Who Are Gravis Marketing and What Do They Know

[Foolmewunz]

I noticed a few weeks ago that Real Clear Politics is now picking up data from Gravis Marketing. This is simply a name I'd never encountered before, but what really caught my eye is that they're always right out there with Rasmussen in having the GOP doing considerably better than what other polling organizations show.

My cynical nature is telling me that RCP have an interest in a close election/close polls. The more it looks like it's up for grabs, the more people click to check at RCP every day, after all.

All I can find on Gravis, prior to this, is that they're a robocall marketing firm. Sounds a little like Zogby - e.g. you give us the answer you want and we'll write you a properly worded poll to get that result.

Does anyone know anything about them? Any comments on my suspicions that RCP need another out-lyer (how do you spell outlyer?) because Rasmussen alone wasn't enough to make Florida and Ohio and Wisconsin look close?


ETA: Or do they just have better information and methods than everyone else and we're all deluded because Mitt's winning everywhere?
And did anyone see O'Reilly yesterday talking about current polls. He went through six or seven races/questions, using only Rasmussen numbers and said, "....because they were the most accurate last time out...". That gave me a chuckle considering the fact that they've been proven to have been one of the worst, week-on-week, and only recovered their accuracy by getting the day before the election correct.

 

[Puppycow]

Don't know about their methods. They don't really have a track record that I know of.

When Nate Silver calculated the "house effect" of various polling firms back in June, Gravis wasn't on the list.

I wouldn't jump to conclusions though, just because some of their recent polls seem to be more favorable to Romney than other firms. All polls have a margin of error, so you really need a lot of them before you can infer a house effect. I see that on Nate Silver's page, Gravis polls figure in the calculations for Virginia and Ohio. The one in Virginia seems to be favorable to Romney, but the one in Ohio shows a 4-point lead for Obama.

So I'll remain agnostic on them, but I think it's reasonable to put less weight on polls from a firm that doesn't have a track record than from firms with longer track records (assuming they have good track records).

 

[Foolmewunz]

Well, you're no fun! I'm looking for diabolical conspiracies.

Seriously - it's not "some" of their polls, though. It's all of them. And I really just like to know how much to "weight" a given organization. E.G. when you see Romney +2 - the latest Rasmussen poll - you know why the GOP is in disarray... because everyone else has Obama +2 to +5, and historically, we know that Rasmussen blatantly weights their polls to the GOP. Even if it's a day when the only major poll that shows up in the morning is Rasmussen, you just subtract 5 to 7 points and get an accurate picture.

For Garvis, I don't have any rule of thumb, yet. Are they slanted? Are they merely incompetent and trying to make a name but with faulty methodology, or are they the new Nate Silver? I figured or hoped some of our election hawks might have some insight on them.

 

[ZirconBlue]

I don't know anything about Gravis, so I'll just tackle the most difficult and important of your questions:

how do you spell outlyer?

"Outlier."

 

[jnelso99]

A company called Gravis made really cool joysticks in the 90s. Still have a Firebird stick in my basement. It was awesome for "Wing Commander". Might have to get a serial card to see if it still works...

 

[JoeTheJuggler]

All I can find on Gravis, prior to this, is that they're a robocall marketing firm. Sounds a little like Zogby - e.g. you give us the answer you want and we'll write you a properly worded poll to get that result.

It sounds like a strike against them that they're also selling political robocall services (and "a wide array of campaign services"). Hard to claim they're disinterested if some candidates are their clients.

http://gravismarketing.com/political-robocall-services.html

 

[Corsair 115]

A company called Gravis made really cool joysticks in the 90s.

That was my first association with the company name as well. I owned a Gravis joystick or two. Quality products, as I recall.

 

[Rashmi]

That was my first association with the company name as well. I owned a Gravis joystick or two. Quality products, as I recall.

I once had a boyfriend, his name was, "Gravis", he had a nice schweet joytick...I miss him more than my first fish tank.

 

[Kaosium]

Regarding Gravis, never heard of them. Perhaps they're new on the scene? At least that's the impression I got from their website.

Well, you're no fun! I'm looking for diabolical conspiracies.

Seriously - it's not "some" of their polls, though. It's all of them. And I really just like to know how much to "weight" a given organization. E.G. when you see Romney +2 - the latest Rasmussen poll - you know why the GOP is in disarray... because everyone else has Obama +2 to +5, and historically, we know that Rasmussen blatantly weights their polls to the GOP. Even if it's a day when the only major poll that shows up in the morning is Rasmussen, you just subtract 5 to 7 points and get an accurate picture.

Hrm.

That's an interesting thing to say, because Nate Silver himself rates Rasmussen one of the least biased polling outfits. You can see that here which is part of this page where he also details their first year failure. Another thread here was dedicated to another mis-projection, the '10 midterms, it's like some people keep track of their every failure! I'm guessing amongst a certain crowd Rassmussen being inaccurate is pleasing news? However I noticed they polled during the Wisconsin recall and they pretty much nailed that one, thus perhaps they fixed whatever methodology problems they might have had in the '10 midterms, and of course to get that high a rating from the sainted Nate Silver they musta been pretty damn good the rest of the time, eh?

For Garvis, I don't have any rule of thumb, yet. Are they slanted? Are they merely incompetent and trying to make a name but with faulty methodology, or are they the new Nate Silver? I figured or hoped some of our election hawks might have some insight on them.

I got to wondering who in the hell is Nate Silver anyway? I'd never heard of him, I was wondering why so many genuflected before the awesome might of him! I guess he did well in the '08 election, huh? That's where his reputation came from, analyzing polls in an election where the GOP had an incumbent leaving office with approval ratings ~25%, running a candidate who'd bared his fangs to the base over and over again for the past dozen or so years, and with the economy in crapper to boot. How anyone thought McCain ever had a chance in that election is beyond me. Then I guess this Nate Silver guy did this deed being fed internal Obama polls, the ones not for public consumption, the ones that tell it how it really is?

Color me less impressed than some by that 'feat.'

At any rate, you may note that Rasmussen had the highest 'pollster rating' amongst any of the ones doing regular nationwide presidential polling this cycle on Nate Silver's list, and it's pretty easy to see where the others get their bias from: oversampling Dem voters for whatever reason. As you can see here since the Gipper's blowout in '84 the number of people self-identifying as Dems over GOP at the polls has been between 0-4 percentage points--with the exception of '08, which was rather an historic occasion in many respects, some detailed above, others due to Obama's qualities as a candidate and his unique bid as a major party candidate.

Now as you can see on that chart, the Dem percentage didn't actually soar, it stayed right about where it had always been, it was the GOP number that nose-dived back to post-Watergate levels. Being as since then there was a historic GOP midterm, and it's not '08 anymore, I'd think it unlikely that the Dems would maintain that +7 percentage point advantage on election day. However if you look at this chart on the next page of the article, it appears the methodology of most of those other polls that RCP uses seems to be aiming for that +7 or so Dem advantage. That article (of which I'm only referring to the data) was two months ago, so I checked the most recent polls on the RCP site and they're all doing the same thing still! One of them has a double-digit Dem advantage built in, I'm guessing you can figure out which one just by looking at the numbers. I checked the state (Fox News) poll for Ohio and found the same +6 Dem percentage point skew.

Now it could be that repeatedly pollsters are finding that self-identification by Democrats is higher than it's been since Reagan's first election and that's going to be an accurate predictor of the turnout as well at the polls, but it may also be that most of those polls don't properly identify the ones likeliest to vote, that being actually more difficult than just asking them because too many will say what they're 'supposed to.' Sometimes it's best done with kinda 'trick questions' by seeing if they know where their polling place was, who they voted for last time etc. It could also be that, due to the curious attraction to Dem+6, that most of those polling outfits have decided to try to skew their polling as a predictor of turnout roughly equal to that of the last Obama election, minus a point for age or something.

So there's your diabolical conspiracy! I hope I chilled you to the bone! p) What if the Rasmussen poll is the most accurate and the others are simply due to pollsters making similar methodology errors and Nate Silver isn't catching them because he's not actually a wunderkind, he just knows how to make cool charts and spit out the answers he's been fed like he was Charles Van Doren or something.

Oh, and here's some new data along those lines to ponder....should you be afraid...very afraid...or am I jus' messing' wit ya?