The poll in question is one of Virginia, released today by a company called Gravis Marketing. It has Romney at 49, Obama at 44. This is a weird result because the nine previous Virginia polls (all in July and August) had Romney up only once, by three, a Purple Strategies poll. In fact, this poll by Gravis is the best poll for Romney in Virginia since a WeAskAmerica poll had him up 5 back in June.
Boy there sure are a lot of new polling companies this year. AngryBear thinks it may not be a coincidence that a lot of them seem to lean right. I don't know, but I'd be interested in your opinions. His implication is that you could move poll averages by having several outfits doing this.
I have found some things out.
The odd VA poll was conducted Sept. 8 and 9. This is an unusual time to find a Romney surge in VA. That was Saturday and Sunday, polling starting less than 2 days after President Obama's speech and 3 after Clinton's. If anything, an Obama+5 number would be less surprising.
You could say "Well, there's a big margin of error for smaller state polls." Problem is, this was no small poll. They report more than 2,000 respondents. That's a big state poll! The margin of error is only 2.2.
Supposedly this is an LV poll. But check out their question wording. They asked people if they were "Very likely, somewhat likely, Not likely" to vote. What did they do with the "somewhat likely" people? Do they get thrown out?
Other Gravis polls have been weird. The last five polls in Florida include only two with Romney leads - both Gravis. This DU post includes a "caution" about the reliability of Gravis despite one of the polls showing Obama up 4 in Ohio. Obama plus 4 in Ohio seem right to you? Me either. Here's Nate Silver on that result:
Gravis has had substantial Republican lean in the other polls it has released this year, making these results look stronger for Mr. Obama by comparison.I think it's time we did some research on Gravis Marketing and see what we find. Can you help?
Below I have more.
As you know, some polling firms turn out to be somewhere between questionable and entirely fictitious. Kos went to court after publishing suspicious results from Research 2000 (that outfit had a left lean which may have been manufactured rather than accidental). Likewise, Nate Silver discovered problems with polls issued by Strategic Vision (a firm that had a right lean in polls).
Now, I don't anything know anything nefarious about Gravis but these previous cases should make us wary. Maybe they are legit but have a biased methodology which gives a strong consistent Republican lean - something many people say about Rasmussen.
Here is what I have found out:
- According to WhoIs.net, their website was registered in 2010 and will expire in 2013. Provider name: Bluehost.com out of Provo, Utah.
- Their website says they are based in Winter Springs, Florida.
- Their website lists no employees by full name. Under the testimonials page you find references to someone named Gravis and someone named Doug. Apparently they are awesome. But not so awesome that they wanted to put their first names on their site. We'll come back those testimonials in a minute.
- This site says they have between 2 and 4 employees. I'm going to go out on a limb and guess that two of them are named "Gravis" and "Doug." The company is reported to be 3 years old even though the website is only 2.
- The same site lists the President as Doug Kaplan and says he's been president since 2011. I guess somebody else (Gravis?) was President back when they were founded.
Have you heard of this firm? Can you find anything out about them? Who is Doug Kaplan? What's his background? Can you find a last name for "Gravis"?