1. The polls today a VERY slight movement back toward Obama and a TIE overall by any reasonable reading. Meanwhile, it’s true enough that Romney has hope in all the battleground states, but he’s slightly behind in most of them. Intrade remains astute in the suggestion that Obama has very slightly less than two chances out of three of winning in the electoral college, although the popular vote is anyone’s guess. There won’t be much movement, it seems to me, between today and election day, but any movement for any reason is very signficant.
2. I’m trying to make this post as flat as possible to rein in both my own flakiness and the occasional bout of anger that has marked this site in recent days. Everyone is free to say what he or she pleases, everyone is free to be lame or ironic or whatever, but nobody is allowed to take more than minimal offense at “the experiments in self-expression” that characterize the blog at its best and its worst.
October 29th, 2012 | 12:38 pm
Hi Peter:
Love your work here (and elsewhere).
But . . . I think you continue to be far to pessimistic regarding the election.
The RCP average has Romney up a point today, so far, which is as high as he’s been, iirc.
By my estimation (excepting Rassmussen’s tracking poll which has moved slightly to Obama) the movement continues to be towards Romney.
And in almost all the recent polls Obama is stuck at 46-48% — deadly territory for an incumbent this close to election day.
No poll has Obama up beyond the margin of error — in fact the only recent one to have him up by more than a single point has a whopping 4% margin of error.
Meanwhile the Gallup poll, with the smallest margin of error, has Romney up by 5% today.
Am I wrong in believing that Gallup, especially close to election day, is usually very accurate?
It’s looking like Romney by 2 at worst, given that its polls that are more than a week old that are keeping Obama afloat in the RCP index.
In addition recent tossup state polls seem to be showing sustained mvt to Romney — especially in Ohio where Romney is now up 2 in Rasmussen today — a massive turn towards him I think.
In any case, if he wins the national popular vote by 2-3% points he’ll win the electoral college too.
October 29th, 2012 | 1:02 pm
“I’m trying to make this post as flat as possible to rein in … the occasional bout of anger that has marked this site in recent days.”
I confess that this has certainly been noticeable on this site as Nov 6 approaches. And I have noted it even in my own demeanor if not the tone in some of my own posts.
I think this would be an interesting opportunity for some industrious pomocon writer to reflect on how elections not only prove to be opportunities for collective catharsis (that was already ably done on this site) but also moments where collective partisan agitation is brought out.
The benefit of elections as cathartic events are certainly true when the result is decisive, but as we approach Nov 6, what are we to expect of the possibility of a contested result, or a result that at least insufficiently satisfies both sides of the partisan divide?
And what is the responsibility of the reflective and thoughtful citizen in those moments of collective agitation?
October 29th, 2012 | 2:24 pm
I don’t know where the author gets his information, but a review of Drudge just now shows this:
Democratic Pollster Celinda Lake’s ELECTION PROJECTION: R 52% O 47%…
GALLUP MONDAY: R 51% O 46%…
OH: R 50% O 48%…
Separately, Rasmussen Tracking poll puts Romney at 50-46.
October 29th, 2012 | 2:31 pm
Is there any doubt a Romney administration would favor the rich and increase the income gap in our country? Mitt is a pariah in Mormon Clothing and will stop at nothing to expand an empire of greed for the rich in this country by expanding tax cuts for the 1% of the wealthiest Americans. He’s out to gut the Middle Class for every red cent he can get. Will his sacred Mormon underwear grant him the protection and money to buy this election? See for yourself as Mitt dons his tighty-whities sent down from the Good Lord Himself at http://dregstudiosart.blogspot.com/2012/05/mitt-romneys-magic-mormon-underwear.html
October 29th, 2012 | 2:31 pm
Just noticed the intriguing phrase “Postmodern Conservative”. Whatever “conservative” means these days, it can benefit from a sassy qualifier.
Unfortunately for conservatives (I mean this very sincerely and from my heart), in the long battle between knowledge and beliefs, science and knowledge appear to be winning, and conservatives appear to be relying ever more emphatically on dogma and beliefs to guide their thinking and their actions.
At times, scenes occur that are so surreal as to make one wonder if time warps aren’t all around us – take a deep breath and look at this intently. Yes, this is real, it is happening in the XXI century, in the US no less:
http://governor.state.tx.us/news/proclamation/16038/
Now, to the concern about Obama’s poll numbers and outcome predictions by the 538s of the world.
Without knowing the details of those models one can still make two general statements. One is that they don’t have sufficient data to properly calibrate them, regardless of their internals. Another is that parameters not captured by those models may play the decisive role this time around – such as last-minute deciders who “avalanche” one way or the other.
October 29th, 2012 | 3:04 pm
That second paragraph is priceless, Domingo. I’m glad to know it came from your heart, it makes the faulty thinking easier to tolerate.
Always a pleasure to have a member of the reality-based community show up and express surprise that people still pray.
October 29th, 2012 | 3:38 pm
…as a liberal who is subject to bouts of anger on the opposing end, I’m wondering why polls (which have proven unreliable at best… even the day before an election) would take precedence over… precedence. Honestly, I wouldnt have the hope that some of your readers do if my candidate were in Romney’s position… and it’s admirable, but unrealistic. There’s little doubt that if Obama wins Ohio, there’s no and if or buts. (As a Liberal who voted in NC … knowing it was futile, but knows my home states of NH and NV well enough to say “blue”… Harry Reid couldn’t lose in NV in 2010, and NH… who has never embraced Romney. there’s simply no way either state will go back Red. Not this yr) That being said, it’s just not prudent for ppl to imagine that Romney will take anything beyond Florida and NC … plus all of the States he’d need to to win the electoral votes he’d need. With all due respect, outside of the author … it is a bit unsettling that polls are given weight over common sense here?
October 29th, 2012 | 4:56 pm
marieke,
If I had as much confidence in my candidate as you seem to, I doubt I would be spending so much energy chastizing my friends on the other side of the isle in such spectacular fashion.
In my experience, when someone does that sort of thing it’s usually to help convince themselves of something they want to believe but are starting to, ever so slowly, doubt in their heart of hearts.
Regardless, welcome aboard the SS Pomocon. We’re all jolly here and love our friends from the otherside as long as they aren’t waving around large blunt objects or cling too much to straw man (or straw woman) arguments.
Not that I’m suggesting your last post is guilty of either, mind you.
October 29th, 2012 | 9:34 pm
“Without knowing the details of those models one can still make two general statements. One is that they don’t have sufficient data to properly calibrate them, regardless of their internals. Another is that parameters not captured by those models may play the decisive role this time around – such as last-minute deciders who “avalanche” one way or the other.”
This is simply hogwash, and to know that it is hogwash, you really do have to know something about what poll aggregators such as 538 actually do, AND what they don’t do. You don’t have to know everything, but you do have to know something.
It never ceases to amaze me how Conservatives can draw such grand and final conclusions from admitting that they know nothing whatever about the subject being discussed.
Once you buy into this sort of foolishness, of course, the trick is easy:
Without knowing the engineering involved, or knowing whether the material used is bubblegum, matchsticks, playing cards, or wrought iron, it is as plain as the nose on your face that a structure such as the Eiffel Tower is far too delicate and spindly to avoid early collapse.
So there, too!
Aggregators start with polling data over time, they manipulate it mathematically to create numerical values that reduce the variability in the raw data. The data is the data and requires no extra “calibration” to do math with it.
They do not start with “parameters”, they start with numbers and manipulate them by doing sums. Some aggregators, (in my opinion the best) such as Princeton Election Consortium or Electoral Vote.com, stick strictly to the numbers and the math and focus on the simple problem of the number of Electoral Votes a candidate is likely to win–51 pigeon holes and 538 pigeons to plug them with.
Other aggregators, such as 538 apply other, non-poling and non mathematical data to the already processed numbers, usually past electoral performance or general economic performance correlated to past electoral performance. And attempt to track aggregate popular vote percentages as well as electoral vote count.
What they do NOT do is draw specific conclusions about groups of actual voters, whether by party, lack of decisiveness, or popular avalanches.
(For the lonely reader out there who would actually like to know something about aggregators and their methods before drawing grand, overarching, and, indeed, Eiffel Towering conclusions, I strongly recommend the posts and comments at Princeton Election Consortium.)
Predicting election results is no more exact than predicting the path of a hurricane. But that doesn’t mean you go do something else and only announce the hurricane after it has already arrived.
You crunch the numbers you have and interpolate the result. The fact that there are types of numbers we have not yet obtained about either hurricanes or elections is irrelevant to the power of the numbers we have and the math we have to manipulate them with. That power is substantial, though not omniscient.
Of course, in the Truly General Case–which requires no fusty facts or specific information because the TGC is true, is general, and is the case–it is sufficient to impeach any analysis whose results you don’t like by pointing out that the analysis fails the test of omniscience.
October 30th, 2012 | 12:10 am
re: Intrade remains astute in the suggestion that Obama has very slightly less than two chances out of three of winning in the electoral college
You make it sound like Intrade sets the prices… it does not. Intrade is a market so the price is set by what buyers and sellers. Like NYSE, etc.
October 30th, 2012 | 9:27 am
“You crunch the numbers you have and interpolate the result.”
Since you’ve now multiple times in the past couple of days gone off the deep end attacking folks criticizing the polls, I feel I must ask you to please explain this sentence.
Because to my idiotic self, despite my Physics BS from Caltech and subsequent PhD, it appears to be the result of a random-scientific-term-generator, along the lines of the classic “flux capacitor” of yore. The words in the order they appear on my screen don’t seem to mean anything at all.
I mean, far be it from me to bring up tired old points of criticizing poll aggregators for attempting to use averaging to reduce systematic errors instead of statistical errors, or to perhaps question the validity of analogizing meteorological measurements to polling measurements, or to doubt whether systems calibrated on prior years can really be extrapolated forward to current electorates, or any of the other ways that folks with a minimum of statistical training and background could criticize how pollsters, or their aggregators, work.
But since you’re clearly such a super-genius who sees deep truths that the rest of us can’t comprehend, please just tell me what “you…interpolate the result” means in the context of our recent discussions, and what it has to do with anything.
October 30th, 2012 | 1:59 pm
Well, whatever my pretensions may or may not be, if human behavior in groups were not predictable by polling, nobody in the private sector would pay to have it done. And hurricane Sandy did not move out to sea and did not strike Mussel Shoals, Alabama. So I think I can say that I bring at least common sense and ordinary reality testing to the table. You can too, if you simply wait a week and compare the electoral results to the predictions.
(As an aside, I really don’t expect you, or most who write here, to have the nerve to do so if the results don’t meet with your desires.)
Once again, your implicit demand is for absolute predictive certainty, for omniscience, which simply cannot be realistically asked of anyone but God, and neither polls nor aggregators pretend to have it. What they offer are levels of confidence (a.k.a odds) in any given outcome.
What I meant by “crunching the numbers” is figuring out the odds. If my colloquial language confused you, I apologize. And when I said “interpolate the results” I meant making predictions, with odds, of who will win the election and by how much of some numerical value or other. I thought that this would be perfectly plain from context, but being wrong about something like that is simply one of the occupational hazards of writing in English.
As a human and not a Divine endeavor, polling and aggregating may in any given case be wrong, but across all cases they are, relatively, right. And nobody but God has absolutely right in stock.
Human behavior in groups is quite predictable, however much it may disturb the self esteem of folks here to be told that. Not as predictable, certainly, as Newtonian physics in the context of ordinary sensory experience, but still predictable none the less. And, when only our common sense and not our self-esteem is involved, we are perfectly well aware of this. We avoid crowds when we judge them likely to turn ugly. Our judgment of this isn’t perfect, but neither is it arbitrary, just like the pollsters and aggregators.
Beyond common sense, however, I bring to the table the understanding that when you count peas, the numbers you bring to the task are exactly the same ones you use to count buckshot pellets. And the mathematical operations you apply to the numbers of both sets of objects are the same thing, too. This is why math can tell us anything at all about the universe. The world may vary, but the numbers we use to describe the world don’t.
This is a little more abstract and sophisticated than common sense, but I don’t think it should trouble anyone educated in the sciences. Do you?
This also means that the numbers and mathematical operations are open to anyone to try. One of the reasons I recommend Sam Wang at PEC is he is quite transparent about the math he is doing and is generous enough to offer his own imput code for anyone to try it themselves.
Whatever criticism you may or may not make of Sam’s predictions, or anybody else’s, you do not appear inclined to examine the evidence of what they are actually doing in order to criticize them.
(I do notice that evidence adversity is a trait many conservatives share.)
It is perfectly possible, and not logically assailible, to generate an endless stream until Doomsday, of fresh doubts about anything at all and not ever bother to refer to the evidence offered for it.
There is a fine old name for this in the Canon Of Western Civilization. It is called Sophistry. And conservatives, frankly, are very lucky to have the Canon behind them since they actually practice Sophistry so constantly.
I’ve never tried to make sophistry my long suit. I much prefer evidence evaluation and reality testing. For one thing it actually gives you something interesting to read and think about, like PEC’s methods, from a perspective outside of your own prejudices about the results.
This really doesn’t require any sort of “super genius”.
But it does require a certain minimum of intellectual integrity.
October 30th, 2012 | 2:30 pm
“And when I said “interpolate the results” I meant making predictions, with odds, of who will win the election and by how much of some numerical value or other.”
Then, as I suspected, you don’t know what “interpolate” means, and you just thought you could randomly throw in somewhat impressive sounding words in a sad attempt to demonstrate you’re just soooo much smarter than anyone else.
I have no problem with someone being an insufferable, pompous, self-important, pedantic windbag. Heaven knows I do so at plenty of times.
But if that’s the pose you’re trying to play, you best get the very basics correct, or you just look like a fool.
As for the rest of your post–tl;dr. It has nothing to do with my point, which has been made.
October 30th, 2012 | 3:32 pm
Brian. Your last line – “Tl;dr” – is the absolute bane of internet commenting. It is literally useless except to indicate that one is lazy. You cannot state with any meaningful certainty that the rest of that post is irrelevant to your point… without reading it.
In fact, that reaction perfectly illustrates the point that the previous commenter DID make about your view – that items/ideas/models of limited or imperfect validity are avoided wholesale instead of analyzing them to the degree that they DO contain valid information.
Moreover, his use of interpolate is not inaccurate. Combining various polls from different pollsters at different points of time does function as interpolation – filling in informational gaps with supportable predictions and data from other, relevant sources. These aggregation methods increase the overall sample size (thus lowering uncertainty) and counteract any inherent biases or systematic errors that one pollster may be making by adding pollsters with opposite biases or potential systematic errors into the equation. If anything, Nate Silver’s (as an example) models are excessively prone to hedging and blurring of confidence intervals by using irrelevant economic data as if it were not baked into the public-mindset pie.
October 30th, 2012 | 3:32 pm
Brian,
I think Joseph Marshall meant “extrapolate” (not “interpolate”). Aside from that, everything he stated made sense to me.
For someone who supposedly has a PhD in physics, your argument and counter contain very little substance.
Sorry for the late response, I just found this site.
October 30th, 2012 | 4:48 pm
Actually predicting presidental elections is quite easy. I have predicted..exactly the last two. Each state has its own history and story..making the electoral college fascinating. 2012………obama 237 guaranteed romney 206…florida(29) romney by 3. fla has a obama distast
e..virginia(13) romney by 1.5 va trending obama with sou. Va more enthused..colorado(9) romney by 1.5 olmpic job well known in colo…obama 237 romney 257. Iowa(6) obama by 2. Obama mania 2008 started here..nevada(6) romney by 1.5. High mormon vote plus bad econ. New hamp(4) romney by 1.5. Mitt home . independents and nascar vote. Wisc
onsin(10) romn
ey by 2.5 vote. Dem walkout and
republin enthusiasm. Ohio(18) obama by 2 auto bailout. Final romney 276 obama 261
October 30th, 2012 | 5:14 pm
Well, if all you intended to do is correct my diction and call me names, you’re certainly welcome to. Anyone who writes is under obligation to write as well as he can and should be prepared to evaluate and make use of serious criticism from anyone. So, point noted.
As to calling me names, I’m not unfamiliar with the experience, but after 60 years I’m rather deadened to it. My grade school playground days were long ago, and responding in kind is simply a bore.
October 30th, 2012 | 5:50 pm
Sorry for the stat-nerdy post to follow, but I feel as though there are wide-spread misunderstandings of polling results. Let’s take an example – suppose a particular poll of likely voters in OH has a margin of error (MOE) of +/- 2 pts and shows Obama 50, Romney 46. This does not mean that the race is a statistical tie because Obama’s lead is within the MOE. The 50/46 result is the mean (or the top tip) of a normally distributed bell-curve of probabilities. The MOE simply puts bounds around the confidence interval (CI) of the poll – most polls use a 95% CI. The lowest end of the CI is a 48/48 tie, and the highest end of the CI is an Obama lead of 52/46. The takeaway of this poll is that we can say with 95% certainty that Obama is leading. There is only a 5% probability that the race is Romney is tied/ ahead OR that Obama is ahead by more than 6.
Of course the poll itself is subject to sampling error, i.e. its assumptions on likely voters is inaccurate. This is why different polls give us different results. However, if a poll shows even a 1 pt. lead for a candidate, then there is a higher than 50% probability of that candidate being ahead among those that the pollster considers likely voters – the true probability can be calculated based on its MOE and CI.
Sites like RCP, Nate Silver’s fivethirtyeight.com, the Princeton Election Consortium, etc. help filter out the noise of sampling error to project true probabilities by state. I think it is pretty likely that these two sites will not be off by more than a few EVs. Silver thinks Obama is a better than 7-to-3 favorite and Princeton thinks that Obama is a better-than 9-to-1 favorite to win. It seems to me that you’d be getting pretty good odds by betting on the President on Intrade today.
October 31st, 2012 | 8:43 am
Joseph Marshall in the last few days has written stuff like this: “Conservatives can draw such grand and final conclusions from admitting that they know nothing whatever about the subject being discussed…Of all the things most annoying about the Conservative mindset, one of the worst is this rhetorical habit of raising questions that could be easily answered by a little reading on the part of the questioner…The amount of sheer willful ignorance in venues like this of what happens when people are polled is both overwhelming and appalling…It is also one of the reasons that I call most Conservatives “fact challenged””
Anyone who comes to a conservative site and writes this sort of nonsense first better bother get his own stuff straight, which apparently was beyond you, and second certainly can’t whine about someone “calling me names.” Your sudden sensitivity is just too precious.
October 31st, 2012 | 4:10 pm
Brian, whenever I read those sorts of comments from Joseph’s posts I always like to go back to his contibution on this thread:
http://www.firstthings.com/blogs/postmodernconservative/2012/09/27/my-take-on-the-polls/#comments
when he said that following:
“I’m sure by now you’ve asked yourself, “Why is this Marshall character over here? He’s obviously a committed Liberal.” I’m here because most of you are intelligent Conservatives of the non-mouth frothing sort, who really ought to have an alternative to both what is in the Republican platform and the Democratic Party that they actually believe in. Probably you already do.
But I can tell you that it is so heavily encoded, self-referential, tied to an arcane set of books that no ordinary voter has read, and articulated in such vague and general terms that it is largely unintelligible outside of your closed circle.
Abstractly, and beyond my own partisanship, I think this is a shame. By definition, a democratic society needs a clear articulation of every point of view in order for the People to make an intelligent choice. The lack of one erodes the intelligence of even your opponents and is objectively bad for everybody.
And encourage you, even for my sake, to start doing the thinking you need to do to talk about the alternative you believe in clearly to everybody else.”
I do wonder what’s changed since that post. Perhaps Romney’s prospects? Perhaps, Mr. Marshall? Regardless, I’m sure he’ll be back to his good graces post Nov 6.
November 1st, 2012 | 7:11 am
Nothing has changed. I still enjoy reading the views of people here, even Brian’s. I prefer my prejudices sandpapered rather than curry combed. But that doesn’t mean that I’ve ceased to be partisan or that I’m no longer forthright about what I think to be nonsense.
I have not spoken my full mind about Mitt Romney because, even if he is elected President, he is a mere man and will pass from the scene in his own time and his own way. As will you, as will I, and as will even Brian.
But ideas and collective habits of mind endure, and I have no hesitation about rebutting those I think are bad ideas and bad habits. And I presume, since these comments are monitored, I have done so without excessive personal rancor to anyone. If this is not the case, then I apologize to any who feel personally wounded or offended by any of my comments.
Beyond that, I try to think as clearly and write as well as I can, without making a parade of my credentials, since I think none are relevant except that of being a citizen of a free and open society.
November 1st, 2012 | 10:16 pm
With a doctor of philosophy in physics and a pseudo-philosopher of metaphysics among us, presidential race polling seems so unworthy a discussion.
We should rather be exploring the cosmos, relativity of time, being, countin’ strings.
The heck with polls. Six days from now someone will be clearly right and
November 1st, 2012 | 10:20 pm
others will be clearly wrong. But it’s very unlikely anyone will ever prove the others wrong on the meaning of our existence or on string theory.
November 3rd, 2012 | 11:56 am
People, reality check. First, if Obama is up 3 in Oh and the margin of error is 3, that does NOT mean it a “statistical tie” as I heard CNN report. It COULD be a tie OR Obama could be up 6. That’s why they say, “plus or minus 3″ and not just “minus 3.” Journalists & pundits should take at least one stat course if they are going to opine on it.
Finally, the trend is clear and Obama ie either ahead or within the MOE in almost all swing states. Unless ALL the polls are systematically biased (they’re not), Obama will likely win. Anyone who denies that probability is simply deluding themselves. Could Mitt win. Sure, but it is far less likely than an Obama win.
Links
Blogs
Find Us
Contact