Obama versus McCain (versus the rest)

[RainMaker]

Yeah, I don't consider him just another blog. He's a really smart numbers guy and has a solid track record behind it.

Plus he's become pretty succesful. Has a book deal and so forth. I can't fathom he'd risk a defamation suit over something he wasn't certain about. He's not one to throw around wild accusations for attention.

[JonInMiddleGA]

Originally Posted by flere-imsaho

I don't count myself amongst the "heavy duty math/stats contingent" here, but in general I think Nate Silver's grasp of statistics is assumed on the bases that a) he developed PECOTA and b) his statistical model correctly predicted the presidential election result in every state but Indiana and correctly predicted the result of every race for the U.S. Senate.

Alls I'm wondering is whether the seemingly reasonable questions about the possible limitations to the method he used to analyze the data are, well, as reasonable as they appeared to be. Like I said, somewhere along the way the questions went beyond the scope of my knowledge so I'm not sure if they're solid questions or just looked like solid questions if you know what I mean.

I haven't had a chance to see whether he eventually answered those questions or not, but while I'm fairly comfortable that he knows his stuff well enough to have dealt with the issues raised about his own analysis that's not a guarantee that he did actually consider them, one of those "only human" kind of things.

[JPhillips]

Originally Posted by JonInMiddleGA

Alls I'm wondering is whether the seemingly reasonable questions about the possible limitations to the method he used to analyze the data are, well, as reasonable as they appeared to be. Like I said, somewhere along the way the questions went beyond the scope of my knowledge so I'm not sure if they're solid questions or just looked like solid questions if you know what I mean.

I haven't had a chance to see whether he eventually answered those questions or not, but while I'm fairly comfortable that he knows his stuff well enough to have dealt with the issues raised about his own analysis that's not a guarantee that he did actually consider them, one of those "only human" kind of things.

You also have to keep in mind that there is a lot of evidence that SV LLC is being dishonest. Added to the statistical analysis done by 538 is the fact that SV LLC refuses to show anything other than the final numbers of their polling, the fact that of the three addresses they list as offices all three are UPS stores, and the fact that their physical location seems to be in a hotel plaza in a small town two hours north of Atlanta.

Of all the issues , the most damning is their refusal to release any poll internals. If they had actually taken polls it should be pretty easy to release a section of a past political poll to prove they are at least actually calling people.

[JonInMiddleGA]

Originally Posted by JPhillips

You also have to keep in mind that there is a lot of evidence that SV LLC is being dishonest.

That's fair enough based on what I can see from a few quick reads but it doesn't make the 538 analysis necessarily without flaw either. That's the part I was really wondering about since at some point the questions about their (538's) methods started going over my head. Since those questions showed up over there a bit before the comments started heading into fully partisan territory on both sides, and they at least sounded reasonable to me, I was as interested as anything in whether they turned out to be on the mark or not from the standpoint of whether "sounding right" or plausible turned out to be right or even plausible. Sort of a "could I tell the difference" thing just for me personally.

Added to the statistical analysis done by 538 is the fact that SV LLC refuses to show anything other than the final numbers of their polling, the fact that of the three addresses they list as offices all three are UPS stores, and the fact that their physical location seems to be in a hotel plaza in a small town two hours north of Atlanta.

I'm going to take a leap of faith here & guess that you know that's not all unheard of as a business tactic even for perfectly legitimate businesses, right? And especially for businesses physically located outside major metro areas whose clientele is largely from/tuned to major markets.

We never did it because we eventually decided it was better to turn our location into a positive instead of trying to hide it as a negative but it's something I run into probably 2-3 times a year at least. Most recently was another ad agency that we briefly considered merging with, who prominently claimed locations in Atlanta, New York, and Washington with another opening soon in Dallas. The DC location was actually a small apartment one of the guys maintained up there for personal use, the New York location was somebody's parents home, and the new Dallas location was the basement office at the home of a freelancer they were hiring largely to claim the Dallas location & he was happy to do it since he could then claim connections to three other cities. Their (the ad agency that is) business is completely legit, it just happens to be in an industry where looking bigger than you really are is a significant advantage if you're good enough at faking it not to get caught.

None of which makes 538 completely legit, completely crooked, or anything in between. Just saying that the address thing isn't exactly unique nor even indicative of substantial fraud either.

[JPhillips]

I don't think any single issue is that big of a deal, but the volume of problems is pretty damning IMO, especially since they won't show any of their work. One poll's internals would put all the speculation to rest.

[DaddyTorgo]

Originally Posted by Mizzou B-ball fan

If you seriously think that, under the best of circumstances, 52% of all voters in PA in November will be Democrats, I've got some fine oceanfront property to sell here in Missouri to you at a deeply discounted price.

Just wanted to bring this gem back to life

United States presidential election in Pennsylvania, 2008 - Wikipedia, the free encyclopedia

PA went 54.47% Obama and 44.15% McCain.

[Kodos]

Huh. Usually his predictions are spot on. Are you sure history isn't wrong?