The public relations firm Strategic Vision, LLC, which we have raised pointed questions about for the past several days, lists on its website a mailing address of 2451 Cumberland Parkway, Suite 3607 in Atlanta, Georgia and describes itself as being "headquartered in Atlanta". However, the firm's physical offices appear to be located some two hours away in rural Union County, Georgia, in a motel park in Blairsville, far outside the Atlanta Metro Area.
The website for the Union County GOP, for which Strategic Vision co-founder Laura Ward Johnson is listed as a webmaster, recently advertised a BBQ, for which tickets could be purchased at the Strategic Vision offices. The address listed in association with the offices is:
Strategic Vision
22 Town Square, Suite 6
(Seasons Inn Plaza)
This appears to match the listing for the Seasons Inn Motel & Plaza, which is located at 16 Town Square in Blairsville. Laura Ward Johnson had also formerly listed the 22 Town Square address in a posting at the Union County Republican Party facebook page, although the reference has since been edited. Last year, Laura Ward Johnson and David Johnson, the husband-and-wife team that founded Strategic Vision, gave an interview to the Philadelphia Inquirer in which they indicated that they had moved to Blairsville from Marietta in suburban Atlanta.
The firm's Blairsville offices are located some 1 hour and 48 minutes away from the mailing address the firm lists in Atlanta, which in fact appears to be location of a UPS Store. Blairsville is a small city (pop. 659) near the North Carolina border and the Chattahoochee National Forest, and outside of the Atlanta-Sandy Springs-Marietta metropolitan area as defined by the Census Bureau. It is approximately as far from Atlanta as Milwaukee is from Chicago, or Philadelphia is from New York City.
A photo of the Strategic Vision office, as posted at the firm's facebook page, is reproduced below. The address listed in the photo indeed appears to be "22 Town Square", although the image is not of a high enough resolution to say so definitively.
In an interview to the Washington, DC daily The Hill, David Johnson claimed that he and his employees had been harassed and threatened by readers of this website. Although I frankly find these allegations somewhat dubious, I would nevertheless urge everyone to strictly and completely avoid any unsolicited physical contact or any verbal threats of any nature with Strategic Vision and its employees, and would further urge the Blairsville police department to keep a close watch on Strategic Vision's offices and make sure that everyone's security and peace of mind is protected.
Thank you to the readers of this website and Pollster.com, which have assisted with the development of this story.
9.27.2009
Strategic Vision Office Is Located in Rural Blairsville, GA, Not Atlanta As Firm Claims
@ 5:35 PM
Subscribe to:
Post Comments (Atom)
131 comments
I can't wait for the take down of Rasmussen as fudging his negatives of everything regarding the Democratic party.
It'll be a hoot.
In the photo of the office, the area containing the "Strategic Vision" company name has very difficult pixel structure than the rest of the image. Though I am fairly expert at graphic manipulation I am not a photo retouching expert. Does anyone else see this?
@Thatcher: I'd guess that's because the name is a cheap plastic sticker attached to the glass; in other words, I'd guess the optical effect you're noticing is also present in person.
Sounds to me like they rented out a mailbox (at the UPS store) to make themselves look like a more centrally located firm, while locating their actual offices farther out to save money.
@Juris: Like I said over on Pollster, the phone number is actually a personal phone number. So...um...while it is listed, and while it does come up in Google results, it might be wise to remove it from future usage.
@Thatcher: Over in the last SV thread, I think someone pointed out the fact that it's JPEG format. JPEGs can do that on occasion, so that in itself isn't indicative.
Actually, it may be an indication that it's genuine. There's some loss around the address on the door as well, and that looks similar to the pattern around the business name in the window.
So after you were told that Strategic Vision had been harassed by readers of your site, you decided it was a good time to tell your readers how to get to their offices? It seems unlikely that they're in any physical danger from your readers, but unlikely is not impossible.
On a side note, I'm sure there are more than a few organization headquarters in Arlington or Silver Spring that publish a DC headquarters address. I'm not sure why Strategic Vision's "real" address was worth pointing out as some sort of insidious pattern of deception.
I imagine one reason Nate's seeing threats of lawsuits from SV is that companies of all political persuasions who contracted SV for polls in the past will take them to court to get their money back.
According to Google Maps for "2451 Cumberland Parkway, Atlanta, GA" and going to street level view, the address 2451 can clearly be seen in the window of a Publix food and pharmacy beneath the large external sign and in the window next to a red sign
Kevin, most companies using PO Boxes as their office addresses do it because the run their business at a less professional sounding address and think having an address of "Suite #109" makes them sound more legit. It's not because any particular city sounds less important. I'd be highly surprised if any business with a legitimate business address in Arlington or Silver Spring would bother getting a PO Box just so they could say DC.
Besides, your comparison isn't a great analogy, given that both cities you named will be pretty much DC in the minds of anyone not from the area. No one could reasonably say Blairsville is pretty much the same as Atlanta.
I feel like if they're claiming that "[David Johnson] and his employees had been harassed and threatened by readers of this website," then it might not be a good idea to post something that comes across as "Here's where you can find them!"
While I understand your intentions here are to provide further questions on their legitimacy, I don't know if this is the best way to do it.
@Kevin: Is Arlington or Silver Spring two hours away from D.C.?
Yeah. Didn't think so. (Both, in fact, are immediately outside of DC. 10-minute drives, in fact.)
To add on to the comparison, it's about the same distance as Newark, DE is from Washington, DC.
(Incidentally, Google Maps has now taught me where Bel Air is. I imagine that Will's taxi fare must have been heinous.)
It seems unlikely that they're in any physical danger from your readers
If I've followed this the claims about harrassment were made prior to the true address being ferreted out, and the address removal in question was for the PO box in Atlanta. As well wasn't it some "religious" person or something? *shrug*
Now they might have been refering to some entirely separate incident and the claim for all the maildrop addresses is to hide from people that are unreasonably pissed off at them.
Frankly the extra satelite addresses being PO boxes does make some sense if the are using parttime or causual contract work in those cities.
The HQ in Atlanta thing I'd even write off as PR bullshit. Not exactly ethical in a larger sense but not the end of the world. This still remains about the data and polling claims.
P.S. I find it a bit more interesting the ties to the GOP. Not exactly end-of-the world stuff either but something interesting to note.
The pixil anamolies in the sign is extremely clear and is in that area only. It is not attributable to the jpeg format itself. It could conceivably be different reflectivity from plastic, but I don't think so. I've edited a lot of pictures and never seen anything quite like that. We need a photo retouching expert.
Google says that Papa's Pizza is 28 Town Square #11. I'd bet that the SV storefront is somewhere in that shopping center, maybe across the parking lot from Papa's. The structure of those buildings looks pretty close to the one in the photo.
http://tinyurl.com/ybvvyba
Sorry for the typos... Here it is again. The pixel anomalies in the sign are extremely clear and are in that area only. They are not attributable to the jpeg format itself. They could be caused by different reflectivity from plastic, but I don't think so. I've edited a lot of pictures and never seen anything quite like this. We need a photo retouching expert.
Inferno…
Bel Air is about three miles from me, just the other side of Beverly Hills and adjacent to UCLA.
:o) :o) :o)
Seriously, this focusing on the firm’s addresses is distracting from what’s really at issue—whether the “polling” done by SV is really polling or just pulling numbers out of the air. While the statistical analysis posted here is very helpful toward resolving that, the flimsy addresses the firm maintains is a sideshow.
If you can prove that SV is slavishly devoted to seeming more sophisticated than they really are, you’re still no closer to nailing them on what matters—whether they really conducts polls or not.
The fact that someone lives next to a junkyard doesn’t prove that they collect junk.
Johnson's accusations of "harassment" appear to be a red herring at best, and hell, fabricated at worst. Most likely case is that someone in the office next to theirs heard about this blog and then teased or joked with someone from the company.
@Thatcher
A person named Thad addressed that in the last section you posted the problem; here's his reply verbatim:
With respect to thatcher's question about the facebook image
http://www.facebook.com/photo.php?pid=2831370&id=66419995840
That pixelization around the logo is a typical characteristic of the discrete cosine transform underlying the Jpeg format. Where there are high frequencies in the image, you get more noise. The DCT uses an 8x8 block, and you will notice that the noise level abruptly changes on 8-pixel boundaries.
Can't deny it as real on that basis. I'm no photoanalyst, but am a visual effects supervisor in real life
On top of that, I personally can't see why they'd want to fake their address somewhere it isn't supposed to be. It'd make more sense if they faked it somewhere in Atlanta, but odds are the photo is probably real.
Thatcher…
The point about the pixels is—who cares? Why would that matter even if they do represent some sort of manipulation? First, why would anybody bother to do that, and second, who cares?
What a bizarre story this is turning out to be. On the one hand this Johnson character is intelligent enough to understand the potential financial gain and brand exposure in creating a polling arm; on the other, he is so dumb as to assume that no one would eventually find out his methods were fraudulent.
Am I the only one who is turned off by 538 playing private investigator in the last few days? I want political commentary and statistical analysis of polls, not analysis of blurry satellite pictures and accusations based on tenuous proof.
If SV's numbers suck and/or are totally made up, you should have a regression model that picks this up. End of story.
It’s also inane to suggest that Nate is exposing SV or its employees to danger by reproducing a photo from SV’s own Facebook page, and reiterating addresses that SV itself is perfectly willing to display publicly.
The suggestion that someone was inciting violence by using a firm’s own representations of its locations would be laughed out of any courtroom.
Am I the only one who is turned off by 538 playing private investigator in the last few days?
While I'm certainly in agreement with Pragmatus that the focus is about the polling methods and data, I'm willing to give this side detour a little leeway to see what shakes out in regards to the polling data.
If you want to understand the 5 W's of party's actions it behoves you to know about that party. Even stuff that is relatively benign.
LAW said...
Am I the only one who is ...
~~~~~~~~~~
Probably, take care
p.s. who doesn't like a mystery ...
@foaloke
Thanks. Missed the previous post. Yes it could be the reflectivity (high frequencies) but the boundry would change abruptly just like this if the section were clipped and re-pasted.
I was wondering if there was an office at all... could this be a stand-in.
Having said that, the google map street view looks an awful lot like the office photo so I think the point is moot. I think we've put this one to bed.
The funny pixels around the text is very typical of a jpeg artifact from a highly compressed jpeg. That it appears most vividly around the text is also very typical - that's the sort of fine, high-contrast detail that shows that sort of characteristic pattern. That's a non-issue.
I've been to Blairsville, which is just a few miles down from Neel's Gap, on the Appalachian Trail. Maybe these guys advise with Gov. Sanford.
Blairsville! What a hick hole. about ten years ago I was involved in a vehicle accident in Blairsville, it was equivocal, at best, but the local cop wrote a favorable police report for the local boy. My insurance company believed me and fought the claim, I was afraid I'd have to go to trial there. Fortunately, the old boy settled and I was spared a return.
"I want political commentary and statistical analysis of polls, not analysis of blurry satellite pictures and accusations based on tenuous proof. "
Dude, have you actually been reading the website these last few days? The final digit analysis was pretty damning and as far as I am concerned the Oklahoma students poll was concrete proof of fraud (11% of students making the exact same wise ass "Communists and Republicans" joke, scores of students guessing that there were 6 or 8 supreme court justices but none guessing 7, not one out of 1,000 students getting more than 7 very easy questions right, everything adding to 100% with no "other" category etc. etc.). As far as I am concerned fraud is basically proven at this point. What more do you want?
It is true though, that anyone who still believes they are a legitimate polling organization is not going to be convinced otherwise by proof that they fudged on the location of their headquarters.
I'm loving every word of this saga. This is what 538 is supposed to be about - Politics Done Right. When someone is suspected of fudging numbers, while Nate's trying to uphold his own reputation through careful analysis of those numbers, it's only logical for Nate to throw the flag when someone tries to make a mockery of the field of study altogether. This saga is perfect evidence of the growing value of the blogosphere - this story would have gone nowhere if not for Nate's persistence. Rock on. I hope this won't be the last time we 538 readers get to see bullshit get called out as just that, bullshit.
jpeg uses an adaptive block size in compression. The pretense of sharp edges in the lettering drives more detail which shows up as the higher contrast pixels. The smoother surrounding blocks are from less detail and more interpolation for display. You can see similar transitions from "pixelated" to smooth all over the image (up and right from the door for example, or around the leafy thing.)
more subtly, the angle of the lettering (slightly down to the right) seems about right by eye .. someone with right tools could put in the perspective lines and check. But if it is a fake, it isn't a hack job.
I agree that the resolution-on-the-sign issue is inane and really pretty trivial, but the confusion over where SV's office is located does illustrate the larger point. This organization posts inaccurate information about itself, presents highly suspicious polling data, and refuses to disclose an accurate methodology. Examining this information seems perfectly within the realm of a polling analysis website. Nothing is proven at this point, but hopefully 538's analysis leads organizations to question whether or not they want to use SV to conduct legitimate polling research.
Guys, you should check out Town Square in Blairsville on Google Streetview. The town is tiny. It's not just a small suburb of Atlanta; it's Mayberry. It's very difficult to imagine that this is a convenient location for a major national polling firm spending millions of dollars on surveys. It looks like the kind of place that has a single FedEx box and only got broadband access last week.
It's not damning but it's certainly a bit fishy.
I'll agree that the location issue matters a little bit, honesty-wise and all, but it just seems so much less important than the numbers analysis Nate ran earlier. Making an entirely new post just to call them out on fudging their office loc seems overly aggressive and entirely unnecessary, when the numbers speak so elegantly.
A few tidbits from Wikipedia to follow up on my previous post:
Blairsville population: 659
Median household income: $14,120
Again, this is where you headquarter your prominent national polling firm?
Don't knock small towns. They are sometimes very loveable.
And some of them grow into big towns and cities.
Forty years ago, there was this small town in Arkansas called Bentonville (larger than Blairsville is now, to be sure), and then Sam Walton founded Walmart, with its world hq right there; add Sam's Club, and Tyson Farms nearby and your have a burgeoning little (if sprawling) city.
David Johnson quote taken from an article on The Hill, in which he responds to Nate's criticism:
“He’s got a good reputation, but I stand by our work,” Johnson said. “We’ve done the work, and we can prove that we’ve done it. The other thing is – if you look at our results, we’ve been accurate.”
Does that sound like an individual who is running a legitimate polling operation? This is an issue of transparency, methodology, and he offers a generic "we stand by our numbers" and "we've been accurate." Can you imagine, say, Zogby or Rasmussen responding in such a way? They would file a $100 million dollar libel suit in a nano second. Strategic Vision is CLEARLY a fake operation.
Okay, to hop off the sign res issue (because honestly, I'm not convinced that it's a doctored picture - though I think Thatcher can tell from some of the pixels and from seeing quite a few shops in his time):
The thing I don't get is...why hide the fact that you're headquartered in East Bumfark, GA? (Okay, NW Bumfark. WHATEVER.) I don't find it weird that they're headquartered in Blairsville - I find it weird that they're headquartered in Blairsville and their nearest mailing address is a PO Box in Atlanta.
Anyway, I'm done with this post for now. I might try the algorithm I was thinking of in the last post soon and post results.
Hey, I grew up in small towns (although, again, still bigger than Blairsville). I like them, and I even understand the appeal of putting businesses in them. And if Blairsville was even somewhat nearer to a big city, I'd be perfectly willing to believe that Strategic Vision just wanted somewhere cheap and less hectic to set up shop. But this town appears to be way out in the sticks. This is not somewhere that it seems very easy to get into and get out of. In particular, it seems like Blairsville would lack the infrastructure for the sort of transaction-heavy, communication-intensive and current-events-oriented business that SV does. But I've never been there, so I could easily be wrong.
One thing about being in the information industry is that you don't have to be located anywhere in particular. I'd bet the internet connections are every bit as good in Blairsville as they are in ATL.
I think Nate posted several months ago that he could write 538 from just about anywhere in the world. But if he wants to be on TV regularly, and to interact personally with political and media and academic types, he's much better off being in or near a large city.
If you want to be in the polling business, you can contract out 98% of the work to "Aunt Millie's telephone bank," to some survey designers and stats jockeys living in, oh, Gatlinburg or Macon or Durham or Tallahassee, or, um, Madison. You don't need to have a large staff under one roof.
So I think the issue of where SV's HQ is located is totall irrelevant in evaluating the quality of the business. The slight deception in making it seem like he has real offices in Seattle, Madison, Tallhassee, and Atlanta may be an issue with respect to general credibility but it has no bearing whatsoever on whether SV's polls are good ones or not.
I really don't know what this post from Nate accomplishes, other than (1) to question the veracity of Strategic Visions' public relations, and perhaps (2) to show it as a rinky-dink outfit. Even if both of these things are true, is fivethirtyeight the place to proclaim them? A Georgia-based investigative journalist might do well to pick up on the story, but I think we all agree that Nate is not a reporter, and that this is not an online newspaper.
I think it would be far more effective to focus on the firm's inaccurate and/or poorly conducted polling, and leave it at that.
Have to agree with Kevin and Peter in that the finish to this column comes across as sneakily Michelle Malkinish (and yes I just made that word up) although I'm certain that is not 538's intent.
I agree with everyone else in that SV is just being paranoid. However, the way this column still makes 538 look bad.
Google street view apears to show the shopping center and motel in Blairsville: http://tinyurl.com/yc6n7mp
Quality is poor but careful observation shows the same stone bases beneath the store windows, similar parking (head in), covered entryway, etc.
Juris said...
One thing about being in the information industry is that you don't have to be located anywhere in particular. I'd bet the internet connections are every bit as good in Blairsville as they are in ATL.
That's possible, but I actually wonder if that's the case. My parents still can't get high speed internet in rural Missouri.
This isn't about rural vs. metro - it's about truth and consistency. Where are SV's offices? Simple enough question. Silver's not judging the value of the local town, he's asking because Johnson has a history of listing fake addresses to appear like a larger operation AND has made a serious accusation that one of Nate's readers threatened a staff member in his office parking lot.
Johnson has may have reason to obfuscate their location simply to conceal the fact that it is a two person operation, with no resources to produce the volume of polls they do.
I'm surprised that so many of you bought the bogus line about the threat to begin with. Where is the police report?
I live not for from the town in question. Probably one of the few 538 reader's who do. Blairsville and most of the counties in the area do indeed have high-speed internet. DSL up to about 12Mbps (I think thats top speed) It is the headquarters of United Community Bank a fairly sizable regional bank chain. The town is also a short drive from Young Harris, the birth place and residence of former Gov. and Sen. Zell Miller. There is also a private college in Young Harris where Miller attended and has taught (not to mention many prominent Atlantans and other Georgians and/or their children). There is also a lot of summer homes for Atlantans, and retirees from FL. and other such places. NASCAR racer Bill Elliott also keeps a farm in Union COunty and other names in local as well as semi-national (Weather Channel people) supposedly have homes in the area.
Not sure how much use any of this is but it is something to be considered when pondering why Strategic Vision located offices there.
It is rather common for companies to essentially have a fake address, especially in the south. The main reason is an Atlanta address "sounds" more credible then an address elsewhere (say Blairsville) people automatically know where you are generally seem to be located and with Atlanta's international reputation, it just doesn't get questioned.
What really bugs me though is apparently their "corporate offices" is a converted hotel room. This seems to preclude them even having "employees". Sure the whole lot could work from home but it seems like they have a physical address for tax reasons more then anything.
If anything, news of their "official" setup makes the company smells even more as they clearly don't even have the infrastructure to support a proper polling company. Of course if you are just fudging a bunch of numbers, then that office is all you need.
Bingo!
Fake polling: interesting. Fake address: uninteresting.
Thought of the day:
Shouldn't it be Republicans who are really pissed off at Strategic Vision? They clearly punked OCPA, which means they've probably also been scamming their Republican clients, taking money in exchange for phony data that could lead their clients' campaigns astray. SV could be the biggest scam to target Republicans since BMW Direct.
Law enforcement should probably be taking notice at this point. Not that Johnson would be punished; he probably claim political asylum in Iran.
I think many of you underestimate how easy it would be to become a pollster. You don't need a large staff. You contract out for most services. You should have a head for finance and a salesman's mentality, hopefully also knowledge of the technology of survey design and sampling, and a modicum of statistics.
An MBA and/or Engineering degree would do just fine. You build your business with just a small working staff.
You don't have to take on sophisticated sample or survey designs. You're not seeking contract work from the CDC or grants from NIH or NSF.
Of course you could still get way over your head -- as I believe SV was on those citizenship polls. Nate has the right word for what they did. I would add one more: flimflam.
Looks like a lot of you guys missed brown's post, which is a pity. The building in the picture matches the color and materials of the buildings on the right here:
http://tinyurl.com/ybvvyba
...down to the stone decoration at the bottom. Since it's pretty certain that such a building exists at that location, I don't see any reason for SV to doctor the rest of the image. Nice try playing internet detectives though, guys ;)
@inferno
Been trying to keep up with all this and maintain a real life but keep getting bogged down in the wrong post. I've been posting at other sites, too (I'll keep your real name private). Nice to get a shout out for contributing to the dialogue. Looking forward to tomorrow's events
Visited Blairsville when I hiked the Appalachian Trail in '99. Nice place to retire. Apparently a nice place to run a two person sockpuppet polling/PR firm as well.
A Georgia-based investigative journalist might do well to pick up on the story
But will one. All the press appears able to do so far is print yet another phone interview with Johnson.
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.
So I made this point in a previous post but it bears restating.
Our expose of the 'office' locations of SV LLC is indicative of a theme of disingenuous intent of the entire operation. Their website seems to indicate that they are a multistate company with many employees and substantial revenues and resources. Just as that is misleading, how; with such limited resources, can we believe the findings of their polls without the methodology?
And then today there was this (from a previous post):
Apparently SV LLC in trying to do damage control has made a public commitment to release all future data on their polls; a song we've heard before.
From the Atlanta Journal Constitution:
[Releasing Crosstabs] But even if it increases the second-guessing, the added transparency boosts public confidence in the results, [Insider Advantage CEO] Towery said.
Events of the last week have caused Strategic Vision to come to the same conclusion about its future polling.
“We’re going to release all the crosstabs, and put an end to this right now,” Johnson said. “That will squelch anybody from saying anything.”
How about that past data, Mr. Johnson. You got totally outed as a fraud. Tomorrow is going to be a landmark day in polling ethics.
http://blogs.ajc.com/political-insider-jim-galloway/2009/09/27/strategic-vision-promises-crosstabs-with-every-poll/
Add this to the Oklahoma data discrepancies and the obfuscations just keep piling up. Everyone seems to agree that something smells fishy. The stench is just getting worse. I'm hoping SV INC weighs in on this.
Strategic Vision CEO David Johnson said Silver’s suggestion isn’t true.
“He’s got a good reputation, but I stand by our work,” Johnson said. “We’ve done the work, and we can prove that we’ve done it. The other thing is – if you look at our results, we’ve been accurate.”
Yeah, we need to see the data and the methodology. The results are what we're calling into question. 'Because we said so' ain't gonna work here. As any undergrad stats student would be expected to do, show your work.
Oh SV LLC, you know you've fubarred past salvation when even dudes like BDP and NJwingnut and others won't come to your defense.
Fascinated to see what the MSM does with this Monday morning. If anything.
What would be great is if some right-leaning news outlet gets really pissed with SV LLC and runs with the story out of vengeance.
wv - kroox. no comment.
Kevin said...
I'm sure there are more than a few organization headquarters in Arlington or Silver Spring that publish a DC headquarters address.
That used to be the case - even many residents of Chevy Chase and Bethesda used a Washington, DC address, but a proper Zip Code for Chevy Chase or Bethesda.
Many businesses and residents also used to use a DC (202) Area Code for their telephone number when they were physically located in the Maryland suburbs (301 Area Code) or in the Northern Virginia suburbs (703 Area Code).
That is not the case now, as the United States Postal Service and (then) Bell Atlantic (now Verizon) made sure that the mail and telephone numbers were correctly listed (used to be, until the late 1980s, that anyone dialing Area Code 202 could call ANY phone in the DC Metro area, whether the actual area code was 202, 301 or 703).
BTW - there are very few 'Beltway Bandit' firms who would list a Silver Spring address - most in Suburban Maryland are in Bethesda, Chevy Chase, Greenbelt or Rockville - popular perception gives those addresses much more 'cachet' than Silver Spring, and transportation options (VERY important to the beltway bandits) are generally much better for those locations than Silver Spring.
So, nice attempt at defending SV LLC, Kevin, but again one of our usual suspects fails in their attempted defense.
Mike in Maryland
Inferno said...
@Kevin: Is Arlington or Silver Spring two hours away from D.C.?
Yeah. Didn't think so. (Both, in fact, are immediately outside of DC. 10-minute drives, in fact.)
Inferno -
Traveling from Silver Spring to DC can be as quick as traveling across Western Avenue or Eastern Avenue, depending on your exact location in Silver Spring, as they share a common border. I think the 'dividing line' is the middle of the street. One step, or maybe crossing six lanes of traffic, but definitely NOT 10 minutes.
As to Arlington County (yes, Virginia, Arlington is a totally urbanized county in the state of Virginia, crossing the 14th Street Bridge, Memorial Bridge, Roosevelt Bridge, Key Bridge, or Chain Bridge will take you from DC to Arlington. Let's say busy traffic (but still moving) and you have traveled from DC to Arlington. Using Metro might be even much quicker. For example, going from L'Enfant Plaza to Crystal City (in Arlington) takes about 9 minutes. National Airport is only 11 minutes from L'Enfant Plaza by Metro.
Getting to Alexandria (City) or Fairfax (county) will take longer than 10 minutes, however, as not that many 'beltway bandits' are that close to the Metro in those jurisdictions, so travel by car is almost always necessitated.
Mike in Maryland
Real Clear Politics removed Strategic Vision from their polls back in October 11, 2008.
http://www.realclearpolitics.com/articles/2008/10/use_of_partisan_affiliated_pol.html
Just John said...
Fascinated to see what the MSM does with this Monday morning. If anything.
I just did a Google news search on 'Strategic Vision', and got several hits - most political blogs and such, but did find some reporting on Strategic Vision LLC from:
Atlanta Journal-Constitution
Slate
Tampa Bay.com (web site of the St. Petersburg Times)
The Hill (not major media for the public, but MAJOR media on Capital Hill)
Washington Post
Interesting to note that some sites, such as National Review Online, NJPoliticker, and other right-leaning (at best) are quoting "Strategic Vision" polls, but not identifying WHICH Strategic Vision they are quoting. You have to do some link clicking through several layers of links to find out the real source.
As an aside, I've found what information AAPOR was asking for, and SV LLC refused to supply. All the requested information was about the polls taken prior to the 2008 New Hampshire Democratic primary, and AAPOR was trying to determine why the polls were so far off. The specific information requested was:
- who sponsored Strategic Vision’s survey(s)
- a description of the sampling frame
- an account of how likely voters were identified and selected
- response rates
- a description of any weighting or estimating procedures used
It sounds to me like most or all of those questions can be provided without giving away any 'corporate secrets', unless you're trying to hide something.
Mike in Maryland
...this is addictive.
@Rory: That's actually pretty inconclusive. They removed PPP and DemocracyCorps - both of which are Democrat-affiliated pollsters - as well because of their partisan affiliation. Which was the same reason that they dropped SV.
And it appears to be standard procedure for RCP to do this - in the weeks immediate to an election, they'll use only non-partisan affiliated polls.
@Mike: Well, the driving directions WERE to downtown DC (i.e., the Potomac), and I believe around where the White House was...
The point, I hope, wasn't obfuscated, though - both actually border DC. It's like...say, having an office in Jersey City or Yonkers and then a NYC address. (Maybe not Yonkers, as that actually borders the Bronx - which, while it's part of New York City, does not have the New York, NY address. Perhaps something more analogous would be the Bronx itself or Staten Island.)
Good work, Nate! Go for the throat (figuratively speaking, of course).
Here's an illustration of foaloke's explanation of JPEG compression:
http://imgur.com/55F0H.jpg
The image on the left is the original (compressed again), and the image on the right has been slightly modified, and compressed with JPEG compression. You can clearly see the pixel artifacts are just block-size changes where the image gets more detailed around the font.
I did this compression with PS4's JPEG implementation, level 4, which may differ a little from whatever software (probably in-camera) was used on the original image.
My analysis: no doctoring.
A Motel Park? In a small town near the Georgia-North Carolina border? Sounds seedy, very seedy. Plus that is some scary country and very backwards and/or conservative. Nate I wouldn't accept any invatations to go there to meet because they'll probably be planning a necktie party for you! Like I said scary!
Why doesn't someone call Papa's Pizza to see if Stragetic Vision is located down the building?
28 Town Sq # 11
Blairsville, GA 30512-8562
(706) 745-1999
http://tinyurl.com/ybvvyba (link from Andrey Fedorov)
While we're at it, could see if pizza orders have been placed by religious fanatics (or the FDIC).
Inferno,
The driving directions for DC are usually based on the Zero Mile Stone, which is placed in the Ellipse, directly South of the White House.
So strictly speaking about driving from (or actually near) the Zero Mile Stone, NoVa is about a mile away, while Maryland would be (most direct and thus closest in driving time) would be to Silver Spring, straight up 16th Street to the intersection of 16th St., Georgia Avenue and Colesville Road, a distance of about 7.5 miles, and about 25 minutes of driving time (in non-rush hour traffic).
The reason I chose L'Enfant Plaza for Metro is because there are several federal agencies HQed within a block or two of the station. It's also much, much faster and easier to go to the Pentagon or Crystal City (where many federal offices and beltway bandits are located) by Metro than trying to drive (plus park) to either (the escalators at those NoVa Metro stations take you directly into the Pentagon, and at Crystal City, you are taken directly into Crystal Underground, a tunnel network that links all the CC buildings).
I know the Silver Spring area fairly well, as I used to live and work within blocks of that intersection. Also, Crystal Underground is a major shopping area for that part of NoVa.
Mike in Maryland
Interesting - I always thought Strategic Vision the Pollster was the http://www.strategicvision.com/ - a San Diego based company. I had no idea there were two separate corporations.
Seriously...this is going nowhere good. Up until now you could always argue that while 538's commentary is left-leaning, it is a site that deals in facts and sound statistical conclusions. What has been going on in the last few days is everything but fact and sound conclusions.
I don't know where this came from, but this whole escapade makes me embarrassed to be a reader of this site. And I've been reading this site daily since the start when it spun off from Poblano's excellent posts on dKos. I know it is fun tracking clues and playing detective and such, but it just seems there is a very strong chance this blows up and there is egg on the face of Nate and this site.
*Maybe* Strategic Vision is a totally fake organization that is just some guy conducting an elaborate scam. But if I were a betting man, I would bet that SV is a real company, and that their numbers are justifiable in some way or another. And I would bet that in the next week, the reputation of this site is going to take a big hit. Just my $.02
@Matthew Fedder
Yeah, there's a few companies with the name. The San Diego firm has been in business for 30 years or so. They're aware of all the stuff that's happening.
SV LLC is a .biz address. There's a suspicion that SV LLC is piggybacking off of SV INC's cachet.
Interesting that the suspects we would expect to come out to defend SV LLC (after all, it is an extremely GOOPer-leaning organization) are completely silent, but still there are defenders of SV LLC on this site.
And to think that SV LLC could end all the speculation RIGHT NOW by releasing the information that, if the polls were conducted honestly, would clear them in almost no time. Since they won't release the information, and it doesn't seem to be 'corporate secrets' type of information, the only blame I can see on who's extending this situation is SV LLC. The more and longer they stall, the worse they appear, especially if they eventually release information that shows they had nothing to hide in the first place.
I'm still amazed at how far into the ethical cesspool the GOOPers have fallen, be they SV LLC, deathers, birthers, 10thers, birchers, secessionists or the TROLLs that are common on this site.
Defenders of SV LLC? How much are you being paid to post your GOOPer trash here at 538.com? And by whom?
Mike in Maryland
I'm a little worried that it's only down here that someone thought to get someone else to go take a look at Strategic Vision LLC's offices again instead of trying to work out if it's legit from the pixels and from having seen quite a few shops in their time.
I do remember Nate doing something like this during the '08 election, except there the group that raised Nate's ire contacted him and talked through what they were doing, and Nate posted that he was satisfied and that everything was hunky-dory. If Strategic Vision didn't want bloggers combing through everything they've done looking for answers, they could have just answered the questions Nate had.
@LAW
Let me see if I can put it in perspective.
Polling and statistical analysis depends on having accurate and verifiable data. 538 makes its living doing analysis of several polls. If those polls are influenced, biased, or outright lies then you don't really have a true representation of public opinion. If any conservative hack can hang up a shingle and be taken seriously then this becomes a big ethics issue.
The enthusiasm of the commentators of this site in our guerrilla detective work has evolved in order to reveal the dishonesty of this particular PR firm that purports to be legitimate is only one little piece of the mystery. We keep unveiling the little white lies and each of these lends itself to the bigger lie.
We don't have vindictive motives. Heck, most of us never heard of this company until the AAPOR brought it up last week. What we want is to see that polling is not unfairly manipulated. And it certainly appears to be that way in this case.
Nate's expertise speaks to us. When he thinks he has reason to believe that someone is rigging the game, we do what we can in support. Not because we have a vendetta but because we have purpose.
Pretty much what Davy said, except I'd change it to be any hack of any political persuasion. The fact that Johnson is a conservative is not of issue here - I myself have noted it once in my 30 or 40 comments in these threads, and fairly few people have picked up on it either.
I don't care if they have Republican affiliations or not. It's fine - I'll still trust Rasmussen's electoral polls highly, even though Scott Rasmussen is a Republican and open about his political leanings. I wouldn't really care if Johnson were a teabagger, so long as he were an ethical teabagger with regards to his polling. Which increasingly seems to not be the case.
...that is not to imply that Johnson is a teabagger (or, in more polite terms TEA-partier). I'm just saying if he were and he were a good pollster, it should have little to no impact on his results. And should be a moot point.
@Mike In Maryland - You do realize you sound just like a birther, right?
In general, my point is that SV is a private company and can do its business the way it wants. The fact that they don't divulge their methodology when a blogger demands it doesn't make them guilty of anything. Yes, SV may be a giant scam, but it may just be run by stubborn people. The latter is not a crime.
There has been no "proof" of anything on this site save for two major items: 1) the final digit analysis of polling numbers and 2) the firm says it's in Atlanta when in fact it is headquarted a couple hours away.
To address in order:
1) The analysis of SV's poll numbers that has been done on this site is not remotely conclusive of anything. The original assertion was that some arbitrary amount of deviation of final digits in polling signaled fraud. When readers correctly pointed out that this means nothing in the absence of a control group, Nate provided 1 single control case.
That's great and all, but what if the control case he picked just happened to be the best possible case of tailing digit consistency? What if 2 or 3 other pollsters match SV's deviation levels? Absent the answers to these kinds of questions, it's impossible to draw any real conclusions about the case he presented. I realize that doing this kind of analysis across dozens of poll takers is impossibly time consuming, but that maybe should have been considered before making the accusation.
2) There are surely many good reasons for a company to keep its "real" offices in the boonies while identifying with a larger city. Cheap rent, lower taxes, and cheaper phone/internet all come to mind off the top of my head.
Again, I'm not "defending" SV. I couldn't care less about Strategic Vision or what happens to it. What I'm trying to defend is the objectivity I've generally found on this site.
When it comes down to it, Nate was quick on the draw with the original accusation, as he didn't do a proper control group. Then, when pressed he provided one control case before realizing it would be far too time consuming to really do the proper work needed to prove his point statistically. At that point he was in too deep to give up, so he moved to the whole "fake location" thing. I'm just saying that at this point the facts are very thin and conjecture has been the dominating factor in this discussion. Not what I expect from 538.
The person who apparently thinks the screen name "LAW" will scare the bejesus out of us (no, it won't) is totally missing the point of ANY polling that is done reputably.
The AAPOR asked SV LLC, and almost two dozen other firm a series of questions. The reason those questions were asked was because AAPOR was trying to determine why the polls were so far off prior to the 2008 New Hampshire Democratic primaries.
he specific information requested from ALL polling companies was:
- who sponsored (for SV LLC) Strategic Vision’s survey(s) (the same was asked of ALL the other companies)
- a description of the sampling frame
- an account of how likely voters were identified and selected
- response rates
- a description of any weighting or estimating procedures used
In what way would any corporate secrets be exposed in any of those questions?
If all companies EXCEPT SV LLC responded to AAPORs satisfaction, not fearing the exposure of any corporate secrets, why did SV LLC ignore the requests of AAPOR to provide a response to those questions?
The ONLY question I can see that SV LLC might not want to answer on grounds of exposing corporate secrets is the one for who sponsored the poll, but even then, if they had responded, but asked the AAPOR to keep it confidential, would this issue even have hit the news? EXTREMELY doubtful.
And why the "we've put in a call to our attorney about a law suit" comment? A PR firm, on the up and up and honest, would NOT say something like that, as the attorney might just say, at a later date, "You have no basis for suing, and if you do, you might (or probably will) not only lose the specific case, but all your client base because of all the dirt that might be exposed."
SV LLC can clear all this up by making their past and future polls quite a bit more transparent. If Rasmussen has made the requested information transparent (and without any pressure), why can't SV LLC?
The longer SV LLC stalls in providing basic information about how they conduct their polling (a very basic and ethical position
ANY firm that conducts public polling takes), the more people will presume that they have something to hide, and the deeper they'll dig in an attempt to find out what and why.
Oh, and all the 'lack of objectivity' you've seen on this site, you must be describing the messages posted by those who normally are defending the deathers, birthers, birchers, 10thers, secessionists and/or the tea baggers (and they, as a group, have been noticeably silent on this issue). If THEY aren't trying to defend SC LLC in any way, shape or form, it really speaks volumes about the weak position that SV LLC is advancing.
So, as I asked before, is someone paying you to post messages in defense of SV LLC? If not, why are you defending them? And your protestations of "I'm not trying to defend SV LLC" are falling on deaf ears. Because SV LLC is conducting public polling, the public has a right to know if the company actually IS conducting polling in an honest manner. If SV LLC is conducting its polling honestly, why won't it let the public know how it conducts the polling? Or is it doing 'behind the green curtain' polling, just like the wizard in L. Frank Baum's book The Wonderful Wizard of Oz originally published in 1900?
Mike in Maryland
@LAW: I understand what you're trying to say, but...I think you protest too much.
1) You're assuming that Nate is out to get SV and thus has given up on statistically showing that the data is invalid.
But it was the commentors that first brought up the address discrepancy. In fact, Mark Blumenthal over at Pollster.com formally brought it up before Nate did.
2) You're assuming that identifying with Atlanta while being that far out is reasonable. Which...YMMV, but it's like someone in the Poconos or South Jersey saying they had a New York City office.
Plus, Rasmussen is based out of *drumroll* Asbury Park. ARG is from Manchester, NH. Neither of them make much effort to hide it (Ras's number is a local Central Jersey number; ARG actually states they're from Manchester on their home page). Granted, both towns are much larger than Blairsville, but I don't think anyone would have batted an eye if they'd came out and said they were from...you know, Blairsville.
3) Oh yeah, tracking back to point 1? The arguments about the Oklahoma HS poll weren't so much that Oklahoma students couldn't possibly be that dumb as a collective (though Nate did say something like that in the beginning of the post, and other people - myself included - noted that). Instead, they focused on the distribution, which seemed to correspond highly to a response rate being independent and a student body being homogeneous. That is, your knowledge of any question seemed to have no bearing on your knowledge of any other question, and all students had exactly the same chance of getting any given question right.
Quoting that post, those assumptions are completely unrealistic. It's like assuming - to use a New Jersey example - that a kid in Millburn (one of the state's top-ranked districts) is as equally likely to answer correctly as a kid in Belleville (which, I believe, is ranked in the middle of the NJ school pack) and as a kid in one of Newark's general public HS's (which are among the lowest-ranked in the state).
Or any student from any of those three districts, for that matter.
It is circumstantial, yes. I do think at least some commentors are jumping the gun in their rush to condemn SV as guilty. But things aren't looking really good for SV.
Oh yeah, and the minor fact that Nate's not the only one who's asked for crosstabs and not received them. This entire dalliance started because AAPOR asked for information and was stonewalled, then only provided with partial information. I believe an Atlanta Journal-Constitution reporter also asked for crosstabs on a SV poll, and never received him. This isn't just a one-person jihad here.
tl;dr - While one may debate the merits of this post and the ultimate relevance thereof, I think you're too quick in deciding that this is just a Strategic Vision witchhunt.
That said: Mike, I'm gonna have to stand up for LAW and point out the fact that he's a Kossack. ;) (Actually, he wrote a rather good rebuttal to Michael Moore's post last May re: GM. It's worth a read.)
Mike (The one posting about the jpeg compression): Just a quick correction. Jpegs use uniform block sizes across the image, though the luminance channel often uses a smaller block size than the other two. The luminance blocks are generally 16 by 16, and the size of the others generally ranges between 16 by 16 and 32 by 32 (16 by 32 and 32 by 16 aren't uncommon).
Each block is represented by a function, and the more complicated the function is, the more noise you generally see. The text you added, for example, has less noise than the original text, most likely as a result of not having a dark shadow behind it.
I'm no expert on such things, but I agree with you - the original image doesn't look suspicious to me.
I'm a retired Physician who moved from Atlanta to Jasper, Georgia [halfway between Atlanta and Blairsville]. Several things:
1. You refer to Blairsville as a "small city." change that to "small town."
2. Then there's this:
AUG 26, 2008 [Small is the new big] ... Fortysomethings Laura Ward and husband David Johnson downsized after living for nine years in a 3,000-square-foot, five-bedroom home in Marietta, Ga. They relocated to a 1,200-square-foot cabin with a large porch in Blairsville, Ga., two miles from their former home and eight miles from the North Carolina border.
“It was simply time for a change because the house represented too much space,” says Ward who works with Johnson in their public relations firm, Strategic Vision. The couple sold or donated furniture that wasn’t needed or didn’t fit the new, smaller scale of their home.
But they found that giving up provided a worthwhile return: not having to clean as much, being able to travel more because of their lower expenses and using all their spaces better. And despite their scaled back quarters – the couple do not have any children – they each still have a room to retreat to when they need a little time alone, Ward says.
[Blairsville is not 2 miles from Atlanta, it's 2 hours]
3. This part of Georgia has been Republican since there was such a thing to be - like with Abraham Lincoln. Appalachian Georgia opposed slavery and so it was both Republican and white. Still is.
4. We Democrats have to vote in Republican Primaries for local elections because that's all there is. In national elections, we vote Democrat [but it's more a "hobby" than a political force].
I've done work for perfectly sound companies that really looked pretty similar to this. They are usually pretty much one man companies, or husband-and-wife companies. They live from contract to contract, and often have gaps between contracts. When they have a contract, they balloon up, bringing in a network of other contract workers, most of whom don't work at the main office. Between contracts, the company can largely disappear, with the office often empty and calls being forwarded to the head's iPhone, as he travels around scratching for another contract.
Many of the companies like this that I've contracted for are headquartered in small towns where rents are low. Overhead must be kept low when income is unreliable. The ones I've worked for are solid and reliable companies with a lot of integrity. It all depends on the expertize and integrity of the head of the company.
None of those I worked for used false front addresses, but they weren't PR firms either. PR firms are supposed to make people look more solid than they are. It'd be weird if they didn't do the same for themselves.
Actually, in this context, the failure of SV to respond to AAPOR's requests makes more sense. If the AAPOR express mailed a request to the false front PO box, it may not have been forwarded, or it may have been delayed. If it arrived at the SV office, it could have hit at a time when the office was between contracts and had no staff, and when Johnson was out of town trying to win a new client. With no staff, response to a strange and complex request from a company the secretary/receptionist never heard of might well be slow.
The statistical questions about SV are possibly troubling, but this info about their address/location really only makes some of their other behavior make more sense. It doesn't contribute much to making them look questionable.
It does probably mean that their polls are done by contracting them out to telemarketers, sometimes piggybacking them on top of other polls that SV was contracted to do. The callers are probably not directly trained by SV, and SV probably does not have good control over them, so it's even possible that SV was being scammed by it's subcontractors, who could have faked some data instead of doing the calls.
Johnson said. “We’ve done the work, and we can prove that we’ve done it."
Then why hasn't he been forthcoming with said proof? Stalling for time to dummy up the more detailed data?
Data talks, bullshit walks.
It does not matter where they are located, it is the fidelity of their data and the quality of their analysis.
I think there work is 2nd rate, at best.
I saw the repulsive Frank Luntz on Fux last night with the Letters PH.D after his name. So, now he needs his creds bolstered for legitimacy. Did he get his degree from the same online joint as Orly Taitz?
@Jan Wolter
Well said.
@Jan Wolter
Your analysis is nonsensical. "With no staff, response to a strange and complex request from a company the secretary/receptionist never heard of might well be slow."
To say that a data request from AAPOR is too "complex" or "strange" for a company like SV that was able to pay for more than 150 polls over the past 4 years, is nonsense. The reason SV hasn't responded to AAPOR's request is because they can't. Not because of being short staffed, but because of a lack of factual data and any real methodology.
Re quoting Ben Smith:
"Another question is how the firm pays for its polls. Its website lists at least 172 public polls, and at a stated cost of $30,000 a poll, that's an expenditure of more than $5 million -- quite a sum for a small firm."
I thgink those of you criticizing Nate for his focus on SV's address are missing the point. Having buttressed his suspicions of SV by pointing out that all their addresses seemed to be mail stops, he felt obligated to finish the story by publishing his discovery tht they actually have a "real" address in a small town a couple of hours away. People can reach any conclusion they want from that, but responsible reporting required him to post that follow-up.
According to Wikipedia, Frank Lutz has an undergraduate degree in Political Science from the University of Pennsylvania, and a doictorate in politics from Oxford. If true, it's undoubtedly a more distinguished set of academic credentials than those possessed by most of his critics on this website.
For what it's worth, RedState.com doesn't like Nate anymore:
"Nate Silver becomes the Joe Morgan of Politics"
http://www.redstate.com/neil_stevens/2009/09/28/nate-silver-becomes-the-joe-morgan-of-politics/
The funny thing is that 2451 Cumberland Parkway isn't even "in" Atlanta, as they claim. It's in unincorporated Cobb County, Georgia, although it does have an Atlanta address. I live about 15 minutes from there, and am DAMN tempted to drive over there to see what's really at that address.
I know Red State is lower than Fox around here, but Nate has to answer that attack. He is accused to a total misuse of basic mathematical tools. He flays his targets for refusing to defend themselves, so he can't now ignore this. I'm a total amateur on these stats questions, but I found some of Red State's criticisms quite plausible.
I'm on the fence as to the substance here. I am curious, given the rage about Rasmussen, Zogby, etc., around here: are there polls that you readers think lean left? Is there ever a finger on that side of the scales? My own sense is that a lot of big polls (Newsweek, CBS/NYT, even CNN at times) play games trying to depress the number of Republicans in their samples. Nate has admitted this in the past, but somehow it doesn't seem to bother anyone. Daily Kos polls are supposedly neutral, but they always seem relatively rosey for the left. Do you folks disregard those polls, as you constantly seem to disregard Rasmussen?
This is the last time I ever post anything but statistics-related comments here.
I regret asking for (off-subject) expertise on pixels. We got a couple of nicely qualified responses, but paid for it in a larger volume of uninformed nonsense.
Jan Wolter posts some well-considered background and gets flamed out of context.
Yikes!
Jeff-
I'll waste two minutes of my time repeating that what is being questioned by Nate in this series of posts is whether SV LLC is actually conducting polls or making shit up.
All other discussions are peripheral and fun and distracting and entertaining and inconsequential and issue-fudging, the latter which is probably your intent. Bottom line: nobody who is being honest can deny that come election time, (Rasmussen+1D-1R) is a winning formula for predicting results. Or that Newsweek's polls are D-slanted. There are many agenda-driven pollsters. And then there are dishonest ones, as SV LLC seems to be. If Nate isn't going to weed them out, who will?
@ Jeff. What Just John just said.
He doesn't need to answer Red State. This isn't a political battle, despite your implications or theirs.
That is all.
Nate is in no more of a position of authority to demand answers from SV LLC than RedState.com, or one of its authors, is to demand answers from Nate.
Key difference. Nate actually has credibility.
Muley, you actually stayed out of this whole thing. Made the site much more pleasurable. Cat of veracity got your tongue?
@ Mule Rider
You are correct. SV LLC does not have to answer to Nate. They always have the choice. They have to judge what impact Nate can have on their business, and whether answering his charges is appropriate for their plans.
Nate must also make the same judgement about RedState's claims, and how it effects his credibility (which is, essentially, his blogging business's biggest asset)
As to him being "just" a blogger, he is, for better or worse, influential. No one is "just" anything, MR. You might say the same thing about Erin over at dressaday.com. She is "just" a blogger, but because of her knowledge and experience, and the exposure that brought to her other blog and work, she is one of the premier sources of insight into lexicography in the US.
The SV LLC contretemps may turn out to hurt Nate's credibility, but until it does, he is an oft-quoted source for media programs of all sorts.
To say nate is "just" a blogger would be to say you are "just" a guy who rides mules.
Semper Fi,
Terry
Sorry about that, guys. I shouldn't have given one of the resident trolls a reason to enter what was largely an intelligent discussion.
@ Davy. Seriously, we can do better than this.
I was going to post this comment at Red State but I don't want to sign up. So here is my response to thier article today:
Have you followed the math at all? Do you even understand how statistics works? Nate doesn't cry wolf. If he makes an accusation, it's based upon reasoned research. And he didn't really even make an accusation, He just suggested that there might be inconsistencies in Strategic Vision's methodology. All they need to do is provide the data or the methodology and the subject will be closed. How hard is that?
That redstate blog is wrong on so many levels, Nate would be beneath himself to respond. Disagree with his politics if you want, question his methodology if you can. But calling Nate a "mathematical illiterate"? Please.
Not only does the piece completely misunderstand Nate's analysis, it offers NO suggestion that Strategic Vision is legit. If you read to the end, you'll see that even the blogger is not willing to put his cred on the line for that firm. ("Is Strategic Vision making up poll results? I have no idea"). The blog is just an excuse to lay a hamhanded punch on 538. Better luck next time.
This isn't about moral high ground. It's about data. And until SV produces theirs, there is actual data on display here to suggest they are frauds.
I'd actually argue that Nate does have the moral high ground. First, he fully admits that none of this is by any stretch a definitive conclusion of fraud. It's suspicious and leads one to that possibility but proves nothing. He's also asked stats folks to look over his work and present alternate theories as to why the numbers come out as they do.
On the distribution, Nate actually states he'd expect something approaching true random but threw out the Benfords's Law as another remote possibility one could argue for. Redstate completely glosses over this distinction in Nate's analysis. Nate made it clear up front that there has, to this point, been no in depth study of such apsects of polls and thus there isn't yet a law or theorum he can plug into a computer to analyze the results. Thus he's coming up with his own tests, posting his methodology (imagine that!) and inviting peer review. That's how science works folks.
As to the further critique of his methods and choice of "arbitrary" comparison or the argument that this could be this way because of A, B or C reason, that's why Nate tried to make his control group as close a match for SV as he could. And yes, we all fully realize that there can indeed be a reasonable explanation. But to presume that the vast plurality of one set of polls triggered a specific grouping of numbers at the expense of another while a second group with similiar circumstances didn't is reason enough to let the whole line of questioning drop unanswered? It's possible sure. But no more or less possible than deliberate fraud or unintentional incompetence. Without getting more details on how SV has been conducting their polling and the after the fact data adjustment which all pollsters do, there is little way to prove any theory. However, releasing the data would make a determination much easier and if it is the first option, would easily vindicate SV and yes potentially harm Nate's credibility. I don't doubt that Nate took this into account and is fairly confident of which direction the end results will go, nor do I truly buy that he's on a vendetta for no reason and has suddenly dropped the math that he's always held so dear.
I was wondering, if it is true that Strategic Vision is "making their numbrs up", would it be possible for Nate or somebody else much smarter than me to reverse-engineer their algorithm? :-)
For example, I stumbled on this blog page:
http://politicalarithmetik.blogspot.com/2006/07/wi-gov-pollster-and-sample-still.html
This is about the 2006 Wisconsin governor's race. The "usual suspects", Zogby, Rasmussen and Strategic Vision all have the Republican challenger a lot higher than everyone else, and higher than the final results would seem to justify.
Looking at the graph (click on it to see a larger version), one could, with sufficient bad will, get the idea that Strategic Vision's numbers are based on the previous numbers from Zogby and Rasmussen. Rasmussen goes up, and two weeks later SV goes up; Zogby goes down and two weeks later SV goes down.
Coincidence? Probably. Major, major, major lack of input data? Yes, guilty as charged.
But again: If Strategic Vision was making it all up, I wonder how they did it, and if it could be determined?
@ Mule Rider
You're cute. Here's a lolly-pop. Now go play outside so that the adults can continue their discussion.
I have done some analysis which at first glance largely seems to vindicate Strategic Vision.
I assume the size of the third party/undecided/no opinion percentage was gamma distributed with parameters Alpha = 3, Beta = 2.
The size of the spread between the top two candidates/for-against/etc. was also gamma distributed, but with parameters Alpha = 1, Beta = 6.
An n=10,000 Monte Carlo simulation produces the following distribution:
0 1856 9.28%
1 1586 7.93%
2 1555 7.77%
3 1562 7.81%
4 1753 8.76%
5 2026 10.13%
6 2361 11.80%
7 2563 12.81%
8 2435 12.17%
9 2305 11.52%
@kinkorknight
@ Davy. Seriously, we can do better than this.
I know, I know. Hell I just got moderated from a previous post where I said something totally inappropriate regarding PK. I consider this a dubious distinction shared by a rare few; a category in which I do not wish to belong.
But I can't help myself. I feel like I'm defending the place. I know I should rise above it but whack-a-troll is a guilty pleasure for me.
I've tried ignoring troll posts. I've tried downloading the troll blocker program (can't get it to work). I just cannot stop responding to the provocation. I mean, somebody needs to stand up to these guys, right? Maybe that resistance makes a difference?
Think of it this way: rational discourse involves people who are willing to examine (but not abandon) their beliefs. People who are unwilling to debate their beliefs often resort to tactics of preying upon weaknesses in the opposing views character. I just think that if you show up here with a weak argument and try to deliver the smackdown, you ought to be prepared to get the smackdown yourself.
Is that wrong?
I thought I would never utter these words, but the guy at RedState has a well articulated counter argument that basically destroy's Nate's "case" here.
Feel free to dismiss it as just some conservative blog, but the RedState authors' points are objectively valid. Nate is simply playing it fast and loose with his assumptions, specifically with the assumption that a collection of polls would follow a certain distribution.
Jeff said...
"I'm on the fence as to the substance here. I am curious, given the rage about Rasmussen, Zogby, etc., around here: are there polls that you readers think lean left? Is there ever a finger on that side of the scales? My own sense is that a lot of big polls (Newsweek, CBS/NYT, even CNN at times) play games trying to depress the number of Republicans in their samples. Nate has admitted this in the past, but somehow it doesn't seem to bother anyone. Daily Kos polls are supposedly neutral, but they always seem relatively rosey for the left. Do you folks disregard those polls, as you constantly seem to disregard Rasmussen?"
Of course there are liberal/dem leaning polls just as there are rep/conservative ones. Rasmussen often gets poster ire here because some folks believe that there are times the questions lean towards push polls. I myself have little opinion on that topic as I've never studied their polls. What they do have is an easily identified republican slant in their methodology and this is easily accounted for. The slant comes from their party weighting and Likely Voter calculations which both tend to lean results towards republicans. So a reader that disagrees with their conclusions would adjust the poll a point or two in the other direction. When reading a poll I know has a corresponding democrat party slant, I again mentally correct or better just split the difference between them and Ras.
Nate actually rates Ras quite highly because while they may have a lean, they also have solid procedures, clear methodology and have released the data needed to confirm both. The issue with SV is not only that they have questionable results but also that they refuse to share the data that would help explain those results. They refused to aid AAPOR in their attempt to investigate the primaries and determine why things turned out so different than predicted. This wasn't a witchhunt on the part of AAPOR, but rather an attempt to determine (by asking for details from 2 dozen pollsters about how they reached the results they had) what happened so as to improve the science of polling. To refuse to send ANY basic details to such an organization as well as several other requesters (Pollster.com and an Atlanta paper both report having requested basic cross tabs from SV) begs the question of why. For SV to state they've never even heard of the American Association of Public Opinion Research and yet want us to believe they are a credible and well researched Public Relations and Polling company also seems off.
Sure, there are plausible excuses for all of this and none of it is a smoking gun. But it all together raises a question of how much weight if any we should give such an organization. They claim that had Nate simply asked for the data they'd have provided it to his satisfaction and that the facts will prove them to be right. Yet they've had requests for this data and it's the fact they completely ignored any and all such requests that had gotten Nate's attention in the first place. All Nate has done is raise a red flag, just as the IRS does routinely when looking over tax returns. The red flag doesn't mean you are crooked. It means the situation needs further review and ultimately that you'll be called upon to provide backup for your claims. SV got a red flag from AAPOR which Nate then looked at, ran some numbers, looked things over and then posted what he'd found. For all Nate new, the results would have vindicated SV, but they didn't. Instead they raised further questions.
I'm mystified by the outrage here at the suggestion that Nate answer Red State. No he doesn't "have" to, so what? He might "wish to" if he wants to convince people like me, who generally respect him but who found some of this critique plausible. If Red State is off its rocker, then Nate will presumably be able to show that, thus increasing his own credibility. This isn't just some political game, and Nate is the one throwing accusations around.
You all don't realize what a nest of political corruption exists in Blairsville. The big bank in this town is made up of the big money which propelled Zell Miller into office and beyond. These same people received the lion's share of "forgiven loans" to the tune of hundreds of millions paid from the Georgia Treasury to friends of Zells who live in Blairsville. Zell even changed the state banking laws to keep these guys out of the scrutiny which would have forced them to reveal the cash flowing in from the Cayman Islands and out to Switzerland. In other words, the state of Georgia laundered drug money for years through Blairsville and a certain bank. The same bank received $200 million in stimulus funds as the Bank President laghingly joked that "we didn't need the money but since everyone else was getting some, we applied."
Jeff said...
"I'm mystified by the outrage here at the suggestion that Nate answer Red State. No he doesn't "have" to, so what? He might "wish to" if he wants to convince people like me, who generally respect him but who found some of this critique plausible. If Red State is off its rocker, then Nate will presumably be able to show that, thus increasing his own credibility. This isn't just some political game, and Nate is the one throwing accusations around."
Eh, note that Nate isn't the one that has said he doesn't have to answer. From what I've seen of Nate, I rather suspect he will answer. Also if you read through Nate's posts in full you'll see that he welcomes critiques and alternate theories to refute his own. He doesn't take these attempts of his as gospel but rather as a start to attempt to see if his belief that the numbers should distribute close to random and if in turn that can help identify pollsters that are being less than honest in their published results through fraud or bad process.
Mule Rider said...
"Again, there are bodies that oversee this kind of thing - (AAPOR), et al - whose demands are legitimate. A stats geek in the blogosphere is not that authority. He can raise questions that may influence initial opinions on the matter, but he doesn't have the final say. That's what you all don't seem to get."
And AAPOR is exactly who asked SV for data and was ignored. Again, that's what started this whole issue. SV claims never to have heard of them and also that since they aren't a member they feel no need to answer to them or anyone else about their process. In today's day and age of polling, that is highly frowned upon and invites questions.
@ Mule Rider.
Wow, somebody's throwing a tantrum. That's not the way to get a lolly-pop, boy-o.
Nate -- stick to the numbers, not this amateur gumshoe stuff. It's perfectly normal for small business outfits to get PO Boxes in major metro areas where they conduct business. Hell, my friend runs a non-profit mostly from his home office in upstate NY, but for most fundraising he uses a PO Box in NYC, since that's where most of his donors are from.
You're gettin a little off-track here. Stick to the math and let the numbers do the talking -- that's when you're at your best.
Look forward to seeing your response to the RedState critique -- which seems to be not much more than a layman's "Now I'm no fancy, big-city mathematician..."
Mule Rider said...
Credibility doesn't insinuate authority.
Actually it DOES. All authority flows from credibility. You want to bring down a government in an insurrection/revolt? First step is to directly attack their credibility as a functioning government.
P.S. Which is why your lies and idiocy define you as the weak, pitiful man you are.
@Mule:
Your point about authority would have merit if Nate was the one originally calling for methodology (etc). But the very group you cited (AAPOR) made the call and apparently was not satisfied by any results they received. I don't recall seeing anywhere Nate personally said, "I call on them to give me their numbers and prove me wrong."
Yes, he has repeatedly mentioned their lack of transparency, but I don't see where he's trying to hold them accountable to him . Please show evidence otherwise if I'm wrong there.
Were it not for that, you do have a point - the burden of proof would be on him to prove their wrong-doing, and I hate to say it, but the "If only they would release their data this would be over" is the same argument the birthers use to perpetuate their idiocy. The underlying question is whether or not you believe Nate, the AAPOR, and others have raised sufficient question to justify supporting that POV. The proof will be in public opinion and in how SV responds.
I suspect they will eventually produce some form of information, and if there was indeed fraud, that information will be torn apart to expose the fraud and their company will rapidly fail. If there's no fraud, they may even get a bump out of this.
Rennie said...
For what it's worth, RedState.com doesn't like Nate anymore:
~~~~~~~~~~
Too funny and when did RS ever like Nate & 538 lol. Bottom line, 538 is getting wayyy more attention than RedState and they have their panties in a bunch.
and to all the concern winger Rep trolls worried about Nate and 538, feel free to relocate to RedState at any time, eh.
This is the frickin' internet people, catch a clue!
Okay, I've updated my analysis a bit using the trailing digit counts we've seen for Strategic Vision and Quinnipiac.
In both cases I use a Gamma distribution to predict the size of the 3rd Party/no-opinion/undecided vote.
I also use a Gamma distribution for the spread between the two leading options with alpha = 1 (because we expect the mode to be zero).
For Strategic Vision I am using the undecided parameters alpha = 3, beta = 2.5 and the spread parameters alpha = 1 beta = 7.
This produces:
Digit MteCrlo StratVis Delta
0 9.47% 10.14% 0.66%
1 8.58% 7.77% -0.81%
2 8.75% 8.51% -0.24%
3 8.51% 8.84% 0.32%
4 9.65% 9.49% -0.17%
5 10.25% 10.80% 0.55%
6 11.41% 9.61% -1.79%
7 11.27% 11.53% 0.25%
8 11.76% 12.19% 0.43%
9 10.32% 11.11% 0.79%
For Quinnipiac I use alpha = 3.5, beta = 5 in the undecided distribution and alpha = 1, beta = 6 in the spread distribution. This produces:
Digit MteCrlo Qnnpc Delta
0 9.48% 9.86% 0.39%
1 10.08% 9.95% -0.12%
2 10.30% 10.98% 0.69%
3 10.46% 10.71% 0.25%
4 10.51% 10.37% -0.14%
5 10.33% 10.17% -0.16%
6 10.13% 9.95% -0.17%
7 10.11% 9.20% -0.91%
8 9.35% 9.67% 0.31%
9 9.25% 9.12% -0.13%
These assumptions make the observed data much much much more believable with the exception of the fact that Strategic Vision's numbers clearly have an unnaturally occuring preference for 5's over 6's and 0's over 1's.
In my opinion, this is more likely to be the result of idiotic rounding than fraud.
Humans probably do tend to inadvertantly round down toward 5's and 0's though, so maybe we can't blame Strategic Vision too badly for being human.
@ Bronxx
the burden of proof would be on him to prove their wrong-doing, and I hate to say it, but the "If only they would release their data this would be over" is the same argument the birthers use to perpetuate their idiocy.
Okay, I take exception to this (particularly since you're quoting me).
The argument that 'if only they would release data' is anything like the birther movement is ludicrous. Did John McCain release his birth certificate? Oh, wait. He wasn't born in America.
Every other pollster was transparent. They showed their work and validated their findings. If you're going to purport to represent public opinion with scientific polling, you should be required to show how you came up with your results. Otherwise, YOU LIE!
@ Davy,
Chill. Wasn't meaning to call you out individually, others have made the point, too. If I actually used the same words you did in my quotes, it wasn't intentional.
My point was about the burden of proof and trying to shift it to the person being accused, rather than on the accuser. Personally, I think Nate has made his argument and that if SV wants to remain a legitimate organization (benefit of the doubt to-date), they should release that info. But as Nate himself has repeatedly said, this is not "proof".
Assuming they're 100% innocent of wrong-doing and this is a weird statistical anomaly, they may feel they have no reason to release the info and even some reason to keep methodology a secret. (I would agree such reasoning is probably wrong, but then again, I'm not them.) Assuming President Obama is 100% a citizen of the U.S. having been born in Hawaii, he no doubt has a good reason for not responding to birthers - possibly to avoid lowering himself to the standards of complete imbeciles.
I was pointing out the similarity of the argument/verbage being used and that it relies completely on whether or not you feel your side has made its point. Someone who doesn't feel the same would be justifiably completely unpersuaded by the "release the info" argument. They may be an idiot (i.e. birthers), but since they disagree on the underlying argument, they're justified in at least rejecting a conclusion based on it.
And just to be clear, let me say again: As far as I'm concerned, Nate has made his point and has circumstancial evidence and well-reasoned logic on his side, but no true "proof" without that source data.
But Hank Aaron is not an authority to demand that such a change be made. There is a governing body for MLB to handle that.
Actually Hank Aaron would have a hell of a lot of influence (AKA 'authority') in the public eye and baseball circles in such a hypothetical situation. Woe to the MLB commisioner that ignored and didn't address his comments. Far more the case than say comments (to Frank Aaron, or the MLB for that matter) from random-guy-off-the-street-whose-baseball-background-having-watched-watched-Bad-News-Bears-II that the RedState author represents represents. :)
Put a sock in it, Dwight, you bitter asshole.
LOL, thanks for that. I do love the smell of irony in the morning (and afternoon, and evening, and at night....). :)
You're still too cowardly to meet me, so that tells me all I need to know.
As I told you time and time before, cowardice has nothing to do with it. Stop lying. Stop posting stupid shit. Stop unhinging. Stop purging from your memory any detail that is in the slightest inconvient to your preformed views. Get hold your posts to at least a small measure of integrity, rather than using some childhood "well, Billy lies and says stupid stuff".
In short remake yourself into someone worth my time and effort to meet and I'll be there in Savanah, or perhaps further NE, in a year or less. (Not sure if Indianapolis is closer for you? I might end up there for a weekend convention in Aug. next year.)
Nate -
I think your witch hunt against Strategic Vision is perhaps a bit off the mark.
Your main line of evidence is that the trailing numbers skew in a non-random direction.
This is not particularly compelling because of the following:
1) Pollsters generally survey close races more often
2) This means that the expected result will be somewhere closer to 50-50 than the average electoral race.
3) This means that you would, absent undecideds, expect to see a U-shaped distribution.
Example: 51-49 produces a 1 and 9.
You would expect a high and low more often than 2 middle values.
4) Pollsters vary GREATLY in their effort to classify undecided voters.
- Some word the question loosely to encourage soft commitment
- Some as a "leaner" follow-up (Rasmussen)
It is the strength of effort to classify undecideds and the method in doing so that will cause a non-random skew in the data.
Your singling out of Quinnipiac Polls felt very much like cherry-picking.
You should have run the analysis for all major pollsters and showed us the distribution for each.
Why did you not do that?
Because some have much greater skews that Strategic Vision.
Here is my experiment.
I took all presidential approval polls from George W. Bush as archived by the Roper Center.
http://webapps.ropercenter.uconn.edu/CFIDE/roper/presidential/webroot/presidential_rating_detail.cfm?allRate=True&presidentName=Bush
This produced 2,894 trailing digits.
What is good about this is that it is that pollsters are measuring:
- The same thing
- In the same geography
- Under similar conditions
- Over the same time period
The result?
The SKEW differed wildly:
FIRM N %0-4 %5-9 Spread
Fox/OpDyn 282 49 51 1
Gallup 226 46 54 7
Gallup/CNN/USA 218 42 58 17
Pew 200 50 51 1
Newsweek 186 46 54 8
ABC/WP 172 52 48 5
CBS 162 49 51 2
Democracy Corp 154 52 48 4
ARG 138 38 62 25
NBC/WSJ 138 41 59 17
CBS/NYT 132 58 42 15
All 2,894 48 52 3
Even among these firms measuring...
- The same thing
- In the same geography
- Under similar conditions
- Over the same time period
We saw a spread on %0-4 vs. %5-9 as high as
ARG: 25 points
NBC/WSJ: 17 points
Gallup/CNN: 17 points
CBS/NYT: 15 points
Strategic Vision's spread was 10 points, and this was...
- Different candidates
- Different geographies
- Different time periods
YOU WOULD EXPECT MORE VARIATION.
Your methodology and threshold indict about 1/3 of pollsters as frauds.
Um, the UPS store you link to is Suite 300, not 3607 as Strategic Vision lists. Further the UPS additional info is "VININGS CROSSING SHOPPING CENTER AT PACES FERRY ROAD AND I-285". Obviously it's some big strip mall. There is a UPS store in the same strip mall as the alleged SV headquarters.
That was real easy to figure out too. Surprised you missed it.
With regard to the Papa's Pizza question, this site has several photos which give a clearer image of the storefronts surrounding the pizza place, which are quite clearly the same as in the Facebook image. Sadly, none of them appear to have captured a Strategic Vision storefront.
http://www.downtownblairsville.com/night.html
@Mule Rider
"Owes" is irrelevant. SV will make some kind of disclosure if they feel it's in their interest to answer Nate. Same goes for Nate with regard to RS. But these are not analogous situations.
RS is unconvinced by an analysis that Nate has already published. Nate could respond further but in any case has already shown the work on his side. The relevant info is in the open; anyone who wants to can read the original analysis and the RS critique and judge for themselves whether Nate's analysis stands up to scrutiny.
SV, on the other hand, is facing questions on whether it did any polling at all, and if so how it was done. The relevant information needed to answer those questions is kept secret by SV. Not showing their work is their prerogative, but it precludes anyone from making a fully informed judgment about how their work was done.
I imagine Nate will respond to RedState as soon as he's back home, though he'll do it here where more people will read it. :)
Further, it seems SV is having the same trouble Iran had with their election. It takes time to go back and fake all the numbers to justify a fraudulent poll. It's not as easy as just faxing over the pages, you have to make them first and do it in a way that's less obviously fabricated.
Notice how much time SV has had to create and back-date the data? Still isn't ready, I guess.
Nobody is saying his opinion would be worthless (as is the case with Silver and SV LLC), but I think you are overstating the importance of his opinion and the weight it could carry
I'm actually working from a similar real situation with relatively recent NHL rule changes.
But really you are getting lost in your analogy.
In the end:
1) all authority flows credibility
2) Mr. Johnson has already bowed to Nate's public credibility by using his name and tossing a platitude of acknowledging the fairly wide perception of Nate as a somewhat authorative voice on the subject (I bet that just burned your biscuts to read that, huh? :D )
3) the RedState post clearly didn't actually read/understand/want to understand Nate's posts, although admittedly there are also poster here that don't and are calling that work "proof" :(
4) you've shown time and time again that you wouldn't notice a logical critique if it snuck up behind you and cornholed you :)
P.S. I'd actually like to see an "answer" from Nate that further fleshed out the first rough pass over SV's data.
Huh, I never yet stayed in a motel with this policy (from the Seasons Inn website):
In observance of the Lord's Day and to give everyone who works here a chance to go to Church and spend time with their families, Seasons Inn is not open on Sundays. All stay over guests receive extra linens on Saturday for use the next day.
It was to help illustrate the point, and even a simpleton could see the parallels.
Congratulations, you just slotted yourself lower than a simpleton.
You simply don't understand in a deep sense what authority is, where it comes from, and where it exists.
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,
электронная почта без регистрации
Post a Comment