FiveThirtyEight: Politics Done Right: An Open Letter to Strategic Vision CEO David Johnson

9.29.2009

An Open Letter to Strategic Vision CEO David Johnson

by Nate Silver @ 11:19 PM

Subject: couple questions
From: Nate Silver 
To: xxx@strategicvision.biz

Hi David,

I'm writing you for two reasons.

Firstly, I wanted to make sure that you had some 
decent contact information for me.  Was really 
looking forward to the lawsuit and figured you 
might be having trouble getting in touch. I'm
at this e-mail address -- xxxxxxxx@gmail.com, 
or at xxx-xxx-xxxx.

Secondly, I wanted to provide you an opportunity 
to clear the air about one singular fact.  What 
call center(s) have you used to conduct your 
public-facing polling?  For every call center that
you're willing to publicly disclose, up to a 
maximum of 5, I will donate $538 to Children's 
Healthcare of Atlanta (http://www.choa.org/).

Best wishes,
Nate Silver

cc: FiveThirtyEight.com

268 comments

Peter said...: Haha, well played.; September 29, 2009 11:26 PM
shiloh said...: Excellent! ... Touché!

carry on; September 29, 2009 11:29 PM
Geoff said...: Nice try, however as a pessimist, I doubt they will bite.; September 29, 2009 11:34 PM
Mr. Universe said...: Hell, if he can come up with even one, I'll make a donation. I'm in.; September 29, 2009 11:35 PM
Mojowen said...: Nate, you're a vindicative bastard. I LOVE IT.

Keep going for blood old boy.; September 29, 2009 11:46 PM
Jake said...: Not sure what this is all about but, I'll jump on the bandwagon. Awesome! (For some reason.)

Also, a stupid nit. Only one space after periods, Nate. :); September 29, 2009 11:48 PM
LAW said...: This is seriously pathetic. Why, exactly, does this company have to disclose its call centers to you? And why do you keep picking on them? Most of your accusations consist of hand waving that you pass as "proof". This is what I expect from Republicans, not the guy who absolutely nailed most of the 2008 election numbers.

I have really lost a lot of respect for you, Nate.; September 29, 2009 11:54 PM
FakeJonStewart said...: you still haven't addressed whether or not strategic vision just samples a smaller other/undecided/no preference pool. if they typically poll close races and have 2-8 points of undecideds, they should always show up as 46-46 thru 49-49 races, as a general rule...

and you haven't even mentioned it. that's embarrassing.; September 29, 2009 11:55 PM
LAW said...: @FakeJonStewart

Don't hold your breath - he hasn't addressed any of the criticisms against his methods. Who has time for boring proof when you have witches to hunt?; September 29, 2009 11:59 PM
Mark Grebner said...: This shifts the terms of the debate somewhat: The rightmost digit frequency disparities only raise the question whether some of the published polls were not conducted as presented. Under Nate's challenge, SV-LLC can collect if any of its published polls were legit.

This feels a lot like Elizabethan dueling. The insulted party must either fight or be branded a coward. And in this incarnation, the worst that can possibly happen to Nate is that he's embarrassed and helps a worthy charity.

I like it.; September 30, 2009 12:00 AM
whitesoxfan said...: Guys, actually look at Strategic vision's polls. They don't poll mostly two-way races with 2-8% undecided.; September 30, 2009 12:00 AM
Juris said...: @Jake: Nate's using a typewriter from the 1960's....

@Nate: If David sends me a copy of the system file and full documentation (questionnaire, sampling frame, response rates) for the surveys that AAPOR asked for information about, then I will donate 53.8 cents per respondent to those surveys -- also to Children's Healthcare of Atlanta.; September 30, 2009 12:05 AM
Regret said...: @LAW

The fact that Nate cannot prove that the polls are fraudulent does not mean that they are not. He is only showing dozens of reasons to suspect that they are fraudulent and asking for the pollster to prove otherwise.

This is similar to the Madoff case. Harry Markopolos, the whistleblower who repeatedly reported Madoff to the SEC, did not have the data to prove Madoff was committing fraud, but he made a very strong case for why he believed it was.; September 30, 2009 12:15 AM
Pan said...: @LAW: "This is seriously pathetic. Why, exactly, does this company have to disclose its call centers to you? And why do you keep picking on them?"

You can say he's going overboard, but I think at the core there are very good reasons for Nate to be pursuing this with the fervor that he is. Nate runs a site that's all about aggregating and analyzing polls. When a pollster appears to be making up data, this causes Nate's site to be less accurate.

If Nate excludes a pollster that frequently winds up polling on the conservative side of the median, conservatives will claim his bias is driving his numbers. This would be death to his credibility. Therefore, he needs to seriously call someone out rather than just ignore them. And when that someone makes vague legal threats, he can't just back down or once again his credibility is shot.

I hope you can see the logic in all of this.; September 30, 2009 12:26 AM
Juris said...: @Mark Grebner. I think there is no, zero, zilch chance that Johnson CAN produce the data I mentioned. If by some miracle he does, then I would ask him to send it to you, and I would give you the cash to donate to the charity. You don't know me, but I know how to reach you. And I'm sure you would enjoy picking through the data. I prefer to remain anonymous.; September 30, 2009 12:26 AM
LAW said...: @Regret

I have seen nothing that I would consider "suspicious" enough to merit the witch hunt that Nate has decided to embark on. Accusing a company of widespread fraud is *very* serious business and really shouldn't be done on a hunch. Nate has used much of the goodwill he has earned by nailing poll numbers to drag SV's name through the mud to the point where many people will never trust their results again. And what proven wrongdoings is SV actually guilty of? None whatsoever. It's pretty shady business as far s I'm concerned.

Frankly, the most suspicious behavior being exhibited here is that of Nate Silver. What is his beef with Strategic Vision? I am pretty sure if he turned such intense focus on *most* of the other polling companies he could find very similar "anomalies" in their data. Why Strategic Vision? What's the deal?; September 30, 2009 12:32 AM
somasoda14 said...: Plenty of bashers on here who want to look smart and discerning by expressing their deep, paternalistic "disappointment."

You guys think you know more about polling than Nate Silver, when all you are is a commenter on his blog. His criticisms are open, humble, and sound, and keep in mind, he

a.) was not the one who initiated the investigation of Strategic Vision (AAPOR was, dipshits)

b.) has at every point made it abundantly clear that he has (grounded) suspicions and nothing conclusive, certainly far from "witch-hunting" as one of you morons thoughtlessly puked

c.) has received NO cooperation or compliance from David Johnson (or anything other than hostile dismissal), nor has AAPOR or anyone else -- or are they just in on the conspiracy as well??

d.) has been, rather than being engaged or corrected by David, threatened by him with a LAWSUIT

So maybe you idiots should think about all that before you get all pissy and judgmental just to make yourself look and feel judicious. Or rigorous. Or whatever science-y adjective that, when you can tie it to yourself in your head, helps you go another day without coming to terms with your overwhelming doucheness.; September 30, 2009 12:34 AM
somasoda14 said...: @LAW: "I have seen nothing that I would consider "suspicious" enough to merit the witch hunt that Nate has decided to embark on."

Well, people come to 538 to read expert analysis on polls. In short, nobody gives a shit whether you think you know more about polling than Nate Silver. If you do, prove it and start your own website where you predict a presidential election better than everyone else.

Or at least demonstrate something other than your capability for being an overwhelming loser who values his own opinion much too highly on technical issues he knows nothing about.; September 30, 2009 12:39 AM
Zee said...: Dear Nate,

Your accusations cannot at this point be absolutely proven. Therefore, it is totally irresponsible and clearly a witch hunt for you to try to disprove said accusations. I hereby demand that until you have proven fraud by Strategic Vision, you cease investigating them and giving them opportunities to prove you wrong!

Fair and Balanced Commenter; September 30, 2009 12:39 AM
Michael said...: Guess what, "LAW".

Nate has dragged SV's reputation through the mud, but he could clear it again in a day. It'd be easy.

All he'd have to do is say "They have provided satisfactory explanations for the anomalies I found in the data, and I am now convinced that they are legitimate."

And I am quite convinced that if they did provide Nate with such explanations, he'd state so promptly.

But they haven't, and it seems quite unlikely that they will.

Why? Why do you think that SV doesn't want to tell us any basic information about their old polls? I mean, recent polls, sure. Recent polls are trade secrets and all that, But the crosstabs on some of the old polls are hardly important matters, are they? They could surely release those to help clear the air - why don't they?

How do you figure the results from the Oklahoma poll are even remotely plausible, either?; September 30, 2009 12:41 AM
somasoda14 said...: @LAW

"I am pretty sure if he turned such intense focus on *most* of the other polling companies he could find very similar "anomalies" in their data."

Why don't you PROVE it, instead of being a fucking asshole who makes sweeping claims with no justification?

Especially while YOU'RE ACCUSING SILVER OF LACKING EVIDENCE FOR HIS CLAIMS. YOU SUCK SO BAD JESUS; September 30, 2009 12:44 AM
Dan Szymborski said...: Jesus, Nate, whose bright idea was it to send this letter? You're never going to get an answer now based on this letter and by writing this letter and immediately posting it, you've upped the stakes on the personal level.

If someone wrote a letter demanding you justify the Matt Wieters projection and posted it on their website, would you respond? Of course you wouldn't (and in fact, Colin Wyers sort of did this at THT and while you might not have been involved, I know for a fact that a few of the BP guys were pissed off).

You gotta take the high road here. Gotcha points aren't that useful, they usually just make the person look petty. What's the end game here? Getting the information that may or may not exist or winning a flame war?

And yes, I note the irony of complaining about writing a private letter and posting it publicly, but I'm at Panera and don't have my normal address book on my laptop. You should know that I don't take potshots at you, either.; September 30, 2009 12:44 AM
Joey S said...: I would like to interrupt this comments-section-argument to say HAHAHAHAHAHAHA. That letter was hilarious.; September 30, 2009 12:54 AM
shiloh said...: Self-righteous indignation or winger trolls upset Nate has probably won again, like the '08 election?

You make the call.

hmm, witch hunt, seriously pathetic, embarrassing, pretty shady business.

Nate appears to have inside info he may not be disclosing re: SV who appear to be all bark and no bite, much like the winger trolls at 538 ...; September 30, 2009 12:57 AM
Inferno said...: I predict reasoned and civil discussion where all parties are open to consideration of all viewpoi...

...oh. And a bonus appearance by proofreaders, to boot!

I will state that the smackdown was amusing, though. Albeit, a bit puerile - Ras or Selzer doesn't publicly divulge which call centers they use, AFAIK.; September 30, 2009 12:58 AM
LAW said...: @Pan

"...When a pollster appears to be making up data..."

And by what method did you come to this conclusion? That's exactly my point. Nate is well-respected, and when he asserts something, people like you believe him. That's why it's important that he not run around throwing about accusations without proof. And let me state it again: There is NO proof in this matter that SV has done anything fraudulent. There are some anomalies, but what Nate hasn't bothered to do is show that these kinds of anomalies don't exist among *all* polling companies.

And by the way, it proves absolutely nothing that SV has opted not to respond to the accusations and bullying tactics of this blog. By the logic some of you people are using, the president must have been born in Kenya because dozens of blogs have asked for the "real" birth certificate and have gotten no response from the White House. Please...don't stoop to that level.

As for somasoda - you're an angry little guy, aren't ya? Jeez.

By the way, for people who are implying that I'm some troll, I've been reading this site since Nate came over from dKos last year. If you need proof, here are all my comments. The general level of civility and objectivity here has been refreshing, and it just sucks to see it spoiled.; September 30, 2009 1:04 AM
juvanya said...: While SV deserves to be fully investigated, I do not think this is your job, Nate. It gets you into a mess. No matter how correct you are, it will mark you as partisan and that is something to be avoided.; September 30, 2009 1:04 AM
Mike in Maryland said...: A shout-out to Nate on the Rachel Maddow show tonight in the second 'Holy Mackerel' story in that segment of her show.

To see it, go to her website at msnbc.com (http://www.msnbc.msn.com/id/26315908) and click on the show for September 29, then the segment titled "Baseball prediction freakily accurate. The first 'Holy Mackerel' story is about changing ringtones on China Mobile phones - the second story (beginning just after the 1:35 mark) is the baseball prediction. The shout out to Nate is at the very end.

Mike in Maryland; September 30, 2009 1:11 AM
Steve Singiser said...: Law--

You wrote: "I am pretty sure if he turned such intense focus on *most* of the other polling companies he could find very similar "anomalies" in their data."

Actually, he already did that. A couple of days ago. Using Quinnipiac for a variety of good reasons (similar states, for example), he found their distribution to be far more random. A reader in the comments did SUSA...same conclusion. A far more random distribution that with SV.

As it stands, Nate has a right to fire off on SV.

Due respect, but your conflation of this to the birther thing is in error--the birther thing is inane because a mountain of evidence exists to counteract the charges. There IS a birth certificate, as well as birth announcements in Honolulu papers, the verification of everyone in Hawaii up to and including their GOP Governor, etc.

Conversely, SV has offered *nothing* as of yet to counteract the charges.

Tangential note--SV's counterclaims to the AAPOR condemnation were just insipid. Johnson claimed AAPOR was trying to upstage his release of...a totally meaningless GA primary poll nearly a year before the election. As if YET ANOTHER poll showing Barnes and Oxendine leading in Georgia was game-breaking information that the AAPOR was intent of sabotaging. Just bizarre.; September 30, 2009 1:12 AM
LK said...: Snarky. I like it.

This isn't about partisanship (and, by the way, the right already thinks you're a partisan, and will continue to think so no matter what you do). If you can out an organization like this, it'll vindicate you as more than just a "smarty pants smarty pants" (Is that what Dylan Ratigan called you? What a weirdo) - it'll make you a journalist.

Who says math can't solve mysteries? It's like a real life episode of that show Numbers, which I've never seen, but from what I understand, my simile is valid.; September 30, 2009 1:13 AM
Juris said...: @Dan Sz: PECOTA definitely overestimated Wieters (mainly probably because Davenport's "input" translations that it relied on), but (a) Wieters has had quite a good rookie season, and (b) in the Wyers article that you refer to I find some other projected OPS numbers that are also wildly inflated, with THT's in particular being way out of line.; September 30, 2009 1:14 AM
Norman Kittrell said...: Dear, Nate
The letter does not become you. It seems to me like something written in anger that should be slept on before being sent. I have always and still do respect your genius and general civility; one discretion couldn’t change my mind. I would only offer this political quote.
“Don’t wrestle a Pig. You’ll just get dirty and the Pig Enjoys it.”; September 30, 2009 1:15 AM
Mason said...: While SV deserves to be fully investigated, I do not think this is your job, Nate.

Who's gonna do it? You? You, Lieutenant Weinberg?

It gets you into a mess.

He's a big boy.

No matter how correct you are, it will mark you as partisan and that is something to be avoided.

Since when is verifiability and accurate methodology a partisan issue? As others have pointed out, Nate would (justifiably in my mind) draw criticism if he simply excluded the right-leaning SV polls without cause.; September 30, 2009 1:18 AM
Zee said...: Dear Nate,

Giving people both opportunity and motivation to disprove your well founded (although admittedly not air tight) accusations about their polling data is totally irresponsible. The responsible thing to do would of course be to just allow Strategic Vision to continue on its merry way whether it falsifies data or not.

Companies should be left alone until they are absolutely, undeniably guilty. You are not even allowed to investigate their guilt until they are proven guilty, as that would be totally unfair. Please stop these posts so I can open up my web browser without having another panic attack.

Concerned Commenter; September 30, 2009 1:20 AM
shiloh said...: juvanya said...

While SV deserves to be fully investigated, I do not think this is your job, Nate. It gets you into a mess. No matter how correct you are, it will mark you as partisan and that is something to be avoided.
~~~~~~~~~~

SV drew first blood by threatening a lawsuit after Nate said that there might be something screwy in Denmark ie he made no proclamation. He only paraphrased Dr. Henry Lee, "There something wrong here."

SV made the first move and it appears to have come back to bite them in the ass!; September 30, 2009 1:22 AM
Zintis said...: juvanya: "No matter how correct you are, it will mark you as partisan and that is something to be avoided."

Oops - too late.

Seriously - are there any 538 visitors who still come here looking for non-partisan commentary? I doubt it. This post was pretty lame, but really not all that surprising...; September 30, 2009 1:24 AM
Mason said...: Seriously - are there any 538 visitors who still come here looking for non-partisan commentary? I doubt it. This post was pretty lame, but really not all that surprising...

We've got to keep things "fair and balanced," doncha know?

The commentary is partisan, but the numbers speak for themselves.; September 30, 2009 1:36 AM
shiloh said...: Trying to think of someone involved in politics who isn't partisan.

hmm, I'll try to get back to you all on that one ...

take care

p.s. I enjoy Nate's blog because it's a progressive blog, shocking, eh.; September 30, 2009 1:38 AM
Skip said...: Well, back in the dark ages of typing class, I learned that there are two spaces after a sentence-ending period to distinguish it from an abbreviation-ending period.

Did that change sometime since the Mac Classic came out?

On a previous thread about SV (BTW, very enlightening, and I've already sent them to my colleague the stats prof) someone suggested not feeding the trolls...just let them stay under the bridge.; September 30, 2009 1:40 AM
Skip said...: Well, back in the dark ages of typing class, I learned that there are two spaces after a sentence-ending period to distinguish it from an abbreviation-ending period.

Did that change sometime since the Mac Classic came out?

On a previous thread about SV (BTW, very enlightening, and I've already sent them to my colleague the stats prof) someone suggested not feeding the trolls...just let them stay under the bridge.; September 30, 2009 1:41 AM
Skip said...: Well, back in the dark ages of typing class, I learned that there are two spaces after a sentence-ending period to distinguish it from an abbreviation-ending period.

Did that change sometime since the Mac Classic came out?

On a previous thread about SV (BTW, very enlightening, and I've already sent them to my colleague the stats prof) someone suggested not feeding the trolls...just let them stay under the bridge.; September 30, 2009 1:42 AM
LAW said...: @Steve Singiser

The issue is that of a proper control group. It is very possible that some significant number of pollsters show the same "anomalies" in trailing digit analysis as SV does. Nate provides a control group of size 1 (I am not aware of the SUSA analysis you speak of).

Nate of all people should have known he needed a control group, and he could have provided a nice scatterplot of a number of polling companies' trailing digit analysis similar to the ones we've seen on this site many times over. If SV was an obvious outlier with some degree of confidence, THEN he could start accusing them of fraud. That would have been the responsible way to conduct this analysis.; September 30, 2009 1:42 AM
John said...: What the heck does LAW mean when he says that there has been no proof? Nate Silver has been showing us his work in a clear and transparent fashion. His work is straightforward enough that readers familiar with statistics can easily confirm he has a right to be suspicious about the trailing digit data or scholastic poll. But even if you aren't familiar with statistics, the fact that S.V. does not have the offices one would need to run a major polling firm is fairly easy to understand.

Nate has been careful, methodical and transparent throughout this process. Calling this a "witchhunt" merely shows that you consider any investigation to be unfair.; September 30, 2009 1:43 AM
Steve Rogers said...: At this point in the debate the evidence that Nate has provided, while not proof of anything,certainly outweighs the nonexistant evidence coming from SV. The burden is now on SV to refute the allegations. They have not done so, which says a lot in and of itself. It would be one thing for SV to take the "we don't have to answer to some stupid blog- we're not even going to dignify that with a response" route. However, they did choose to respond, only instead of respondign with evidence of their own, they responded with threats of legal action. If they were a legitimate polling organization, they could make all of this go away instantly. The fact that they didn't provide the info originally makes them seem unprofessional. The fact that they won't release it now makes them seem suspicious. There is no "witch hunt." Strategic Visions brought all of this upon themselves.; September 30, 2009 1:44 AM
Ahab said...: Concern trolls are concerned.; September 30, 2009 1:48 AM
Aaron said...: Strategic Vision uses a .biz TLD!?

I thought .biz was prima facie evidence of fraud. Why even bother to question their numbers after that? :-); September 30, 2009 1:49 AM
T. J. Hairball said...: And now we move from the "polite academic" stage of the confrontation to the "showman and gambler" stage of the confrontation.; September 30, 2009 1:49 AM
Bradford said...: I was polled by Rasmussen tonight and now KNOW they are biased bastards. The question on healthcare asked if I thought the healthcare plan of "Obama and the CONGRESSIONAL DEMOCRATS" with a nice evil emphasis on "congressional" would lower costs, etc. Also, the the question order, emphasis, and wording was aimed to make you both pick Republican and avoid being a liberal. It was genius polling - how to be biased while acting unbiased.

Question - is fake polling worse or using your polling skills for evil (e.g. to create biased polls to effect the public discourse)? I call it about even...

My view - Strategic Vision might poll 30 people (all the husband and wife can reach) and extrapolate to a full sample, while Rasmussen calls a full sample but makes sure the question bias skews the results...they both suck and they both use polls to make money through lies. Essentially they are the same animal though. All you biased pollsters with huge house effects on both sides need to think about that.; September 30, 2009 1:59 AM
Bradford said...: I was polled by Rasmussen tonight and now KNOW they are biased bastards. The question on healthcare asked if I thought the healthcare plan of "Obama and the CONGRESSIONAL DEMOCRATS" with a nice evil emphasis on "congressional" would lower costs, etc. Also, the the question order, emphasis, and wording was aimed to make you both pick Republican and avoid being a liberal. It was genius polling - how to be biased while acting unbiased.

Question - is fake polling worse or using your polling skills for evil (e.g. to create biased polls to effect the public discourse)? I call it about even...

My view - Strategic Vision might poll 30 people (all the husband and wife can reach) and extrapolate to a full sample, while Rasmussen calls a full sample but makes sure the question bias skews the results...they both suck and they both use polls to make money through lies. Essentially they are the same animal though. All you biased pollsters with huge house effects on both sides need to think about that.; September 30, 2009 1:59 AM
kinkorknight said...: I'm almost disappointed that the comments are back on. The flame wars as of late have been getting irksome (and I freely admit I got involved. My sin. I'm moving on).

I'm of two minds here. On the one hand, it's a fun little email. On the other, I doubt it will accomplish anything. Sean Hannity didn't submit to waterboarding when Olberman put out a similar offer, though another conservative pundit did go through it.

I do find it fascinating how we still get folks who say "who are you to demand such information," as if the right to demand information is conferred by the gods. He can't legally demand information, but he sure as hell can ask for it. Arguing against that fact merely because it doesn't appeal to your own beliefs is sophistry of the worst kind. Sadly, it's a well-worn practice, especially these days.; September 30, 2009 2:01 AM
LAW said...: @John

I mean exactly what I say. There is no proof. The trailing digit analysis means almost nothing in the absence of a control group. The location of the office is complete conjecture. Who's to say that they don't have a small office in a strip mall for tax purposes and then do their polling elsewhere? And how do we know that other polling firms don't do this as well?

The general problem here is that Nate is picking on this one particular company without really having a baseline to judge against. Things that look "suspicious" might just be par for the course. In the end, he simply doesn't have the point of reference needed to make the kinds of accusations he is trying to make.; September 30, 2009 2:03 AM
Pan said...: Re: two spaces after the period

I know this is a tangent, but the topic recently came up and I found out that everything I knew was wrong.

http://www.chicagomanualofstyle.org/CMS_FAQ/OneSpaceorTwo/OneSpaceorTwo03.html

Generally, the whole two spaces after the period was from people BREAKING the normal rule of one space. And it was mostly an American thing. I have to wonder if HTML would have ignored whitespace by default if it had been invented by an American who had typing class in the 50s rather than a Brit.

Next thing you know, you'll be claiming that it's incorrect to put a period outside a quoted piece of text at the end of a sentence, like "this".

Look that "rule" up to - it may surprise you.

(Of course, it's pretty obvious that Nate meant to have two spaces, since he did it in other places in the letter. Then again, this reinforces that the CMS statement about it not being worth it because of the extra hassle.); September 30, 2009 2:16 AM
Bradford said...: LAW-

...and the moon might be made of green cheese - do you personally have a moon rock and can prove it came from the moon? That face on MArs is real too, I swear it and the CosttocoastAM third graders say it is real, please prove tobe beyond a doubt it is not true. Also, while you are at it, prove to me that the Strategic Vision offices that all map to UPS Stores are truly huge suites owned and occupied by real people calling real people from call centers.

Proof is a difficult when proving a negative, but live in fantasy land and ignore the facts. You are no better than Glenn Beck...

As to polling proof - please prove to me the house effect at Rasmussen shows real bias, prove it to a certainty. Also, prove to me the Iran and Afghanistan elections were biased enough to change the result, please, prove it to a certainty.

Get over yourself, smart guy.; September 30, 2009 2:17 AM
Mike in Maryland said...: Jake said...
Also, a stupid nit. Only one space after periods, Nate.

I agree - a stupid nit.

People write in order to communicate. Communication means the writer AND the reader need to be able to comprehend what is written. The inclusion of a double space after sentences, in my opinion, assists in that comprehension, as that double space clearly separates one sentence from the next. This is especially important when the person reading the material has some type of visual impairment, temporary or permanent.

The US government follows the dictates of the GPO Style Manual. In the 1959 edition, it specified that sentences should be em-spaced even when typeset, and defined a double-space as a synonym for an em-space.

Other people have differing views on this subject. The four main points of view can probably be summed up as follows:

1. Some authors believe that text using double-spaces between sentences is more readable than text written with only one space after the period.

2. Some authors believe that proportionally spaced fonts have made double-spacing redundant, and that it should only be used in a monospaced (nonproportional) font. The argument here is that double-spacing was an attempt in a monospaced font to create the effect of proportional fonts' spaces, and that the ready availability now of proportional fonts renders a double-space redundant. However, the width of spacing in a proportional font width is nearly always narrower than an em space. This argument is supported by renowned typographer, Ben Worrall. When questioned about the practice of double spacing, he replied: "It makes me physically sick."

3. Some authors (particularly professional designers or typographers) believe that double-spacing creates an unappealing appearance. [Personal note - this is style over purpose. Written communication is not about appearance, but about communicating. If appearance does not interfere, then it should be of no consideration.]

4. Some authors believe best practice is dependent on the particular typeface being used.

And finally, the Modern Language Association (MLA) tolerates either one or two spaces after a period, colon, or semicolon (http://www.mla.org/style_faq3).

So yes, a stupid nit to pick.

Mike in Maryland; September 30, 2009 2:20 AM
LAW said...: @Bradford

How am I asking someone to prove a negative? Nate has more or less made the assertion that Strategic Vision is committing fraud and releasing poll numbers that are not legitimate counts of voter intention. That is a positive statement, and one that can certainly be proven with some degree of confidence given the right methodology.

I simply want to see some proof that isn't massively flawed. And since you seem to be such an expert on SV's offices - I need to ask: Have you ever even stepped foot in Blairsville, GA? Or does all your expertise on the subject come from the infallible knowledge of The Google? Just curious.; September 30, 2009 2:32 AM
Mike in Maryland said...: A follow-up note to my post of 2:20 AM:

I inserted a double space after each and every period in my post, but the software that posts the material compressed all spaces of more than one space to a single space.

I know that this type of software also converts a tab space into a series of spaces, then lops off all spaces greater than one, unless that space is at the beginning of a paragraph, in which case it lops off ALL spaces prior to the first character of the paragraph.

In other words, the software dictates how your post will appear, and there is very little you can do to prevent it from doing that.

Mike in Maryland; September 30, 2009 2:32 AM
Bradford said...: LAW-

I asked you to prove it. Prove to me the offices are real, not using Google. Prove it, to me personally. Prove that SV polling is real and well done. Just prove it smart guy. You are really brainy, just quit sitting outside the tent and pissing in and try to prove your point.; September 30, 2009 2:35 AM
Bradford said...: LAW-

I still await actual proof that you can prove anything about polling to an actual certainty. The Iran thing is easy , right? For a guy as smart as you...

My moon rock point was an easy one to prove, right? Just go to the moon and get a rock and then actully prove to me it is real after you get back.

Get it? "Proving" anything to a certainty is very hard, that is why intelligent people attempt to understand science and proof and don't sit around and "Glenn Beck" other people points.; September 30, 2009 2:40 AM
shiloh said...: LAW said...

I don't know where this came from, but this whole escapade makes me embarrassed to be a reader of this site.

September 28, 2009 2:28 AM
~~~~~~~~~~

Again, if one is embarrassed, stop reading and posting, otherwise one is not embarrassed and just a concern troll.

take care; September 30, 2009 2:41 AM
Bradford said...: To those who think the comments section here is out of control - LOL! You should have been here BEFORE the election...; September 30, 2009 2:44 AM
TFLive said...: Just so I understand it, LAW is demanding transparency from Silver...but thinks its a witch hunt to demand the same thing from Strategic Vision?

One of these I will meet a die hard Republican that isn't a hypocrit (everything allowed only when my team does it). There has to be one out there, the odds demand it, but its been years now and I am starting to lose hope.; September 30, 2009 2:45 AM
shiloh said...: Bradford said...

To those who think the comments section here is out of control - LOL! You should have been here BEFORE the election...
~~~~~~~~~~

or yesterday, just sayin'; September 30, 2009 2:46 AM
Bradford said...: sorry shiloh, you could be right. I was not on much yesterday - I actually had work at work and watched some Dexter episodes on Netflix last night...; September 30, 2009 2:50 AM
shiloh said...: Bradford said...

sorry shiloh, you could be right. I was not on much yesterday - I actually had work at work and watched some Dexter episodes on Netflix last night...
~~~~~~~~~~

Actually it was yesterday and the day before and yes before the election the ad hominems were flying fast and furious :) as I was a casual observer back then.

So hard to keep track, stayed on the sidelines mostly lol was mainly at Kos and RCP back then.

Yes Virginia, the party of No! still hasn't come to grips an African/American family is living in the White House.

As Colonel Kurtz would say, The horror ... the horror! ;); September 30, 2009 3:00 AM
LAW said...: Sigh...

@Bradford

Obviously, I can't prove any of the things you ask short of flying to Georgia. Never said I could. I am simply stating that I, and you and Nate for that matter, also cannot provide any real proof of fraud. And for what it's worth, where is your demand that we hunt down the offices and methodology of Zogby, Research 2000, CA Field Poll, et al? Why this one particular pollster?

@TFLive

I am not a "diehard Republican." I am a liberal Obama supporter that has been following this site since its inception. I gave the link above showing this and I'll give it again. I guess I didn't get the memo that being a liberal means that one has to join in on lynch mobs based on spurious evidence...; September 30, 2009 3:06 AM
Mike in Maryland said...: LAW said...
@Bradford

How am I asking someone to prove a negative? Nate has more or less made the assertion that Strategic Vision is committing fraud and releasing poll numbers that are not legitimate counts of voter intention. That is a positive statement, and one that can certainly be proven with some degree of confidence given the right methodology.

Idiot - Nate has NOT made any assertions that SV LLC is committing fraud. What Nate has presented is indications that the information that SV LLC has presented cannot be easily explained, and has always invited SV LLC to present the cross-tabs to show that the polls they have released can be supported by the background information (i.e., the cross-tabs).

Nate is not the only person, nor the first, to ask for this information. After the 2008 New Hampshire Democratic Presidential primary, when the polling was very far off, AAPOR asked 21 different polling organizations to show how they conducted their polls, and try to explain what went wrong. All but SV LLC responded, and responded in a quicker time-frame than AAPOR asked them to respond. SV LLC didn't respond, and when AAPOR extended the time period, SV LLC didn't respond, again. AAPOR asked again, and SV LLC didn't respond.

Why did AAPOR request the information? To see if something was wrong with the current practice of polling, and if so, what might be done to make polling more accurate in the future.

The Atlanta Journal-Constitution also asked for supporting documentation from SV LLC to back up the results of polls that SV LLC conducted. SV LLC ignored the AJC's request for information.

It was after the above two organizations asked and were refused the information that Nate then started to investigate.

Has Nate PROVEN that SV LLC has not reported accurate information? No. And that was not his intent. What he has shown is that the results do not seem to be able to be accurate, as analyzed in many different manners.

SV LLC can end this real quickly and easily - release the cross-tabs that will, in the words of Matt Towery, CEO of Insider Advantage, ". . . prove that you’ve got a poll".

But instead of releasing the requested information (cross-tabs), SV LLC responds with a "We've got a call into our attorney about filing a law suit against Nate Silver" comment.

Now, SV LLC says that it will release cross-tabs for future polls.

The reason for the focus on SV LLC, though, goes back to the beginning of 2008, just after the 2008 New Hampshire Democratic Presidential primary, and thus calls into question each and every poll released by SV LLC to date, including the polls of students in Arizona and Oklahoma that purported to show that 77% of the students in each state don't know that George Washington was not the first President of the United States.

A lot more background can be found here:
http://blogs.ajc.com/political-insider-jim-galloway/2009/09/27/strategic-vision-promises-crosstabs-with-every-poll/?cxntfid=blogs_political_insider_jim_galloway

Mike in Maryland; September 30, 2009 3:08 AM
beavis said...: I hereby demand that until you have proven fraud by Strategic Vision, you cease investigating them and giving them opportunities to prove you wrong!

So let me see if I understand you. Unless you have proof, you can't investigate? I guess the police better not investigate that murder until they can prove who did it?

How can he prove fraud without investigating?

I wonder if SV knows that paying people to shill is actually illegal. Lots of new names here trying to defend SV. hmmm; September 30, 2009 3:19 AM
can8tiv2000 said...: When a purported poll's results has serious policy implications then knowing a lot more about it is important.

Case in point: the 2 polls of high school students (AZ, OK) that Nate reported on earlier without getting into details, showed a critical lack of knowledge in those students on basic civics, e g who was the first president of the US, what are the 2 houses of Congress, who wrote the declaration of Independence, etc.

Arizona's poll covered over 2,500 students. It was sponsored by the Goldwater Institute, an organization which certainly does have an ax to grind. The report was written by their VP of Research who is also a PhD.

The Oklahoma report was written by the same person, with numerous (and acknowledged as such) sections lifted from the AZ report.

The percentages of correct answers in both states were pretty parallel. Passing was 6 correct out of the 10 answers. In neither state did a single student answer all 10 questions correctly, not even 9 or 8. Only a handful of private school students in AZ answered 7 correctly. Not one student in Arizona's combined 3,500 plus got more than 7 right.

The author suggested cutting out state spending for civics education, substituting a mandatory test (given by a 3rd party) to qualify an applicant for a driver's license. Yep, promote the idea of teaching to the test, a notion he rejected earlier in the same report.

A snide description of an appraiser's credential MAI is "made as indicated." Given that the Goldwater Institute produced the report, is it not possible that Strategic Vision LLC created the data to fit the client's needs?

That's not an accusation - I'm just sayin'; September 30, 2009 3:30 AM
Joseph said...: @LAW:

From your whining one would think Nate operated his own news network, and that like Fox News, his megaphone large enough that he can create his own 'reality' regardless of truth or falsity. But he doesn't, and he can't. If Nate is wrong he'll have to eat crow on his blog. It will be his reputation that gets hurt, not Strategic Vision's. So there is zero need for your trollish concern and nagging.; September 30, 2009 4:49 AM
Opus 132 said...: beavis-

Zee was parodying other posters who were attacking Nate.; September 30, 2009 5:21 AM
Bradford said...: @LAW-

You are a good republican aren't you? You do not even understand the base case against this pollster. Go find a fact.

Why SV? Because the polling organization AAPOR sanctioned them, and only them, as they would not release their internals to their polls. They were the only member to do so and the only one to be sanctioned, thus they have essentially been accused of fraud by their fellow pollsters, even before Nate started to eat their lunch.

This is the problem with the Fox News approach, uninformed idiots like you think it is OK to go attack people without doing any research or have the knowledge of basic facts. Go research any right wing news story, and you will likely find an idiot like you who does know what he is talking about.

Leave, and leave Nate alone.; September 30, 2009 5:45 AM
Mr. Universe said...: That was an awesome baseball shout out to Nate on Rachel Maddow.; September 30, 2009 5:46 AM
loner said...: The Law 538 Commentary Package

Ah, but I was so much older then,
I'm younger than that now.; September 30, 2009 6:20 AM
Rick said...: Dear LAW,

By your argument, Nate should not offer accusations of misconduct until such time as there is solid proof of such. I then ask you to abide by your own argument and not accuse Nate of misconduct until you have solid proof that he is wrong in his assertations.

You clearly have doubt in Nate's approach and that is fine and in fact a good thing. We should always look to question the logic behind what we are told. However, you have yet to prove that Nate is actually wrong any more so than you feel Nate has proven Strategic Vision LLC to be in the wrong.

As for why this specific pollster? He's already gone after the methodology of a few in the past such as some of those done by Zogby, particularly the interactive ones. This pollser however has repeatedly refused requests from news outlets, other polling focused sites and also AAPOR. These requests were all in order to gain a better understanding of the results they are presenting. The latter in particular was looking into all of the pollsters involved in the 2008 democratic primaries in an effort to help improve the process globally. Lots of folks were very wrong and so they wanted to help to determine why. Yet SV constantly refused to participate in any fashion. This raised eyebrows all over the place, but of course few in the mainstream media would ever dare question the fact handed to them so it didn't generate much real buzz outside of the inner circle. Nate happens to be a new comer to that but is a number/fact happy individual and so began looking at the numbers that had been presented.

He did the digit analysis as we all know and compared it against his entire 2008 database and the results have been posted. In response to requests for a better control, Nate then selected a pollster who's states covered were not only similiar but overlapped heavily and who polled the same types of contests and questions. He did his best to select a control as close to SV as can be done in this type of environment and the results of Quinn matched that of his full db more closely than SV.

However, so long as SV refuses to release any additional details there simply is no way to verify they are authentic or proof them to be in error and that gets us to the root of the problem that really needs addressed and thus the requests/demands for more info and transparency.; September 30, 2009 6:21 AM
ecarlson said...: I think this post was out of line by Nate. I think Nate should take the high road. Let SV threaten lawsuits, while Nate quietly and consistently cites evidence that SV's numbers indicate fraud. Nate doesn't have to even accuse them of fraud - the numbers speak for themselves.

I think the evidence based on final digit analysis is convincing, though not proof. Some commentators have complained that Nate only made one comparison. I agree, but it's a big job to do many comparisons. Commentators have also claimed that unusual digit distributions might be commonplace in certain types of polling. Although this is true, there's no evidence that the polls SV performed should put a large bias in particular digits. Furthermore, the random results from sampling error should smear such biases, so that, for example, we can perhaps imagine a range of digits like 4-9 coming up more often, but it is hard to justify adjacent digits having wide disparities.

As for the analysis of the Oklahoma test, I think Nate did himself a disservice by trying to argue on a statistical basis. The main argument presented was that, while the data was consistent with a homogenous group of students with rather poor skills taking a group of questions of equal difficulty, it was not really plausible when you take into account the fact that students are heterogenous, so that a student doing well on one question makes it more likely the same student will do well on another question. This effect broadens the distribution. However, he ignored the correlation across students on the same question. If a question is hard, it may be missed by most or all students; if easy, they may all get it. This narrows the distribution. The two effects may cancel, resulting in a distribution akin to the one observed.

This leaves us with the argument that OK students can't all be that dumb. Though I agree with this statement, it isn't a statistical one. And a few commentators here have asserted that OK students ARE that dumb.; September 30, 2009 6:36 AM
widmerpool said...: SV is a fraudulent operation. I'm all for exposing them, especially since their polls are regurgitated by other media.

i do think Nate goes a little overboard with the personal stuff.; September 30, 2009 6:52 AM
Rick said...: Gonna pick a few statements to comment on here.

ecarlson said...
"Some commentators have complained that Nate only made one comparison. I agree, but it's a big job to do many comparisons."

Actually Nate intially ran a comparison against his entire 2008 database and found that it didn't match up in comparison to SV's entire range of released results that Nate analyzed. Some commented that this wasn't sufficient as a control and that it might have to do with the types of polls SV was running in comparison to the conglomerate in the DB. Then Nate picked the closest match in states and types of polls he could find, Quinipac (sp?), and redid the comparison as closely as he could, including keeping the total number of questions nearly equal. It may be that there aren't that many other pollsters as similiar and thus trying to force say Rasmussen or Zogby into the same comparative analysis would be less valid because of it.

"As for the analysis of the Oklahoma test, I think Nate did himself a disservice by trying to argue on a statistical basis. The main argument presented was that, while the data was consistent with a homogenous group of students with rather poor skills taking a group of questions of equal difficulty, it was not really plausible when you take into account the fact that students are heterogenous, so that a student doing well on one question makes it more likely the same student will do well on another question."

Actually, that was exactly Nate's point. If he took the percentage of people that got each question right, assumed all students were at that exact same knowledge level and ran his simulation, it produced results similiar to the overall number of questions right graph. Yet if he varied for students above and below that baseline the graph of total questions correct changed in outlay conciderably. Add in the other non-analytical issues such as the fact that no student got 8 or more out of 10 right, etc and it becomes a very convincing argument for further investigation.; September 30, 2009 7:09 AM
Bradford said...: ecarlson-

I was a college professor at a good liberal arts school, i left partially because OK students, even at a good school, are that dumb.; September 30, 2009 7:14 AM
mob said...: I find it odd that all the talk about spacing periods has come up, yet all but ignored was his clumsy writing.

The letter really shouldn't begin with Hi David nor should Nate have begun a sentence with was.

This isn't a beer conversation with his pal David or is it? A little more professional attitude would have been nice.

If Nate has a professional problem with the practices of this cmpany and wants clarification, being cute usually doesn't get on very far.; September 30, 2009 7:30 AM
Jeff said...: Bradford,
Your evidence of bias at Rasmussen leaves me a bit cold. The fact that they mention "congressional"? Come on. Congress is writing the bill - they are driving the train. Obama is off in Europe awaiting his marching orders. This is nothing like the bias in CBS/NYT/Newsweek polls. For instance, asking about the public option and always mentioning "competition", as if there was no dispute about that. Nate has exposed this as a very biased wording.

On the broader point, all of you slavishly adore Nate Silver, but somehow ignore his many defenses of Rasmussen polls. This despite the fact that their accuracy has been regularly demonstrated. They certainly pegged the steep decline Obama's popularity, the rise of the GOP on the congressional ballot, etc.; September 30, 2009 8:21 AM
Pragmatus said...: Jeff…

“Rasmussen…certainly pegged…the rise of the GOP on the Congressional ballot…”

Um—since there’s been no election yet to demonstrate the accuracy of this I think it’s safe to say Rasmussen hasn’t predicted anything here except the course of your wishful thinking.; September 30, 2009 8:41 AM
Dwight said...: Pragmatus, I assume you (or at least Dorothy Lauer) learned to type as I did, on something that looked like this?

However with the advent of variable width fonts with per-letter kerning, the need for double spacing after a period has largely disappeared.

Arguably, unless you are using a fixed width font.; September 30, 2009 9:27 AM
JF Isher said...: Ras just has larger rep. numbers and smaller dem. numbers.

PPP (democrat polling) has shown similar numbers - they claim (I'm not defending them, just saying what they claim) that people are more likely to give a true response if they aren't speaking to a human - that the same numbers that elected him were actually the numbers that disapproved of him generally all along, and that the polling really hasn't changed in the past 8 months, people were just trying to sound "fair" to human pollsters on the phone.

I think it's a fair argument. People keep acting like Obama not doing better than 51% is surprising, when 53% was the amt. that voted for him (and that was with excitement way up).; September 30, 2009 9:31 AM
Dwight said...: Incidentally that last line, given that the 'letter' is in a fixed-width font, is an indication that people should just drop it. Nate's just kicking it in an acceptable old-school mode.

P.S. If you want to get your pedantic on focus on the sentence beginning with "Was ...". :P Of course you are missing the point of casual speech tone but....; September 30, 2009 9:32 AM
Bradford said...: Jeff-

I worship noone and hand Nate his ass when he deserves it. He does not on SV.

As for Rasmussen - were you on the phone? Did you hear the tone used? Do you know they did not etio or describe ay public option? Do you read minds and thus can hammer something for which you have no knowledge?

The question order, tone of the questioner (taped in this case, as with all Ras polls), and push for Mike Castle in my state of Delaware were all distasteful and not good polling. But since you already know everything I guess you have a baseless opinion to offer on it.

Are you a teabagger?; September 30, 2009 9:42 AM
Dwight said...: Incidentally I'm kinda disturbed that the pedantics in the crowd missed an actual error. The subject line is "couple questions". There is in fact only one question in the email. He talks about a couple of things but only one of them is a question.; September 30, 2009 9:54 AM
Pragmatus said...: Dwight…

*sigh* Yep, that’s what I learned to type on. No fancy-schmancy electric typewriters in Miss Lauer’s class.

It is strange how completely typewriters have been routed from the modern office. For some things they are better than computers, such as addressing an envelope quickly.

If you ever want to completely puzzle someone under 25 years of age, ask them what a “carriage return” is…; September 30, 2009 10:04 AM
Wa - 7th said...: Even if Nate has a recorded confession of fraud in Al Gore's secret lock-box, this post is far below any standard of personal behavior publicly displayed thus far by Nate.

Nate, your letter adds nothing new to the mix other than sophomoric macho chest-thumping. Please give it a rest. Both you and your blog are better than this.; September 30, 2009 10:04 AM
markymark said...: Juvanya said
'While SV deserves to be fully investigated, I do not think this is your job, Nate. It gets you into a mess. No matter how correct you are, it will mark you as partisan and that is something to be avoided.'
------------------------------

I don't think Nate has ever come across as anything but partisan. Nate is clearly a liberal.

Personally I think the SV LLC investigation gives 538 some bite, that it has perhaps lacked so far in this off election year. I'd persoanlly like, though do not expect, SV to come on to the site and answer some of Nate's concerns.; September 30, 2009 10:12 AM
bitguru said...: Most of the people who advocate two spaces after a period were simply taught so and are passing what they learned along.

Many were taught two spaces in typing class but just as many were taught two spaces on the computer, presumably because that's what their teachers learned in typing class. If they learned it in school it must be true, right?

Every once in a while you find someone like Mike in Maryland that has given it some thought. Some of these have constructed elaborate arguments for what they irrationally believe, while others actually understand some typography and have more reasoned arguments. Sometimes it's hard to tell which is which.

Anyway, in my opinion two spaces is acceptable though not necessarily recommended in typewritten text (or monospaced faux-typewritten text, such as Nate's letter) but should be avoided in typeset or proportional-spaced text. But that's only my opinion.

I will quibble with Mike an Maryland when he says, "Some authors believe that text using double-spaces between sentences is more readable than text written with only one space after the period." Sure it's true that some authors believe that, but I don't think there's any actual evidence that it's true. (It's hard to say there is no evidence in situations like this because there could always be an old study hiding under a rock somewhere.); September 30, 2009 10:16 AM
Brother Wolf said...: I love the letter Nate!

and

@Law

"The general level of civility and objectivity here has been refreshing, and it just sucks to see it spoiled."

I agree - and you started it. So cut it out. Nate is doing a public service to a firm that may be trying to influence the body politics by lying through their teeth.

I bet he can find other historical oddities with this method in American political history.

I'm I supposed to use one space or two in front of my sentences?
Storytelling; September 30, 2009 10:19 AM
Erik said...: @Skip "Well, back in the dark ages of typing class, I learned that there are two spaces after a sentence-ending period to distinguish it from an abbreviation-ending period.

Did that change sometime since the Mac Classic came out?"
(and @pragmatus too)

Actually, the rule "changed" back to the previous rule. Typset text did not use two spaces after periods, because it looks ridiculous, wastes space, and interrupts the eye on the line. When typewriters were introduced, they had a fixed-width for all characters. Thus, a new rule was created, just for typewriters (and other fixed-character-width media), to put two spaces after periods. Nate used a fixed width font, so he correctly put two spaces after sentence-ending periods. The vast majority of writing today is proportional-width and should not have two spaces after the period. Most writers probably recognize that intuitively. Some writers (possibly the same who believe there is some grammatical rule for using which or that), believe that the two spaces after period is a grammatical rule. But this is not the case: two spaces after period is a typing rule.

(Also, Skip, we aren't using carbon paper any more either, so I think you can stop posting in triplicate if you want :)

Oh, and nice letter overall, Nate. Go kick some ass while we debate typography in the cheap seats; September 30, 2009 10:21 AM
Pragmatus said...: I agree with everybody who said that the “two spaces rule” was really only useful when not using proportionally spaced fonts, i.e. banging something out on the old Smith-Corona.

donse: Someone really stupid; September 30, 2009 10:22 AM
Juris said...: Hey I love this! A picture of the typewriter I learned to type on. Back when the only proper way to type was to leave two spaces after the period.

Then along came IBM's Selectrics with different font and symbol balls. Proportional spacing fonts along the Courier and Prestige Elite fixed space fonts.

The leading edge of the modern age.

And when IBM came out with their PC's they used fixed space fonts as standard. Then Steve Job got inspired by some lectures on calligraphy he'd heard at Reed College, and decided that Apple should start a new age (for computers). ,

Also once MS Word took over the market, it didn't matter if you put 1 or 2 spaces after the period. It reapportioned the characters on the line either way. Unless, of course, you insisted on using a fixed space font. Which is what Nate did. Thanks for the trip down memory lane, Nate.; September 30, 2009 10:23 AM
Pragmatus said...: Erik…

Wow, thanks. It never occurred to me that the two-space rule was put in place because of the invention of the typewriter, but you’re right. In typesetting it never applied.; September 30, 2009 10:24 AM
Rennie said...: "Opinions are like assholes, everybody's got one."

Unfortunately, a lot of people seem to think that everyone's opinion is equally valid. Except for their own opinion, which is much better.

In reality, the opinion of an intelligent and experienced person is usually much, much more valid than the average opinion.

Nate Silver is very, very smart, and he is very, very experienced in working with polls and statistics. (He pays me to say things like this. :-)

So when he implies that SV is probably cooking the books, I believe him. He has the smarts, the knowledge and the experience to be able to make an assesment like that better than 99% of the general population.

In addition, he probably has more information than he has divulged at this point. For example, on 9.25.2009 he requested that anyone who had ever been polled by SV contact him and tell him about it. This current blog article would seem to indicate that he has gotten zero response to that request. Maybe he has other items of information too that support his suspicions.

Flamboyant? OK, but don't we usually admire those who show off, as long as they don't overdo it?; September 30, 2009 10:27 AM
bitguru said...: I don't know if it was Rasmussen or not, but I was polled on the telephone last night. Most of the questions were about the Virginia governor race and local contests, but I was also asked if I was in favor of bills that "advocate the government takeover of the health care industry." Cute, eh?; September 30, 2009 10:28 AM
Andrey Stroilov said...: There is a difference between appearing partisan in that one prefers a certain outcome, and appearing partisan in that one cooks numbers to "prove" something that is not true. Nate preferred Obama to McCain, but there were points in the election when the numbers were not that good for Obama, and Nate called it out; similarly, right now (as far as I know) Nate probably supports public option legislation but his numbers and analysis suggest it's not particularly likely unless something changes.

Simply ignoring a conservative-leaning potentially-fraudulent poster would give the impression of the latter problem - that is, of being partisan to the point where Nate's numbers couldn't be trusted. By addressing them head-on, he's giving a clear reason why he doesn't use their numbers, and it's one that SV can address and remedy if they do have the facts on their side.; September 30, 2009 10:31 AM
Juris said...: Here's the commencement address by Steve Jobs in which he discusses fonts!

wv: beherall (the blogger's wish); September 30, 2009 10:32 AM
Pragmatus said...: Fifi…

Didja ever think it’s ’cause you’re such an immature whiner?

Surely this isn’t the first place you’ve been thrown out of as a pest.; September 30, 2009 10:38 AM
Juris said...: Which reminds me, is anybody here besides me old enough to remember the rightwing slogan "Better Dead than Red?"

Now that the radical right "redstaters" have suddenly fallen in love with that color, I am waiting for them to march behind a new banner, "Better Red than Dead". No consistent fools are they!; September 30, 2009 10:51 AM
Michael (mbw) said...: @LAW- Perhaps it would help to go through the statistical anomalies systematically. Nate somewhat flubbed the description of the trailing digit anomaly by getting caught up in irrelevant aspects of the underlying distributions, rather than focusing on the robust statistical predictions.

1. No plausible prior distribution of 'true' percentages has any major sharp fine-structure. Any structure that might somehow be there is, at any rate, filtered by the Gaussian spread of the sample around the true vale. To put this in terms of the direct digit counts, the differences between adjacent digit counts should show a distribution very close to that predicted for sampling from a uniform distribution, even if the actual distribution is a bit non-uniform. To put it in terms of the Fourier representation, the Fourier component with period 10 may show some unknown effects of the distribution, but those with periods 2.5, 3.33, and 5 should be just statistical. (Period 2 could show a rounding artifact, though it appears not to.)
The data from Quinnipiac and from SUSA look completely unremarkable by these tests. The data from SV look extremely unlikely.

2.The OK data look bizarre in detail, given reasonable expectations for students based on their performance in school exams. However, it's hard to be quantitative about how much the student response would fall ff on a non-graded survey.

3. The OK data show an extreme internal anomaly. The distribution of student scores is close to what would be obtained from the single-item scores if the performances on these different items were almost completely uncorrelated, even though the items were closely related. This flies in the face of every known academic test result on anything, for which even unrelated items tend to be positively correlated. It is exactly the result one would get, however, from the most naive Monte-Carlo program designed to generate net scores from item scores.

Now it could be an accident that these statistical flukes all just happen to occur to the one pollster who refuses to divulge normal information about their data-gathering. Any standard Bayesian calculation of those odds would not favor such an explanation.; September 30, 2009 11:03 AM
Carey said...: Since there is so much interest in proportional spacing in fonts today, and how much white space should follow a period, I thought I'd mention my experience with a mid-century executive secretary's typewriter.

This was a high-end mechanical typewriter that had a "half-space" and a "one-and-a-half" space key (actually I think one of these was a shift of the other). Normal practice was to use the "one-and-a-half" key after periods.; September 30, 2009 11:09 AM
svpharper said...: Clearly LAW has not read much of Nate's criticism of SV. He is not picking on them alone. He has "picked" on other pollsters who demonstrated bad methodology or none at all in this case. SV is the ONLY large pollster who does not publish methods and cross tabs. And the results posted fall into possible fraud category. Nate is the best in the business and doesn't need to be criticized by some idiot poster who knows nothing about polling or statistics.; September 30, 2009 11:17 AM
Pastor Alan said...: Microsoft Word won't flag either one or two spaces after a sentence, if that means anything.

Like others here, I learned that double-spacing was to be used only for a non-proportional font. Since I assume most typographers are already made physically ill by the very ugly non-proportional fonts in use, I doubt that it would make much difference. I switched to single-spacing after a sentence sometime in the late 90s, after years of using precious USENET bandwidth using double-spaced sentence breaks.; September 30, 2009 11:20 AM
JamesY said...: @LAW
"By the logic some of you people are using, the president must have been born in Kenya because dozens of blogs have asked for the "real" birth certificate and have gotten no response from the White House"

umm, Hawaii has, i believe three times now (maybe it's just been two) shown Barack Obama's birth certificate. Has that quieted anyone down? Nope. Just because there is proof to the contrary show people who don't like looking at proof that their case is baseless. There are people, in our government no less, that believe the earth is flat, because the bible says so.
and no amounts of proof can tell them otherwise.; September 30, 2009 11:22 AM
JamesY said...: @Skip
currently, if you look at writing standards, they have changed the space after period to 1 space. I know, it irked me the first time someone told me, i was like "but, i've ALWAYS been taught this way."
But, indeed, after looking up the new guidelines myself, i now trust someone who's job it is to publish content on a multitude of websites daily...; September 30, 2009 11:27 AM
shiloh said...: Stating the obvious: there are many, many whiny, trolls in varying degrees, conservative winger Reps at 538 who really, really, really wish Nate was a conservative Rep.

Alas ;) not to be. Know it, learn it, believe it, live it!

And I hesitated using the word Republican since most of the conservative wingers at 538 refuse to admit they are or ever were Reps lol

carry on; September 30, 2009 11:31 AM
shma said...: @ mbw
I've been away from this site for a long time so it's good to see solid posters like you still here. I wanted to comment on your analysis

"No plausible prior distribution of 'true' percentages has any major sharp fine-structure. Any structure that might somehow be there is, at any rate, filtered by the Gaussian spread of the sample around the true value."

I'm not sure that's necessarily true. Since pollsters are generally asking questions about subjects that are not decided issues, I would expect a peak around 50 with no 'don't know' option and a peak in the 40s with a 'don't know' option for two option polls. For general polls, I don't think you can say that a predicted uniform distribution is correct unless you have actually verified it by looking at the trailing digits for polls from all major pollsters. I think a control like LAW suggests would be a good idea. And it wouldn't take too much time to do.

BTW, I haven't seen the trailing digit analysis for SUSA. Do you have a link for it?

The OK data:

I agree it looks extremely weird. But in the end, it's one poll out of probably tens of thousands. And you could conceivably find one such weird poll in the databases of each pollster, so it doesn't surprise me that "statistical flukes all just happen to occur to this one pollster". It was found because people were looking for something like it.

The OK data is circumstantial evidence, maybe even strong circumstantial evidence, but you're not going to get anything conclusive from one poll.; September 30, 2009 11:42 AM
loner said...: The My Back Pages earlier applies to me.

I was never much on pointing out spelling, punctuation and grammatical errors in online posts (too many of my own for one thing,) but I was into warning other posters/commenters that they were putting their reputations for this and that at risk by writing this or that. What a goof.

My parents were schooled in journalism and worked in public relations in Hollywood before starting a family. Skeptics always.

I don't miss typewriters. (That sounds like fun, Carey.) I do miss editors. Two spaces after periods until cutting and pasting stopped being painstaking work, but I still mostly leave two and as little a paper trail as possible.; September 30, 2009 11:45 AM
LAW said...: @Brother Wolf

What, exactly, have I done that is "uncivil"? Disagreement is not uncivil. I have been called all kinds of nasty names and have chosen not to retaliate.

@Michael (mbw)

You can claim whatever you want about the shape the distribution "should" take, but that does not make it so. There is no precedent for this kind of analysis on this kind of data, and as such Nate really needs to prove something about the shape of a "typical" distribution. My instincts tell me that the cyclical nature of the sampling plus some natural biases that occur among poll numbers may create a distribution that is quite non-standard.

As for the OK poll, I will readily agree that it is odd. But one odd poll doesn't not make widespread fraud.; September 30, 2009 11:49 AM
Mark Grebner said...: @shma - "I'm not sure that's necessarily true. Since pollsters are generally asking questions about subjects that are not decided issues, I would expect a peak around 50 with no 'don't know' option and a peak in the 40s with a 'don't know' option for two option polls."

You've missed his distinction - any difference as large as the ten point gap you refer to can be "resolved" by polling and would be called by analogy "gross stucture'. MBW is using the term "fine structure" to refer to structures too small to be resolved. In particular, distinctions on the order of 1% or 2% in size.

BTW, yesterday one of SV-LLC's loudest defenders announced that he had a data set which would show the same contrasts in Rasmussen's polls. He promised to post his data later that night - and has now disappeared. If anybody finds either him or his data, I'd appreciate a link - I can't imagine what he claimed was true.; September 30, 2009 12:02 PM
Valpey said...: -Trailing Digit Distribution Revisited-

Trying my best to give Strategic Vision the benefit of the doubt here:

http://valpeyblog.blogspot.com/

I still have to go through some pretty absurd gymnastics to come up with a trailing digit distribution which could have produced Strategic Vision's distribution by chance alone.; September 30, 2009 12:07 PM
this.is.certainty said...: To all the moralistic nitwits ITT:

Nate is simply trying to have a reality-based conversation on SV methodology. If that offends you, too bad.; September 30, 2009 12:09 PM
Mark Grebner said...: @valpey: I see the charts, but the point isn't to draw a similar picture, but to describe a process. What did you do to generate your charts? Was it a plausible path by which SV-LLC might have generated their results?; September 30, 2009 12:27 PM
shma said...: @Mark Grebner:

"You've missed his distinction - "

It seems I did. Thanks for pointing that out.

Does anyone here know which commenter did the SUSA analysis (or which post the comment appeared in)?; September 30, 2009 12:28 PM
Freedem said...: @Juris -
I have also been amused at the big color switch, thinking of movies like "Red Dawn"(like Dec. 12, 2000) and such though others have made similar comments.

What I have seen little of however is any (but my own)commentary about why the Republicans have the Elephant for a totem in the first place. Wasn't it about something like never forgetting or forgiving the very kind of folks who now make up almost all their ranks?; September 30, 2009 12:37 PM
George Taylor said...: @Wayward

Indeed.

Concern trolls waste time wringing their hands and furrowing their brows and telling us how Nate is overstepping his bounds by being snarky in his letter to SV, or debating One Space vs. Two Spaces after periods. Republican trolls measure their success by how often they hijack substantive threads and turn them into discussions of whether or not the bible says that the world is flat, making high schoolers in Oklahoma and Arizona seem brilliant in comparison.; September 30, 2009 12:39 PM
shrinkers said...: One thing that would be suspicious about Strategic Vision - we should wait to see if it happens -

Nate has pointed out serious problems with the final digits of the polls, and with the fact that they always add up to 100.0 % with no rounding errors. These are reasonable issues that reasonably imply the numbers might be invented.

Having been put on notice, it would be interesting if suddenly these anomalies disappear in future polls. (OMG, I hadn't thought of those things! I'd better be more careful when I make up my numbers next time!); September 30, 2009 12:49 PM
Pragmatus said...: George Taylor…

Well, as a participant in the font discussion I can assure you it was not “concern trolling” that prompted me to engage in it.

People with a certain amount of intellectual curiosity frequently get waylaid by odd things in the midst of discussions, and then form side discussions about the oddities, without any purpose of killing or derailing the primary discussion.; September 30, 2009 12:54 PM
lindi said...: Nicely done Nate.

Despite the ridiculous posturings of the many amateur rationalists in this thread, your reputation is impeccable and your word is quite credible (hence the threat of a lawsuit).

Thanks again for your great work.; September 30, 2009 12:55 PM
bleepul said...: Boy, the dynamic never changes here.

So can someone answer the following question: why does strategic vision matter in the first place? Is this about silencing opposition or garnering web-hits or is it something of an ethical obligation? If it's the latter the open letter seems like a bad idea. Personally I think the whole attack was a bad idea ... not sure what is to be gained except an increase in traffic, but maybe that's enough.; September 30, 2009 12:59 PM
loner said...: shma—

The last post from "steve" is here. An additional comment in the same thread is here. He was getting his bearings in this comment thread. (Note: You have to click on Post a Comment to see the entire thread.); September 30, 2009 1:04 PM
shrinkers said...: @bleepul
So can someone answer the following question: why does strategic vision matter in the first place? Is this about silencing opposition or garnering web-hits or is it something of an ethical obligation?

Bleep, Pan answered your question fully in his 12:26 AM post above.; September 30, 2009 1:05 PM
Michael (mbw) said...: @shma-

The sort of structure which you discuss as plausible is low-frequency structure (on the scale of 10 digits). I agree that there's no obvious reason to use uniform priors, or 'Benford' or any other particular form on that scale. However, the sort of effects you mention would have very small effects on adjacent digits, or on Fourier components with period shorter than 10. These higher Fourier components are also strongly filtered by the Gaussian sampling error. It's in just these rapid variations that SV has anomalies, in contrast to Q and SUSA. (Another poster gave SUSA data in I think the 2nd SV thread.)

@LAW- I don't know what this is supposed to mean: "the cyclical nature of the sampling plus some natural biases that occur among poll numbers may create a distribution that is quite non-standard." It sounds like gibberish, but perhaps you could explain it. At any rate, whatever it is you're talking about makes no high-Fourier anomalies in either SUSA or Q, who also happen to be among the 95% of pollsters who don't stonewall about their methods.; September 30, 2009 1:10 PM
George Taylor said...: @bleepul

SV matters because it is a pollster that portrays itself as legitimate and which has conducted polls that seem to give credence to a specific ideological viewpoint--in the matter at hand, the opinion that public schools in Arizona and Oklahoma are failing to educate high schoolers adequately, at least with regard to civics and US history.

If it turns out that SV was not adhering to established survey protocols, but was instead manufacturing data to meet an ideologically pre-determined result, it has committed fraud. Any legitimate pollster, or any person who understands the need for legitimate polling, should understand the value in exposing any group that claims to have scientifically surveyed the public but which available evidence suggests may in reality have fabricated the numbers.

Not to address possible inconsistencies with how SV gathers and processes data is to risk reducing the level of trust the public has in all pollsters as a group.; September 30, 2009 1:14 PM
Michael (mbw) said...: @Mark Grebner-

Whoops, missed your prior response to shma. Thanks.; September 30, 2009 1:19 PM
Pragmatus said...: George Taylor…

Or, I would add, that the public just accepts polling data which are in fact lies and forms their opinions accordingly. You can’t stop politicians or factions from lying, but you can call them out on it, and everyone is better off when this is done.; September 30, 2009 1:19 PM
MarkinIL said...: I just completed running some analyses using the Strategic Vision LLC, Presidental and Senate polling, Survey USA, and Quinnipiac data that has been posted by Nate and other commenters.

I assumed that the digits should be random and used the Chi-Square test to see if this was true. Because traditional p-values are heavily biased by sample size I also computed phi, a measure of effect size. If there is a better measure of Chi-Square effect size please let me know.

Here are the analyses

Strategic Vision
Chi Square = 99.593
P-value < .000000001
Phi = .134

Quinnipiac
Chi Square = 17.678
p-value = .039
Phi = .057

2008 Survey USA
chi square = 3.406
p-value = .946
Phi = .030

Senate and Presidential Polls, 2008
chi square = 22.828
p-value = .007
Phi = .060

As you can see many of the data sets produce statistically significant chi-square tests, which suggest some degree of non-randomness (as indicated by the p-value).

However, when one looks at the effect sizes it is possible to suggest that Strategic Vision LLC is significantly more non-random than the other three examples which all produce similar effect sizes.

If others have data, such as MidPointMan, I would be happy to add to the analysis. No reason pollsters should be left out. I just don't have the time to collect this stuff myself.; September 30, 2009 1:24 PM
shma said...: @loner,

Thanks!

Those SUSA numbers look more uniform than even Quinnipac's numbers.

@shrinkers

...the fact that they always add up to 100.0 % with no rounding errors. These are reasonable issues that reasonably imply the numbers might be invented.

The rounding issue doesn't really imply anything except that they are not using the 0.5 rule for rounding. There are other rounding methods that preserve the total sum (in this case, 100%).

For instance, a commonly used method is the following:

1)Initially round all numbers down. Add up all the leftover fractions to get an integer number of leftover percentage points.
2) Apportion each of the N extra percentage points to the N numbers which had the largest fractions initially.

Ex. yes=33.4, no=40.3, maybe=36.3

We have .4+.3+.3 = 1 percentage point after rounding down, which goes to yes because it has the highest fraction initially. After rounding:

yes=34, no=40, maybe=36

Note that the total using the 0.5 rounding rule would give you 99.

Ex.2 yes=34.7, no=25.7, maybe=39.6

We have 2 extra percentage points after rounding down which go to 'yes' and no'. After rounding:

yes=35, no=26, maybe=39

Note that the total using the 0.5 rounding rule would give you 101.

This rounding method can be used, for instance, when you are trying to fill a parliament with a fixed number of seats using proportional representation. If a party wins 45.236 seats, you can't give them .236 of a seat. So you use this rounding method to guarantee that, when you are done rounding, every seat is filled.; September 30, 2009 1:30 PM
Michael (mbw) said...: Here's the old SUSA data post from "steve":
"OK, here are my trailing-digit figures for all 2008 Survey USA polls that fit the criteria of Nate's analysis.

0 - 401
1 - 407
2 - 384
3 - 375
4 - 383
5 - 402
6 - 396
7 - 370
8 - 392
9 - 385

As mentioned above, I have excluded 3rd party candidates, primary polls with more than two named candidates, and issue polls with more than two options."

@MarkinIL: Try the same analysis but first removing any period-10 component from the (mod 10) digit distribution. What you'll find is that chi^2 goes way down except for SV. That's because the distributions are in fact a bit non-uniform, but have no non-random fine structure, at least after filtering by sampling eror, except for SV.; September 30, 2009 1:34 PM
yoink said...: @LAW

You keep saying that Nate has leveled "accusations" against SV. I believe I've read all of the SV posts on this blog and I'm yet to see a single "accusation." What Nate has done is identify a range of troubling anomalies in their polls. He has pointed out what those anomalies are, explained why they are troubling, and given SV the opportunity to explain them. He has been very careful--always--to say that nothing he has found amounts to "proof" of wrongdoing.

The ball is really in SV's court now. If they can readily explain these anomalies and care, at all, about their reputation, I'm at a loss to explain why they don't do so (and no, there is no analogy to the whacko birther claims here; President Obama's birth certificate is available online for all to see and the State of Hawaii has confirmed it's authenticity. In other words he did address the questions when they were first raised).

In the case of this letter, all Nate has done is to give SV a very public chance to clear its name, benefit a charity and prove that Nate's concerns were misplaced. Again, I cannot understand why SV would not seize this opportunity if, in fact, its polling is legitimate. Obviously they are under no obligation to respond to Nate's open letter, but equally obviously we are free to draw whatever inferences we like from their unwillingness to do so.; September 30, 2009 1:38 PM
bleepul said...: GT, shrinkers,

With respect to Pan's post, if the objective is to avoid criticism of those on the right who scream "bias" I'm not sure this strategy is going to pay dividends. In my opinion it merely opens the door to the charge that Nate is acting out of political motivation himself. He won't get buy in from the right in any event and becomes less credible to those in the middle (like myself).

If it were me I'd simply point to the fact they never revealed their methodology and let it go. This crusade makes me wonder what the motivation truly is.; September 30, 2009 1:39 PM
Pragmatus said...: MarkinL…

I think we can call SV out empirically too—there is simply no chance that 400+ of the students they sampled listed the two American political parties as “Communist” and “Republican”, all 400 having come up with the answer independently.; September 30, 2009 1:41 PM
mclever said...: shma:

The SUSA analysis was done by "steve", and it was done in steps.

Steve's first efforts were posted here:

http://www.fivethirtyeight.com/2009/09/strategic-vision-polls-exhibit-unusual.html

(Note, be sure to go all the way to comment #270 for interesting supplemental statistical analysis by "Mark".)

Steve's next updates was posted here:

http://www.fivethirtyeight.com/2009/09/comparison-study-unusual-patterns-in.html

Preliminary numbers at September 26, 2009 5:57 PM, and final numbers September 26, 2009 7:05 PM.

P.S. Sorry for not doing the fancy hyperlinks...; September 30, 2009 1:45 PM
loner said...: bleepul—

In my opinion it merely opens the door to the charge that Nate is acting out of political motivation himself.

In my opinion your comments here open the door to the charge that you're not one of those "in the middle." So it goes.

Again, this is about pressing the transperancy issue until Strategic Vision, instead of making excuses and stalling in the hopes that the questioners will move on to something else, adequately explains their polling methodology.; September 30, 2009 1:53 PM
Michael said...: Oh look, Mule is lying by deliberate omission. Is anyone surprised? I'm not.

Let's present the FULL quote, shall we?

"But this data is not random. It's not close to random. It's not close to close. Which brings up the other possibility: Strategic Vision is cooking the books. And whoever is doing so is doing a pretty sloppy job."

Nate is quite clear in presenting this as a possibility, not as the absolute statement Mule lies to make it seem to be.; September 30, 2009 1:59 PM
Hunter L. Cook said...: @Freedem:

Credit for the GOP elephant goes to Thomas Nast in the 1870's, and no, it doesn't have anything to do with the elephant's memory or grudge-holding.; September 30, 2009 2:00 PM
bleepul said...: loner ...

I'm in the middle. Pro-choice, pro-environment, pro-LGTB, anti-death penalty, non-religious, anti-big government, anti-entitlement, pro-free market. Don't know what to tell you.

As for strategic vision ... who ... really ... gives ... a ... shit. If they are cooking the books they'll eventually go down without Nate dragging himself through the mud playing police-man.; September 30, 2009 2:04 PM
yoink said...: Michael writes:

Nate is quite clear in presenting this as a possibility, not as the absolute statement Mule lies to make it seem to be.

Well now, be fair. I mean, it is possible that Mule Rider is simply too stupid to understand what the word "possibility" means.; September 30, 2009 2:04 PM
Pragmatus said...: I have a feeling that our brand new friend “Mule” is none other than “Mule Rider”. Maybe we should get SV to do a poll on the possibility.; September 30, 2009 2:04 PM
yoink said...: Bleepul writes:

If they are cooking the books they'll eventually go down without Nate dragging himself through the mud playing police-man.

How, exactly, is a polling firm that is cooking the books supposed to "go down" if nobody every calls them out on it for fear of "dragging themselves through the mud"? Are we supposed to wait for their uneasy consciences to get the better of them?; September 30, 2009 2:07 PM
loner said...: bleepul—

If they are cooking the books they'll eventually go down

Why?; September 30, 2009 2:07 PM
Pragmatus said...: bleepul…

You keep claiming you are “middle of the road” etc, ad nauseam but whenever a cherished right-wing icon comes under attack you get all flustered.

If SV is lying and thus manipulating public opinion, why do you dismiss their crookedness so easily?

Sorry, but you’re about as middle-of-the-road as Joe McCarthy was.; September 30, 2009 2:08 PM
bleepul said...: yoink,

I thought they were sanctioned or something by the governing body that oversees licensed pollsters. I assume that is their job and not Nate's ... but perhaps no such body exists and they self-police. However, if that is the case you might as well go after every foxnews focus group which seems like a waste of time.; September 30, 2009 2:11 PM
bleepul said...: My cherished right wing icon? What exactly are you talking about?; September 30, 2009 2:12 PM
shrinkers said...: @bleepul
With respect to Pan's post, if the objective is to avoid criticism of those on the right who scream "bias" I'm not sure this strategy is going to pay dividends.

Not quite, Bleep. The Whignuts are going to accuse him of bias no matter what.

The question is more - should we exclude SV from the polling data? If we don't, we may be creating a biased result. If we do, we may be creating a result that will be seen to be biased, although it isn't.

If SV is being dishonest, it simply makes sense to call them out. There is no reason - and no excuse - not to.

If they are not being dishonest, it's easy enough to prove that, and I'm certain Nate will be most gracious and open in acknowledging his error.; September 30, 2009 2:14 PM
Judicial Politics said...: @ LAW

It's almost as if you're not reading Nate's actual posts

"dI posed that question largely as a hypothetical yesterday. But today, I pose it much more literally. Certain statistical properties of the results reported by Strategic Vision, LLC suggest, perhaps strongly, the possibility of fraud, although they certainly do not prove it and further investigation will be required."; September 30, 2009 2:19 PM
bleepul said...: Shrinkers,

In any other profession you would make a complaint and let an organizational structure investigate and judge. Now maybe that type of oversight doesn't exist in polling, I really don't know how it works. And maybe it's likely that SV is cooking the books ... but Nate is taking a big risk and undermining his "profession" by taking the vigilante position. I'm actually on Nate's side here believe it or not.; September 30, 2009 2:19 PM
Hunter L. Cook said...: @LAW

Your objections here are becoming ever-more shrill even as they ignore many thoughtful responses. Let's start with the facts:

1)SV is the only pollster that refuses to provide any supporting data.
2)SV claims to be "Atlanta-based" but apparently has only one "real" (i.e. "not a UPS store") office in a small town in another part of GA. None of their published addresses appear to be suitable as call-centers.
3)SV found that only 23% of high-schoolers in OK could identify our first president. This requires no statistics to look funny.
4)The trailing-digits are dramatically (5.7 SD!) non-random, and much more non-random than a)the whole set of 2008 polls taken together or b)the closest available control group (Quinnipiac).
5)SV immediately threatened to sue Nate as soon as he made a statistical analysis of their data. They made no attempt to answer his concerns...just the legal threats to scare him off.

Now, looking at all that together, it truly boggles the mind that you say you don't see anything suspicious. You can say you don't see conclusive proof (and we should note that Nate makes this very clear) but when you say "I have seen nothing that I would consider "suspicious" enough to merit the witch hunt that Nate has decided to embark on" we have to wonder what would make you suspicious. As others have pointed out, requiring conclusive proof before starting to investigate suspicion doesn't make a lot of sense.

And that really leads into the ridiculous part of this: you compare Nate to the "birthers" for asking for evidence to disprove the fraud. But it's not the asking for evidence that defines birtherism...it's the way they completely ignore any and all evidence that has been provided. The birth certificate has been shown; public officials in HI have said on-record that they're valid. Courts have thrown out the cases. The evidence is strong that Obama is a natural born citizen, but this evidence is ignored by the birthers.

By contrast, Nate asked for evidence where there was none. And here's the fun part: in these comments, you have consistently, birtherishly demanded a control group. And a few times people mentioned that Nate started by providing a control group (all 2008 polls taken together) and then specifically addressed concerns that said control wasn't good enough by providing results for the best control group available, the Quinnipiac polls. Since then other researchers have taken other pollsters' (notably SUSA) data and done the same thing. And it always comes out the same: SV's trailing digits are way, way, way less random than anyone else's that anyone has tested thus far.

And even though that's been mentioned several times now, you keep ignoring the evidence and shouting about how Nate's method is invalid "in the absence of a control group." It's not in the absence of a control group. The control group is there. You're ignoring the evidence you don't like. In short: You're the birther.; September 30, 2009 2:24 PM
Mark Grebner said...: @Michael (mbw) said... Whoops, missed your prior response to shma. Thanks.

But your response included reference to Fourier analysis, while I am merely a humble blacksmith of a statistician. I liked your answer better.

But could you please explain what you mean by "first removing any period-10 component from the (mod 10) digit distribution"? Unlike half our fellow posters, I don't suspect you of deliberately speaking gibberish, or not knowing what you're talking about. In your case, you DO know - but I don't. And I bet if I don't, neither to many others. Could you take another run at your point?; September 30, 2009 2:25 PM
shrinkers said...: @bleepul said...
but Nate is taking a big risk and undermining his "profession" by taking the vigilante position

No, actually, he is increasing his reputation as a careful and thorough statistician, who has a legitimate concern about the integrity of the data he's using. He is showing himself to be careful about ethics. He has shown that he isn't wiling to be manipulated - or to allow us to be manipulated - by dishonest con artists.

He's in a position to notice discrepancies, and to point them out, and possibly to reveal an unethical pollster. Being in such a position, I'd say he has a responsibility to do this.; September 30, 2009 2:31 PM
yoink said...: Mule writes:

I got a hearty chuckle out of the double-standard you guys apply to me...I knew you'd fall for the bait and hope like hell you're so eager to tear apart a quote taken out of context the next time a conservative makes it.

Hey, maybe SV should use that defense! "We were deliberately cooking the books because we wanted to teach people to be vigilant and maintain a healthy skepticism about polling data!"

I mean, we'd believe them exactly as much as we believe Mule in this instance, wouldn't we?; September 30, 2009 2:32 PM
loner said...: bleepul—

Nate is taking a big risk and undermining his "profession" by taking the vigilante position.

I'm assuming Nate knows exactly what he's doing because there's an available history.; September 30, 2009 2:37 PM
bleepul said...: shrinkers ...

He may have increased his reputation in your eyes. If he was ethical he would have pursued a just course of action rather than calling out a group publicly, which make me think this is really about attention and a test of power. Understandable that he now works analyzing politics rather than baseball but nonetheless wrong.

What you fail to appreciate is that he has set himself up for the same type of vigilante-style attack going forward. I don't want MD's bad mouthing other MD's becuase of their surgical history. I want a review board that monitors and decides fairly. That's ethical.; September 30, 2009 2:38 PM
Zetal said...: Very well put Hunter L. Cook. Way to refute the troll.; September 30, 2009 2:39 PM
shrinkers said...: The Whignuts are going to attack Nate no matter what. It's fun to see them attacking him for being ethical and responsible. Kind of makes them look even sillier.; September 30, 2009 2:42 PM
bleepul said...: loner ...

My case in point. Why does Nate feel the need to wade into the middle and make it personal? Very unprofessional.; September 30, 2009 2:42 PM
yoink said...: Mule writes:

I simply pointed out the double-standard. I'm not surprised you are too thick to understand that.

Actually, Mule, if you consult the record you'll find that what you did was deliberately quote a partial sentence of Nate's out of context so as to imply he'd said something he clearly did not say. When you were caught red handed deliberately distorting Nate's argument you then claimed that you had done so deliberately in order to provide you with the occasion for pointing out some supposed "double standard."

The one thing you have not, in fact, done is "point out" a double standard. You have alleged a double standard, but in order to "point" one "out" you would need to actually provide evidence of some compelling pattern of commentators on this board who are opposed to your views deliberately quoting others out of context. Not having done that we simply have a case of someone being caught out in a lie and then compounding that lie by making up a very silly story to account for it.

Or, in other words, Mule Rider's usual practice for any day with a "y" in it.; September 30, 2009 2:43 PM
loner said...: bleepul—

Did you read the link? The last sentence:

Last Friday, I attended a conference on survey quality at Harvard University, where UNC professor Phil Meyer said that our best hope is a "real accountability system" based on public pressure, "a more efficient market on the demand side." He is absolutely right.

Nate is applying public pressure. More power to him.; September 30, 2009 2:47 PM
Valpey said...: I've made my argument a bit more clear as to how we might better develop a reasonable expectation of trailing digit distribution:

http://valpeyblog.blogspot.com/

I am happy to discuss methodology. Let's see if we can't somehow exonerate Strategic Vision.

Without some kind of absurd rounding practices, the best liklihood I can get that Strategic Vision's numbers to be a result of random variation is 0.011%.; September 30, 2009 2:51 PM
shma said...: @Everyone who pointed me to Steve's SUSA results: Thank you all.

@mbw:That is an interesting point, and if you're right, that's strong evidence against SV.

Although I should point out that a leading digit Benford's Law distribution does have a sharp drop off between 1 and 2. Obviously, that's not what we're looking at here, but that kind of drop off does happen for some plausible distributions in certain cases.; September 30, 2009 2:51 PM
shrinkers said...: bleepul
He may have increased his reputation in your eyes. If he was ethical he would have pursued a just course of action rather than calling out a group publicly,

Actually, since issues have already been pointed out with SV in a less public context - and since they have not responded to the concerns - "calling them out publicly" is the "just course". Duh.

They've had their chance in private. They didn't take it.

IF they are cooking the books, then this fact deserves the widest possible exposure.

If they're not, SV can make it all go away very easily.; September 30, 2009 2:52 PM
Mr. Universe said...: Seriously? You guys are eating up a discussion thread in a political forum debating archaic typing styles? C'mon.

Besides, I think the joke that Nate was trying to suggest with that font is that he was dumbing down the technology for SV LLC's benefit. The implication being that they are using old technology in running their business. A subtle slap that Mr. Johnson might not pick up on. HE PROBABLY SHOULD'VE SENT THEM A TELEGRAM [STOP] BUT I DON'T THINK WESTERN UNION DOES THAT ANYMORE [STOP]

[END MSG]; September 30, 2009 2:54 PM
shrinkers said...: @bleepul said...
Why does Nate feel the need to wade into the middle and make it personal? Very unprofessional.

If someone threatened to sue me, I'd probably take it personally, too.

Until that time, it was entirely professional. Nate revealed some legitimate and professional concerns, and invited response. The response he got was a threat.

Suspicious, I'd say.; September 30, 2009 2:56 PM
bleepul said...: Didn't realize I had to go through it with a comb :).

Yeah, I don't know. I think what the Professor is talking about is people like me, as consumers, taking a deeper interest in more polls asking deeper questions and for the more background information ourselves. Nate has a stake in the industry which makes his stance dubious to say the least. It's unethical for him to jump in and try to sway opinion. I'm a scientist and there are reasons why refereed journal article reviews are done by multiple people and anonymous ... namely so that no accusation of bias or benefit can be leveled. What is AAPOR even for then if it is Nate's job to uncover false pollsters?; September 30, 2009 2:57 PM
Samuel said...: I agree with LAW in that this most recent exchange of volleys is unprofessional and Sophomoric. Well, it's funny so that is okay, possibly this is a generation gap? Welcome to the 21st century.

Yes, the evidence that Nate has presented is merely pretty damning and wouldn't win a criminal case. But public opinion is not like criminal court - British-style libel, in which you need absolute proof before you can destroy someone's reputation, is anathema to free speech and to the maintenance of healthy public dicourse.

The public has such an interest in trustworthy news and political reputation that in either arena, the mere likelihood of guilt can and *should* be damning.

There's a further line of objection that reduces to:
1) Who is Nate Silver to demand an explanation of these apparent anomalies?
2) Nate Silver's reputation is such that his scorn will destroy them.

Well, 2 answers 1. You're right that wingnut bloggers will issue similar challenges - and if you want the respect of those who care about their opinion, you must answer those challenges.

In any academic endeavor, and in any business that relies entirely upon trust, you are responsible to answer the criticisms of those who your colleagues or customers trust. So, to recap:
in policy, politics or the professions, you must avoid even the *appearance* of impropriety
therefore
Nate is right to call them out

the SV,LLC people will lose all professional credibility if Nate Silver thinks they're a fraud
therefore
they must answer to Nate Silver; September 30, 2009 3:01 PM
Delorian said...: Were there posts that were posted then removed between 12:09 and 12:27 PM? I ask because I made a post, I think during this time, which is no longer shown. Also, later 'shrinkers' references a 'pan' post from 12:26 that is not shown. It is not a huge deal to me, the posts removed I think mostly revolved around the 'does the Bible say the earth is flat' discussion, I just wondered why a group of posts (mine was not offensive) would be removed like that.; September 30, 2009 3:01 PM
loner said...: bleepul—

The point is that currently it's everyones job and Nate, being part of that group, has taken it on. Again, more power to him.; September 30, 2009 3:03 PM
Rob322 said...: bleepul:

It's not unethical for Nate to ask them to explain themselves simply because he has a stake in the issue. Unless of course you can show how he will personally benefit from beating SV. He, as well as professional pollsters, have a great stake in this. Bad pollsters cause damage to the whole organization of polling and the public often will apply a broad brush to all groups that do polling or rely on them. It's easy to say the public should "take a deeper interest" but I think we know that's not likely to happen; you can barely get 50% to actually show up and vote half the time so getting them to fascinate themselves with the statistical methodologies and strategies that pollsters use (and getting them to understand them) is a ridiculous hope at best.

Quite simply put, if people like Nate do not call attention to these things then no one else will.; September 30, 2009 3:05 PM
MrHill said...: Hey Mule, are you from Jonesboro Arkansas? How is weather down there?; September 30, 2009 3:07 PM
shrinkers said...: @Delorian
Also, later 'shrinkers' references a 'pan' post from 12:26

The post I referenced was from 12:26 AM, not PM. It's still there. You'll find it here
http://www.fivethirtyeight.com/2009/09/open-letter-to-strategic-vision-ceo.html#comment-3395828532433239640; September 30, 2009 3:07 PM
Mr. Universe said...: But, hey as long as you're interested, here's a portion of a response to an article I did just recently:

Like many things in the past, technology has supplanted the need for cursive. Just as the invention of the pencil and pen bypassed the need for calligraphy. The typewriter likely had an effect on writing as well. Likewise, word processing didn't require as stringent a need for accuracy since you could go back and correct your errors. I remember the days when the correction tape on the old Selectric (tm) typewriters was considered a Godsend. Or how about White Out (tm)?

Incidentally, you might find it surprising to know that the qwerty keyboard (which is actually not that effective of an arrangement of keys) is a result of keeping the old typewriters from jamming up. Qwerty is actually meant to slow you down. Somehow I doubt we'll ever be able to break that habit any more than Americans could convert to metric (despite the fact that this quirk caused the loss of a billion dollar Mars landing).

Personally, I find it difficult to imagine a time without spell check, cell phones, internet, Google, Wikipedia, or remember when we had dial-up modems, less than 8KB of ram, etc. Even libraries may one day become obsolete. Technology does provide us with new possibilities but I also worry that it makes us lazy and stifles creativity.

Take instant messaging speak, for example. I fnd it irritating 2 read IM speak and be reqired to no what LOL, BFF, ROTFLMAO, and other wrds i need to no to fit in 140 chctrs. no what i mean?

Imagine if Shakespeare wrote in IM speak:

2 B or not. That's the ??? WTF?

It makes me wonder if our collective brain capacity is being rewired for the better or the worse. Maybe it's freeing up brain capacity for other pursuits. I no longer need to remember what the capitals of the states are; I can just google it.

So long, fancy writing.; September 30, 2009 3:09 PM
Robert said...: Jake said...

Not sure what this is all about but, I'll jump on the bandwagon. Awesome! (For some reason.)

Also, a stupid nit. Only one space after periods, Nate. :)
-----
Just an FYI, one space after a period is acceptable, but for monospace fonts (such as the Courier New font that Nate uses here), two spaces is the correct usage.; September 30, 2009 3:20 PM
Rob322 said...: Wasn't SV the group that last October released a poll purporting that McCain/Palin was leading Obama by 50 points or more with youth voters? I mean, that's clearly suspicious right there given that every other pollster went the other way and, more importantly, when the election was held it was shown to be utter crap. Something with them is screwy, either accidentally or deliberately and asking them to be accountable for what they choose to say in public (no one forces SV to publish their polls or even conduct them after all) is not wrong, not at all. SV makes the statements, what's the harm in asking them to back them up?; September 30, 2009 3:20 PM
Mr. Universe said...: Oooohh, that's where the name MuleRider comes from.; September 30, 2009 3:22 PM
Jason Henriksen said...: 1) Should liars be punished?

2) Should suspected liars be investigated and called out?

3) In a world where anyone can post anything anonymously, who *should* act as the arbiter or trust?

4) Would you still post if you were required to show your real name, your photo, and your phone number along with the comment? If not, why?; September 30, 2009 3:22 PM
Polymeron said...: I've been rabidly following this story ever since Nate started posting about it; thought I'd finally put in my two cents.

First, I see the question being raised whether or not it's ethical for Nate to be posting about this publicly. The answer is: Absolutely. Nate has every right to make public his opinions and his findings, and call attention to it. Of course, he has the (moral, not legal) responsibility to take seriously any evidence he is presented (or he comes across) that alleviates the suspicions. So far, there has been none - I've done a lot of independent looking up, and the closest SV has come to addressing these suspicions at all was by promising cross-tabs on future polls. And that isn't close at all.

But, and this is the second thing I'd like to address, that doesn't mean any form of saber-rattling is kosher or warranted in this debate. I'm very very interested in seeing how this develops, and specifically, here's what I'd like to see in future related posts:
1. More damning evidence, circumstantial or not, the more conclusive the better. This both serves to clear the fog and to possibly attract more media attention to this.
2. Reports on moves that have a tangible chance of getting much-needed information from SV.
3. Thorough, journalistic work that attempts to adhere to journalistic standards while providing us readers with as much reliable information as possible.

This last post? Doesn't answer to any of the three. Oh, it's entertaining to see the chest-thumping and cast our bets on these sudden rivals, but it doesn't really get us any closer to resolution. There's no new information, the chances of such an approach to garner anything other than hostility are slim to none, and the general demeanor of the letter doesn't lend journalistic credibility to Nate or the site. That is unfortunate, and it's not what I'm here to see.

Am I the only one who would prefer Sean Quinn to show up in GA with a camera, asking some honest journalistic questions to this provoking letter? Or some letters sent out to relevant authorities? Or some more statistic hocus pocus?

You know I'm cheering for Nate here, but please tell me I'm not the only one who thinks that, legal threats notwithstanding, this can be handled in a more respectable manner...; September 30, 2009 3:28 PM
shrinkers said...: Great questions, Jason!

1) Should liars be punished?
Not if they're Republicans, no.

2) Should suspected liars be investigated and called out?
Only if they're Democrats, and the charges are obviously invented.

3) In a world where anyone can post anything anonymously, who *should* act as the arbiter or trust?
Glenn Beck.

4) Would you still post if you were required to show your real name, your photo, and your phone number along with the comment? If not, why?
Yes, it should be required. I can make those things up, too.

< /sarcasm >; September 30, 2009 3:29 PM
bleepul said...: loner, robb,

I disagree and think we can as dispense with the citizen's arrests. The end result is that in a month, or two months, or a year, someone or some group will try to take down 538 under a similar rationale. The site has already begun to move from reason to rationalization and I think that's too bad. It was a good site.; September 30, 2009 3:40 PM
loner said...: Polymeron—

Past attempts strongly suggest to me that it can't be handled in a more "respectable" manner.

Rob322—

That was Investors Business Daily/TIPP.; September 30, 2009 3:42 PM
Dwight said...: bleepul said...
Shrinkers,

In any other profession you would make a complaint and let an organizational structure investigate and judge.

Actually, no. For example scientific fields largely use self-policing and investigation as frontline defense against fraud. More formal overarching organizations and also the legal system can be involved. But that generally comes later, after flags get raised by exactly this sort of investigating.

Further it is very much the role of the press (Nate stratals a couple professions on this blog) to investigate and bring things to the public light.; September 30, 2009 3:43 PM
Minstrel said...: bleepul wrote:

What is AAPOR even for then if it is Nate's job to uncover false pollsters?

---------------------------------------

It is everyone's job to put forth any evidence they have of wrong-doing. If it is actionable wrong-doing then, after the evidence has been put forth, it's up to the appropriate authorities to take action.

You tried to analogize this to doctors "bad-mouthing" other doctors. You're shifting the debate with that analogy...Nate Silver isn't "bad-mouthing" StrategicVision (which implies making unsubstantiated attacks)...he's been detailing statistical evidence that raise questions about the the legitimacy of their polls.

If a doctor had actual evidence of malfeasance by another doctor, I absolutely want that doctor to speak up. The relevant medical board can handle judgment and penalties (like removing licensing) but any doctor with evidence of another doctor's potential unethical behaviour should definitely put it forth.

Saying that it's not someone's job to bring to light evidence of wrong-doing is nonsensical.; September 30, 2009 3:45 PM
shrinkers said...: @bleepul said...
The end result is that in a month, or two months, or a year, someone or some group will try to take down 538 under a similar rationale.

Not likely to have any effect, Bleep. Nate shows his work, which is all SV is being asked to do.

I don't doubt the right-wing fruitcakes will try to swiftboat Nate - they're good and inventing lies and other idiocy (think: birthers, death panels, cutting Medicare, Federal funds for abortion, "gummint takeover", other Whignut nonsense). And they're more rabid about it the more effective someone is at exposing their cr@p.

But if I were Nate, I'd be flattered at any such attempts. It would prove he's been doing the Right Things all along.; September 30, 2009 3:47 PM
bleepul said...: Mistrel, Dwight,

Yes. You are right! I don't have a problem with questioning the data. I don't have a problem in reporting it. I have a problem with the letter. It's personal, petty, silly, antagonistic, and self-destructive. It's not a good decision.; September 30, 2009 3:51 PM
Michael (mbw) said...: Great to have a real discussion running through this thread.

@ Mark Grebner: Here's what I meant. We have a sequence of 10 occurrence numbers N(0)...N(9). This sequence can re-expressed as 10 Fourier coefficients:

F(0) (the irrelevant mean)

F' and F" (1 through 4)These are the two coefficients (real and imaginary or sin and cos, however you want to call it) of the sine waves with 1 to 4 periods in the 10 number sequence, i.e. periods 10, 5, 3.33, and 2.5.

F(5): the single coefficient for the period-2 sine. (even-odd difference).

Now the sorts of things people have mentioned that would cause non-random coefficients other than F(0) would show up almost entirely in F(1). F(5) might also show some artifact of the rounding algorithm. That leaves F(2 through 4, 6 real coefficients, which should be almost entirely due to random sampling statistics for the range of N considered. The net variance from pure randomly sampling shows up as the sum of the squares of the 9 coefficients other than F(0). So the 6 that we should look at should have 2/3 of the total random variance, which you as a statistician can easily calculate. The form of the distribution function for the sum of the squares, x, would look like
x^2 exp^-x/a where a is 3*mean of x. (You have some name for that distribution.)

Another way to do this is to Fourier transform, remove the suspect coefficients (F(1) and maybe F(5)), then inverse transform to get a time series whose variance can be taken. Same thing, really.

@shma- Yeah, but leading digits are really different from trailing digits! They correspond to very different log ranges, unlike the trailing digits.; September 30, 2009 3:52 PM
Shaun said...: While I support Nate, I think he's going about this nearly as wrong as possible.

I could not recommend this book to Nate enough: "How to Win Friends & Influence People"

In one of the example stories in the book, before Abraham Lincoln was a politician he publicly lambasted a person in the local paper and embarrassed him to the point that this individual challenged Lincoln to a duel--he was only spared from a fight to the death by their seconds calling the fight off at the last minute. Needless to say, Lincoln was much more careful from then on and took another approach to resolve disagreements.; September 30, 2009 3:55 PM
bleepul said...: Shrinkers,

God, can we move on from the swiftboat and birthers and all the other nonsense? The left has just as many idiotic positions and it's pointless to continue to throw them back and forth. Rabid liberalism isn't going to do anything for this country any more than whignut philosophy.; September 30, 2009 3:55 PM
bleepul said...: Michael,

Is the distribution the third moment?; September 30, 2009 3:58 PM
Dwight said...: bleepul said...
Mistrel, Dwight,

Yes. You are right! I don't have a problem with questioning the data. I don't have a problem in reporting it. I have a problem with the letter. It's personal, petty, silly, antagonistic, and self-destructive. It's not a good decision.

While the tone is a nose tweaker (remember, two professions at work, and at least two audiences ;) ) it very quickly gets to the heart of the matter. "Show me the data", and "your threats of a law suit are empty bluster."; September 30, 2009 4:01 PM
bleepul said...: Dwight,

Seems like someone trying to impress their friends (or draw traffic) to me.; September 30, 2009 4:02 PM
shrinkers said...: @bleepul
God, can we move on from the swiftboat and birthers and all the other nonsense?

Geez, I sure hope so. Maybe the Republicants and trolls will just shut up. You think?; September 30, 2009 4:03 PM
Michael (mbw) said...: @Bleepul- I don't follow you. A distribution is not a moment. The third moment of the distribution isn't even particularly interesting here. I meant that I think that's a chi-square distribution, but not being a statistician didn't want to get the nomenclature screwed up.

But while we're at it, I don't see the point of most of what you've been arguing. The only way these things are ever caught is by volunteers. Perhaps you are not a scientist and don't understand the visceral horror we have at having actual fake data contaminating our results, despite the errors we are used to making.; September 30, 2009 4:05 PM
bleepul said...: Shrinkers,

Doubtful, but you can certainly be the one to start. I think the Democrats would have far better success if they reasoned rather than started and ended every sentence with "After 8 years of Bush".; September 30, 2009 4:06 PM
shrinkers said...: Yeah, but, after 8 years of Bush, what else should we do???; September 30, 2009 4:07 PM
bleepul said...: Michael,

I actually have a PhD in biophysics but was just trying to sound smart with the distribution comment. My statistics is admittedly weak and I have nothing of value to add to your discussion. Thought I might get lucky ... haha.; September 30, 2009 4:10 PM
bleepul said...: Shrinkers,

If I were a democrat I would let Bush go ... he served his purpose and you got a very liberal president as a nice prize. I would spend more time worrying about WHY people believe in death-panels and participate in tea parties and address their concerns rather than labeling them quack jobs. When that happens most of us in the middle see the left as just as messed up as the right.; September 30, 2009 4:13 PM
shrinkers said...: Mule
However, the way he's going about it is doing nothing more than filling some of the most rabid partisans in the blogosphere - the Pragmatuses, shrinkers, Dwights, loners, etc. of the world

I am honored to be named in such company. Thank you. Honestly.

Clearly, Nate has struck a chord, requiring response from the concern trolls. His witty invitation to SV, to put up or shut up, is principled, honest, and straightforward. I was impressed.

And I am glad there is someone like Nate keeping the pollsters honest. Clearly, he has a need to have reliable data - and so do we. Nothing else has had an effect to get SV to show their work. Maybe this will.

If not, as Nate said, the lawsuit should be fun!; September 30, 2009 4:14 PM
Juris said...: Mule said...

Hey Mule, are you from Jonesboro Arkansas? How is weather down there?

Uh, you're in the neighborhood. I was raised and went to college in Magnolia, Arkansas.

See:

http://muleriderathletics.com/

I'm now living in Bartlett, TN, a suburb of Memphis.

Observation:

Just 10-11 miles from your place of work, it would appear.; September 30, 2009 4:16 PM
loner said...: Mule—

I'm hurt.

Gales of laughter; September 30, 2009 4:16 PM

Pollsters and Poll Compilations

Other Resources

Wonks

Blog Archive

Labels

9.29.2009

An Open Letter to Strategic Vision CEO David Johnson

268 comments