and the Infinite Sadness Posted: October 29, 2012 at 12:43 AM (#4287058)
But that _would be_ a problem with Nate's model.

It would be, but not one he really has information to deal with at this point. Party ID in polling is more fluid than people think. Nate doesn't have a polling firm - he's at the mercy of the pollsters. He doesn't have a number of Granderson's strikeouts, he has a bunch of polls that try to ask how many strikeouts people believe Granderson has. If Nate was labeled the Dictator of Polls and got to redesign the polling industry to give the best data possible for making predictions, he'd no doubt make some big changes.
   5458. Random Transaction Generator Posted: October 29, 2012 at 12:43 AM (#4287059)
how much "You blew it! You let us down! We trusted you!" backlash there will be from angry Democrats in the event of an Obama defeat, yes, Silver is "out on a limb."


I'm not a Democrat (Canadian, after all), but I'd be surprised if people who visited 538.com and voted for Obama would turn to the guy who put up a prediction and say "You blew it! You let us down!".

What's the logic, other than misplaced rage. It's not like Silver said "Obama's gonna win, so don't bother voting." or "Obama's going to win, so go to Intrade and put some money down on it."

I can understand the Republican voters who followed his site swarming the comment section with "Haha! You were wrong!" gloating.
That's like getting angry at the sports columnist that predicted the Tigers were going to win the World Series. Yeah, he's wrong, but why would Tiger fans get mad at him?
   5459. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 29, 2012 at 12:45 AM (#4287063)
You asked last night about the electoral cartoon. The "1992" caption said that a Democrat had never won without a majority of the Catholic vote. Did JFK not accomplish this feat in '60?

I doubt if JFK lost the Catholic vote.

The wrong answer was the one for 1952.


If so, then 1992 was wrong too.


No, because the caption for 1992 says that "No Democrat has won without a majority of the Catholic vote....until Clinton did." And that's correct. Clinton won only a plurality (44%) of the Catholic vote that year. You can look it up.

   5460. Joe Kehoskie Posted: October 29, 2012 at 12:48 AM (#4287069)
And, to quibble slightly, there's also a divergence between the state-specific polling and Nate's model.

Understood, although it seems unclear why there's such a wide divergence between Nate's model and the RCP state-by-state polling. On RCP's no-tossups map, Obama wins 290-248, which basically means the election comes down to Ohio and a state like New Hampshire. But RCP has Romney up 2.0 while Nate is projecting Obama to win by 1.7. Even if Romney got creamed in Ohio and some other state, it wouldn't result in a 3.7-point swing. Perhaps it's just a matter of Nate using a lot of very pro-Obama polls (e.g., PPP) that RCP doesn't use, but 3.7 points seems like a very wide gulf just 8 days from Election Day.
   5461. DJS and the Infinite Sadness Posted: October 29, 2012 at 12:54 AM (#4287072)
Ah, the French. While it's certainly a pretty irrelevant factor, if the French want Obama to win next week, I'm not sure that endorsements from French Socialists are all that helpful.
   5462. rr Posted: October 29, 2012 at 12:56 AM (#4287075)
What's the logic, other than misplaced rage.


None, and most of Silver's regular readers won't do it, probably. But there will be a bit of that if Obama loses; people are emotional about Presidential elections, even if neither candidate is inspiring the masses.

That said, I think that the "Ha ha ha" from the right will outweigh anger directed at Silver from the left in the event of a Romney victory.
   5463. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 29, 2012 at 12:58 AM (#4287077)
Nate makes predictions on over 500 elections every 2 years. I'd say he is out on a limb every two years and so far he has been very very accurate. If Romney wins that doesn't prove his models don't work.

A better way of putting that would be to say that no model is a machine, but that some models provide more consistent results than others. We don't discard Rasmussen after he way overestimated the Republican gains in 2010, but we do note that he's entering 2012 without the same sheen he'd brought into that year after 2008.

Likewise, any prejudgments about "failed models" don't mean anything until we see to what extent their predictions failed, and how their predictions compared to those of others. If we were going to be tossing out pollsters and analysts on the basis of one narrowly ended election, then the truth is we wouldn't have anyone left.

And Joe can start kvetching once again, as Nate's just revised Obama percentage is back up to 74.6%, with 80.1% in the now-cast. And Intrade is up to 63.1% for Obama, which means that there's still plenty of time and room for Joe to cash in on his opinions. I've heard all it takes to place one of those bets is a valid credit card and a set of unfiled fingerprints, so go at 'em, boy!
   5464. JE (Jason) Posted: October 29, 2012 at 12:58 AM (#4287078)
"No Democrat has won without a majority of the Catholic vote....until Clinton did."

Yikes! That's a reading comprehension fail. Thanks.
   5465. Monty Posted: October 29, 2012 at 01:02 AM (#4287079)
That said, I think that the "Ha ha ha" from the right will outweigh anger directed at Silver from the left in the event of a Romney victory.


I think -- well, I hope -- that after the election, people's attention will move from the polling back to the actual candidates. If Romney wins, the right will be having victory parties and making lists of Obama policies they want to overturn.
   5466. Joe Kehoskie Posted: October 29, 2012 at 01:02 AM (#4287080)
That said, I think that the "Ha ha ha" from the right will outweigh anger directed at Silver from the left in the event of a Romeny victory.

Interesting; I've been thinking the opposite. I have no doubt that some Romney supporters will gloat over at FiveThirtyEight, but Nate has been telling Obama fans for over a year that Obama is highly likely to win reelection, if not coast to reelection. I suspect he'll have to contend with a lot of angry lefties if Obama loses.

(Just for the record, I'm rooting against Nate simply because I don't want four more years of Obama, not because I want Nate to suffer any personal or professional embarrassment.)
   5467. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 29, 2012 at 01:03 AM (#4287081)
Ah, the French. While it's certainly a pretty irrelevant factor, if the French want Obama to win next week, I'm not sure that endorsements from French Socialists are all that helpful.

Well, it didn't kill his chances the last time, so WTH. And anyhoo, other than Netanyahu and a few Eastern European leaders, what world leader isn't for Obama? It doesn't exactly take a wizard to see who's pulling Romney's strings in foreign policy, and they saw enough of that between 2001 and 2008 to last them a lifetime.
   5468. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 29, 2012 at 01:05 AM (#4287084)
Interesting; I've been thinking the opposite. I have no doubt that some Romney supporters will gloat over at FiveThirtyEight, but Nate has been telling Obama fans for over a year that Obama is highly likely to win reelection, if not coast to reelection. I suspect he'll have to contend with a lot of angry lefties if Obama loses.

I'll go on the record as saying that Nate Silver is the last person I'd be mad at in the case of a Romney win. He's not God, and that being the case, he's not going to nail every projection.
   5469. DJS and the Infinite Sadness Posted: October 29, 2012 at 01:06 AM (#4287085)
I'm not a Democrat (Canadian, after all), but I'd be surprised if people who visited 538.com and voted for Obama would turn to the guy who put up a prediction and say "You blew it! You let us down!".

You ever read the comments? There's pretty dumb stuff all around.

Now, if Nate made mistakes in the past, I think it was more in some of the people who wrote for fivethirtyeight regularly before the transition to the Times. Schaller and Quinn were nearly unreadable and Quinn's Obama Fan Fiction (I'm not making this up) has to easily be the worst thing I read there.
   5470. Jolly Old St. Nick Is A Jolly Old St. Crip Posted: October 29, 2012 at 01:07 AM (#4287086)
No Democrat has won without a majority of the Catholic vote....until Clinton did."


Yikes! That's a reading comprehension fail. Thanks.

Don't feel bad, Jason. I had to double check myself to make sure he had that word "majority" in there.
   5471. Random Transaction Generator Posted: October 29, 2012 at 01:18 AM (#4287092)
(Just for the record, I'm rooting against Nate simply because I don't want four more years of Obama, not because I want Nate to suffer any personal or professional embarrassment.)


I'd say that my rooting interest is 65% Obama, 25% Nate, 10% against the Unskewed guy.

I loathe the idea that his straight-up poll manipulation will get any traction for being "right".

From a view on the Republican side, I can imagine that Romney winning the election will provide great joy in visiting the Princeton election website for maximum gloating.

BTW, he has a post about how to judge all the different election prediction/poll aggregators.

   5472. JE (Jason) Posted: October 29, 2012 at 01:25 AM (#4287097)
Don't feel bad, Jason. I had to double check myself to make sure he had that word "majority" in there.

On second thought, society is to blame.
   5473. Machine Posted: October 29, 2012 at 01:26 AM (#4287099)
Nate makes predictions on over 500 elections every 2 years. I'd say he is out on a limb every two years and so far he has been very very accurate. If Romney wins that doesn't prove his models don't work.


Not really. I think Ray is right. Thus far, Silver has made one prediction. He got it right. The other 499 predictions are for nerds and nobody really cares about them.*

If Silver ends up predicting an Obama victory and Romney wins, he's a prognosticator that only gets things right 50% of the time. Good luck staying employed with that track record.

* Exaggerating to make a point. The other 499 is what allows Silver to have his job. But on the job posting to be a poll analyst the abilty to project the 499 are considered "preferred qualifications" and the presidency is "required qualifications". He can't be wrong on the presidency.
   5474. McCoy Posted: October 29, 2012 at 01:36 AM (#4287101)
He can't be wrong on the presidency

I can't see why he can't be wrong. It's one thing to predict that candidate A is going to win when in fact candidate B wins by 15 points but that isn't what Nate is doing. His model shows that it is going to be a really close popular vote election but that Obama has enough electoral votes to win. If Romney wins by a half a point or point and 1 or 2 states goes to Romney instead of to Obama as predicted by Nate that doesn't really negate Nate's work or findings.

At this point it seems like the biggest issue is that some people take issue with the 75-25 probability that Nate is listing which Romney winning a close election doesn't prove was the wrong probability to have.

The national polls basically say it is pretty close to a coin toss between the two of them and the electoral map says it is a slight advantage to Obama. So if it 50-50 to start with or 49-51 to start with and then throw in the "locked", "leaning", and leading state numbers I can certainly understand why Obama would have a probability of winning north of 60%.
   5475. Monty Posted: October 29, 2012 at 01:49 AM (#4287102)
Isn't Nate really making fifty separate predictions for the presidential election? If he's correct on all but one state, that wouldn't necessarily invalidate his methodology. I think we'll want to look at the actual numbers on each state, which ought to reveal if there's a systemic flaw, either in Nate's methods or in the way polls are currently being done.
   5476. Machine Posted: October 29, 2012 at 02:32 AM (#4287110)
@5474 - pretty much agree that just because Silver is "wrong" he isn't really wrong. He's simply suggesting probabilities and long shots do win. Because his probabilities are not at the 98% confidence rate, you would expect him to get a small number wrong.

But since that took a whole two sentences to explain and used some elementary math, that's way too detailed and complicated of an explanation for most people. "Silver got the election wrong" is much easier for folks to wrap their heads around. It's stupid, but if he doesn't get the prediction right, I think it comes back to burn him. That's might even be true if it ends up being a 2000/04 style election and his confidence on Election Day is only 55%.
   5477. DJS and the Infinite Sadness Posted: October 29, 2012 at 02:59 AM (#4287117)
If Silver ends up predicting an Obama victory and Romney wins, he's a prognosticator that only gets things right 50% of the time. Good luck staying employed with that track record.

Have you actually seen the people that are employed? Dick Morris, David Gergen, John Zogby, and that's just the very tip of a gigantic iceberg. Nate's competition for prognostication is a whole lot of people that couldn't predict a coin flip if they got to make two guesses.
   5478. Joe Kehoskie Posted: October 29, 2012 at 03:43 AM (#4287121)
"Silver got the election wrong" is much easier for folks to wrap their heads around. It's stupid, but if he doesn't get the prediction right, I think it comes back to burn him.

If the polling aggregators like RCP and the major national pollsters like Gallup and Rasmussen have Romney at something like +2.0 next Monday, which is about what Romney's lead has been in the national polls conducted over the past 7 days, and Nate still predicts an Obama win but ends up being wrong, I don't know that he can just shrug his shoulders and say, The result was bad, but my methodology was good. In the preceding scenario, Nate will be predicting that Obama will win the Electoral College while losing the popular vote by more than a point, which is something that has never happened in any presidential election for which polling data is available. And not only will he be predicting that, but he's likely to be doing so on the basis of state polling in Ohio that is also going against ~40 years of that state's history. Such a prediction would be a bold bet based on a bold bet.

With Nate relatively new to the political scene and with just one presidential election every four years, this would be a major whiff, especially if he keeps Obama as substantially better than a coin-flip favorite.
   5479. Greg K Posted: October 29, 2012 at 05:12 AM (#4287123)
Understood, although it seems unclear why there's such a wide divergence between Nate's model and the RCP state-by-state polling. On RCP's no-tossups map, Obama wins 290-248, which basically means the election comes down to Ohio and a state like New Hampshire. But RCP has Romney up 2.0 while Nate is projecting Obama to win by 1.7. Even if Romney got creamed in Ohio and some other state, it wouldn't result in a 3.7-point swing. Perhaps it's just a matter of Nate using a lot of very pro-Obama polls (e.g., PPP) that RCP doesn't use, but 3.7 points seems like a very wide gulf just 8 days from Election Day.

I'm a newbie when it comes to polling (my only contributions to the polling discussion in this thread have been to say polling ought to be outlawed), but it seems to be an engaging discussion so I'm trying to get the hang of it. Apologies if this is wrong, I'm laying out here what I seem to be getting from this thread in the hopes that someone corrects me, or clarifies some things.

1) The national polls and the state-by-state polls are diverging widely
2) Silver is taking both into account but weights the state polls more (because there are more of them?)

As per the quoted comment, which my limited knowledge in the area is leaving me in need of a clarification.

I'm reading this as, RCP has Romney +2.0 in its state-by-state polls, while 538 has Obama +1.7, is that correct? Which seems to imply that there maybe isn't necessarily such a divergence between state and national polls.

Or is the RCP +2.0 based on national polling while the 538 is based on a weighted combination of national and state, in which case the popular vote gap surely could be made up by differences in all 50 states, rather than just the swing ones.

EDIT: Also as a clarification. By "state-by-state" polling do we mean that the polls in each state are used to project vote totals in those states, which are then combined together to get a national total?
   5480. Gonfalon Bubble Posted: October 29, 2012 at 06:05 AM (#4287128)
Over the past eight or ten elections, the GOP candidate has never underperformed in Ohio relative to the national popular vote, and the average has been a +2.0 gain.

McCain: Ohio 46.80%, USA 45.63% (+1.17)
Bush: Ohio 50.81%, USA 50.73% (+0.08)
Bush: Ohio 49.97%, USA 47.87% (+2.10)
Dole: Ohio 41.02%, USA 40.72% (+0.30)
Bush: Ohio 38.35%, USA 37.45% (+0.90)
Bush: Ohio 55.00%, USA 53.37% (+1.63)
Reagan: Ohio 58.90%, USA 58.77% (+0.13)
Reagan: Ohio 51.51%, USA 50.75% (+0.76)
Ford: Ohio 48.65%, USA 48.02% (+0.63)
Nixon: Ohio 59.63%, USA 60.67% (minus -1.04)
Nixon: Ohio 45.23%, USA 43.42% (+1.81)
Goldwater: Ohio 37.06%, USA 38.47% (minus -1.41)

The average Republican advantage in Ohio over the last ten elections isn't +2.0%, it's +0.66%. A third of that is from George W. Bush in 2000. In seven of the last ten elections, the difference was less than 1%.

In the 1940s and 1950s, Ohio's discrepancy used to be higher:

Nixon: Ohio 53.28%, USA 49.55%
Eisenhower: Ohio 61.11%, USA 57.37%
Eisenhower: Ohio 56.76%, USA 55.18%
Dewey: Ohio 49.24, USA 45.07%
Dewey: Ohio 50.18%, USA 45.89%

Herbert Hoover averaged +7% in Ohio in his two elections.
   5481. Matt Clement of Alexandria Posted: October 29, 2012 at 07:22 AM (#4287137)
1) The national polls and the state-by-state polls are diverging widely
2) Silver is taking both into account but weights the state polls more (because there are more of them?)
Exactly right.

The RCP and Pollster national averages are just the averages of the national polls, which are showing a small Romney lead. The state-by-state polling (not just in Ohio, also in pretty much all the swing states) is consistent with a somewhat larger, but still small Obama lead. The 538 model uses a weighted average of all the polls, and since there are more state polls than national polls (and they have larger samples and other things like that), the 538 model is tilted more to the state-by-state polling average than to the national polling average.

If you projected the election using only the polls from the states - ie, you threw out the national polls and used only polls of Ohio, Virginia, Wisconsin, North Carolina, Florida, etc to produce your national polling average - you would see Obama in a lead about a point larger than the one the 538 model gives him.

It is possible that folks in swing states are just sort of all leaning Democratic, compared to the national average, perhaps because Romney's leap in the polls after the first debate was concentrated in non-competitive states. This would mean that both the national polling and the state polling could be correct. There is, as far as I know, no positive evidence of such a wide divergence between swing state and safe state voting preference. So I am doubtful this is the explanation.

It is possible that either the state-by-state polling is wrong, or that the national polling is wrong. People have put forward the "party weighting" argument, but I think that's just mistaking cause and effect. Party identification is very fluid, and people who've just told a pollster they're voting for the Republican are more likely to identify as Republican or Independent than as Democrats, even if they have a history of supporting Democrats. And vice versa. So if a poll is good for Republicans, it's underlying party id sample will be good for Republicans, and vice versa. I guarantee that if Romney wins, the national electorate will have a much higher number of people who identify to pollsters as Republicans than if Obama wins. That isn't evidence of anything beyond the ways in which people self-identify after revealing their voting preference to a pollster. So I don't think there are currently any convincing explanations out there of the state/national polling divergence.

Given a situation where you can't explain an odd effect, I think Silver's solution - treat all the data equally, weight it and average it - is the best response.
By "state-by-state" polling do we mean that the polls in each state are used to project vote totals in those states, which are then combined together to get a national total?
Basically, yup.
   5482. Bitter Mouse Posted: October 29, 2012 at 08:02 AM (#4287150)
Have you actually seen the people that are employed? Dick Morris, David Gergen, John Zogby, and that's just the very tip of a gigantic iceberg. Nate's competition for prognostication is a whole lot of people that couldn't predict a coin flip if they got to make two guesses.


Darn it Dan, I was reading along and going to make fun of everyone who thought Nate might be unemployed if he got it wrong in 2012 and you post almost exactly what I was going to! Oh well, let me second what you said.

Nate is not employeed to accurately predict elections. Really. He is employed to get eyeballs (which is helped by being accurate, I admit). Still he drives plenty of traffic and is a very good analyst and writer and much of his competition regularly gets outsmarted by house pets. He'll be fine.

And Gonfalon Bubble - thanks for the Ohio figures. The whole +2% bit seemed fishy to me, but I was too lazy to actually check. You rule.
   5483. Rickey! trades in sheep and threats Posted: October 29, 2012 at 08:06 AM (#4287152)
Amen. Moreover, Barone, Rick Klein (ABC), and Sam Stein (hello, Huffington Post?) must be hacks too, since they have also commented on the tightening race in Pennsylvania.


1) Intentionally feed a specific narrative that favors your guy.

2) Have the horserace hacks pick up your narrative as a "story."

3) Claim the horserace coverage vindicates your narrative.

Keep dancing, J. Keep dancing.
   5484. Jay Z Posted: October 29, 2012 at 08:54 AM (#4287174)
I think -- well, I hope -- that after the election, people's attention will move from the polling back to the actual candidates. If Romney wins, the right will be having victory parties and making lists of Obama policies they want to overturn.


Obama will continue as usual, and Romney will attempt to favor financial and military interests more than we already do. Perhaps people don't think it's possible to favor those interests more than we already do, but I'm sure he can find a way. Why else would they want an R in the office?
   5485. Bitter Mouse Posted: October 29, 2012 at 08:56 AM (#4287176)
I think it comes down to ground game.* Does the Obama machine deserve its reputation? How much better is the Romney GOTV machine than McCain's was (It has to be better, but how much)?

The supplimentary question is how does enthusiasm fit in? Not voter enthusiasm, but volunteer enthusiasm. How much will the drop in Obama enthusiasm hit their GOTV workers/volunteers? Will the evengelicals (one of the main GOP GOTV groups) be more motivated by their hatred of Obama or turned off (but still vote) by Romney (both meh and Mormon)? Do unions still matter in the Dem GOTV univese (continue to slide, but the GOP war against unions is very fresh in many minds)?

I wish I had answers. I have faith in the Obama GOV operation, but since every election is in some sense De Novo it is faith.

* I think Obama has a small but real advantage even outside ground game, but ground game could easily make the difference. Also I am including early voting in ground game, and here is appears the Dems are doing good, but it is not overwhelming in their favor or anything.
   5486. Matt Clement of Alexandria Posted: October 29, 2012 at 09:05 AM (#4287183)
* I think Obama has a small but real advantage even outside ground game, but ground game could easily make the difference.
It might be the case that Romney has a small but real advantage outside of the ground game, if the national polls are right. We really don't know right now. It seems mostly likely that you're right - small but real advantage to Obama - but there's a lot more uncertainty in the polls right now than there was in 2008 or 2004.

It will be effectively impossible on election day to determine if the outcome is the result of the ground game beating the polls or the result of certain polls being right and others being wrong. It could be the ground game, but I just don't know, and I don't know if we'll ever know. At best, we might get information by December or January as the big post-election polls and analyses come in.
   5487. Ray (RDP) Posted: October 29, 2012 at 11:04 AM (#4287247)
That said, I think that the "Ha ha ha" from the right will outweigh anger directed at Silver from the left in the event of a Romney victory.


And boo hoo to that. When you position yourself as the One Who Knows How To Do This The Best and you fail because others did it better, you deserve to be marginalized.
   5488. Ray (RDP) Posted: October 29, 2012 at 11:10 AM (#4287250)
And Joe can start kvetching once again, as Nate's just revised Obama percentage is back up to 74.6%, with 80.1% in the now-cast.


And going to decimal places is in itself false precision.

And (to anticipate the response) this is not like projecting Pujols to a .338 batting average, because people think of batting average that way, so there's aesthetic value in saying ".338." But there's no value at all in projecting to tenths of a percentage point for what Nate is doing.
   5489. Rickey! trades in sheep and threats Posted: October 29, 2012 at 11:10 AM (#4287251)
And boo hoo to that. When you position yourself as the One Who Knows How To Do This The Best and you fail because others did it better, you deserve to be marginalized.


So now Nate has committed the sin of arrogance or something? How very odd a critique. Especially coming from Ray. All Nate has done is apply the things he did well at BPro to political polling. The idea that he's some sort of evil mastermind is quite possibly the stupidest meme to emerge from this electoral clownshow season.
   5490. Matt Clement of Alexandria Posted: October 29, 2012 at 11:11 AM (#4287252)
When you position yourself as the One Who Knows How To Do This The Best and you fail because others did it better, you deserve to be marginalized.
So if Murray Chass put together a set of projections based on intangibles and feel, and he outperformed ZiPS one year, you'd say ZiPS should be marginalized and Chass knows better? We have to have a balance of methodology and results in evaluating things like this.

Also, I think that the RCP and Pollster models are cheating. They're currently projecting both a Romney win and an Obama win. By not aggregating state and national results, they can say they projected the election correctly either way. It's kind of bullshit.

Obviously, if Romney wins the popular vote and Obama wins the electoral college, by a relatively wide margin, then they'll be right. But I expect that they'll claim a win in pretty much any case - a Romney win, hey, they got the national polls. An Obama win, hey, they got the state polls. The 538 model deserves credit for trying to make sense of the divergence between state and national polls.
   5491. Matt Clement of Alexandria Posted: October 29, 2012 at 11:12 AM (#4287253)
And going to decimal places is in itself false precision.

And (to anticipate the response) this is not like projecting Pujols to a .338 batting average, because people think of batting average that way, so there's aesthetic value in saying ".338." But there's no value at all in projecting to tenths of a percentage point for what Nate is doing.
I do agree fully with this. Nate has a tendency to do this - the "breakout" chances on PECOTA were always extended into decimal places for no good reason other than marketing.
   5492. Ray (RDP) Posted: October 29, 2012 at 11:15 AM (#4287255)
Have you actually seen the people that are employed? Dick Morris, David Gergen, John Zogby, and that's just the very tip of a gigantic iceberg. Nate's competition for prognostication is a whole lot of people that couldn't predict a coin flip if they got to make two guesses.


But Morris and Gergen aren't pimping mathematical models; they're just giving their opinions, biased at that, and everyone realizes that their opinions really aren't worth a sack of peat moss.

Nate is doing something different, and is holding himself out as an objective prognosticator.

He can't get it wrong. How silly would it look if he misfires on this election (and yes, the presidential election is the only thing his reputation really hinges on), and then in four years he's back to quoting 74.6%. It would look ridiculous.
   5493. Ray (RDP) Posted: October 29, 2012 at 11:21 AM (#4287258)
So now Nate has committed the sin of arrogance or something?


I do think he comes off as very arrogant. Not that I care.

He came off as very smug on his Maher appearance the other night, nyah-nyah'ing about Romney. Granted most of that was probably because Maher was fellating him during the entire appearance, but still.

I see no reason to provide cover for Nate simply because he's One Of Us.

   5494. Ron J2 Posted: October 29, 2012 at 11:24 AM (#4287261)
Nate is not a pundit; he's someone who claims to have designed a model that works. If his model doesn't work, he doesn't work.


But here's the thing. He's not claiming the election is in the bag, just what .. 2-1 right now (though it was higher at various points)

It'll hurt -- mostly because most people don't understand arguments based in probability and thus (as you are doing here) restate his argument as "Nate calls Obabma"
   5495. Matt Clement of Alexandria Posted: October 29, 2012 at 11:24 AM (#4287262)
He can't get it wrong. How silly would it look if he misfires on this election (and yes, the presidential election is the only thing his reputation really hinges on), and then in four years he's back to quoting 74.6%. It would look ridiculous.
What do you mean by "can't"?

Obviously a good method can get a call wrong. This happens all the time. It doesn't wholly discredit the methodology. ZiPS isn't ridiculous when it misses a big call. Neither is the National Hurricane System ridiculous when it projects a storm to land 20 miles from its actual landfall. Uncertainty is the nature of the world. No "objective" system can account for all unknowns.

If you're saying that dumb people who write nasty letters to Dan about how stupid he was to project the Giants to only 85 wins will think that Nate is discredited, I can't necessarily disagree, but I doubt it will affect Nate's career terribly much. If you're saying that he'll be legitimately discredited, then you're entering into dumb-people-emailing-Dan territory.
   5496. Ray (RDP) Posted: October 29, 2012 at 11:26 AM (#4287264)
Darn it Dan, I was reading along and going to make fun of everyone who thought Nate might be unemployed if he got it wrong in 2012 and you post almost exactly what I was going to! Oh well, let me second what you said.


To clarify (I forget exactly what I said on the matter and I'm too lazy to go back and look) I don't really think he'll be unemployed/out of a job. There is value in what he does, e.g., in taking events like the 47% gaffe and analyzing how much of an effect they will have on the race based on objective data. His blog is very interesting, and is different and unique from what's out there. He deserves a lot of credit for that.

But I do think he will be marginalized as far as the overall predictions game goes. Wasn't there a polling outfit that misfired badly in 2008 and nobody takes them seriously anymore?
   5497. Matt Clement of Alexandria Posted: October 29, 2012 at 11:27 AM (#4287265)
But here's the thing. He's not claiming the election is in the bag, just what .. 2-1 right now (though it was higher at various points)
Also on point.

I have a six-sided di. There is an 17% chance it will land on four, and an 83% chance it will land on 1-3 or 5-6. If I roll it and it comes up 4, my probabilistic model has not been shown to be "ridiculous".
   5498. Guapo Posted: October 29, 2012 at 11:33 AM (#4287268)
Wasn't there a polling outfit that misfired badly in 2008 and nobody takes them seriously anymore?

There was this guy who erroneously called the AL Wild Card race in 2011.
   5499. McCoy Posted: October 29, 2012 at 11:38 AM (#4287270)
But I do think he will be marginalized as far as the overall predictions game goes. Wasn't there a polling outfit that misfired badly in 2008 and nobody takes them seriously anymore?


Sure, if Nate gets 30 or so states wrong and a bunch of Senate and House seats wrong he'll take a huge hit in credibility. But if Romney wins by a point and Nate gets 47 or 48 states right his credibility will only take a mild hit and the segment that will knock is largely the moron segment.
   5500. Ray (RDP) Posted: October 29, 2012 at 11:39 AM (#4287271)
There was this guy who erroneously called the AL Wild Card race in 2011.


And what happened to me? People made fun of me.

That's exactly the point. And people can claim he's not saying it's over, but (a) a 75-80% chance is pretty high, and (b) the liberals on this board were crowing when it was 80% that it was over, so they certainly took it as that before. All of a sudden now, they've backed off.

If it was 95% and Romney won, would people here concede that Silver blew it? What are we talking about here? Silver can't be graded? Then what good is he? Anyone can offer up a probability of who is going to win the election and if nobody can be wrong because it's a probability, then everyone is just as useless. Here: I say a 60.3% chance of a Romney win. And I can't be wrong, because that still leaves a 39.7% chance of an Obama win!! Woohoo!!
Page 55 of 62 pages ‹ First  < 50 51 52 53 54 55 56 57 58 59 60 >  Last ›

You must be Registered and Logged In to post comments.

 

 

<< Back to main

BBTF Partner

Support BBTF

donate

Thanks to
Dock Ellis on Acid
for his generous support.

Bookmarks

You must be logged in to view your Bookmarks.

Hot Topics

NewsblogOT: NBC.news: Valve isn’t making one gaming console, but multiple ‘Steam machines’
(980 - 1:29pm, Oct 30)
Last: madvillain

NewsblogOT: Politics, October 2014: Sunshine, Baseball, and Etch A Sketch: How Politicians Use Analogies
(4655 - 1:28pm, Oct 30)
Last: Bitter Mouse

NewsblogMadison Bumgarner, World Series legend - McCovey Chronicles
(62 - 1:27pm, Oct 30)
Last: Greg Pope thinks the Cubs are reeking havoc

NewsblogOT: Monthly NBA Thread - October 2014
(588 - 1:22pm, Oct 30)
Last: vagab0nd (no longer an outl13r)

NewsblogPrimer Dugout (and link of the day) 10-30-2014
(13 - 1:19pm, Oct 30)
Last: JJ1986

NewsblogJoe Maddon is to become Cubs manager, sources say
(98 - 1:19pm, Oct 30)
Last: McCoy

Newsblog2014 WORLD SERIES GAME 7 OMNICHATTER
(1438 - 1:16pm, Oct 30)
Last: Edmundo got dem ol' Kozma blues again mama

Hall of MeritMost Meritorious Player: 1959 Ballot
(11 - 1:05pm, Oct 30)
Last: neilsen

NewsblogHeyman: Pablo Sandoval is on Boston's 3B wish list, but so is Chase Headley
(30 - 1:05pm, Oct 30)
Last: villageidiom

NewsblogSan Francisco Giants at Kansas City Royals - October 29, 2014 | MLB.com Box
(67 - 12:57pm, Oct 30)
Last: snapper (history's 42nd greatest monster)

NewsblogRoad maps to pitching success in Game 7 | FOX Sports
(10 - 12:39pm, Oct 30)
Last: The Chronicles of Reddick

NewsblogVanguard after the Revolution | NBC SportsWorld
(52 - 12:13pm, Oct 30)
Last: McCoy

NewsblogStatcast: Butler chugs home
(1 - 11:38am, Oct 30)
Last: Davo's Favorite Tacos Are Moose Tacos

NewsblogJapan Times: Nakamura belts three-run homer in 10th to put Hawks one win away from Japan Series title
(10 - 11:32am, Oct 30)
Last: RMc is a fine piece of cheese

NewsblogAngell: The World Series is Almost Over
(1 - 9:22am, Oct 30)
Last: sotapop

Page rendered in 0.9087 seconds
52 querie(s) executed