Pollster.com

Articles and Analysis

 

Bialik Asseses ARG, Teases Quinnipiac

Topics: 2008

Carl Bialik has a nice assessment of the ARG poll in Pennsylvania that includes this tease of the new Quinnipiac numbers due out tomorrow:

Other pollsters’ numbers disagree with ARG’s. Clay Richards, who runs the Quinnipiac University Polling Institute’s Pennsylvania poll, said he doesn’t expect his poll that will be published Tuesday to show much difference from the last one, which had a Clinton lead of six points. “I don’t see that much movement in Pennsylvania myself,” Mr. Richards said by phone from Harrisburg on Monday. He declined to comment specifically on his rival’s contradictory numbers.

 

Comments
FlyOnTneWall:

Mark-

Any response to my earlier comments (on the ARG thread) regarding the possibility of response bias introduced by their use of omnibus surveying? There's gotta be some reason for ARG's inconsistency. What makes them weird, among polling outfits, is that they don't appear to have a house effect. It's just that their results are all over the map. What say you?

____________________

kingsbridge77:

ARG is tied with Rasmussen when it comes to pollster accuracy during these primaries, although they both rank among the lowest:

ARG is tied with Rasmussen when it comes to pollster accuracy during these primaries, although they both rank among the lowest. Why does no one then trash Rasmussen as hard as they do with ARG?:

http://www.surveyusa.com/wp-content/uploads/2008/02/hi-level-median-error-through-022008.JPG

____________________

Fourth:

Probably because ARG has a unique ability to swing wildly from poll to poll. This led them to have South Carolina in the margin of error, be all over the place on Wisconsin, and now show not one but two huge swings in different directions in PA. The final median error might be the same as Rasmussen, but the difference is that ARG is not internally consistent which makes it pretty useless for looking at trendlines leading up to an election (which is pretty much the purpose of this site).

____________________

FlyOnTneWall:

I'd second what fourth has to say.

There are many, many problems with using SurveyUSA's tables as anything more than a rough gauge of pollster accuracy. But the biggest is that when other polling firms have been wide of the mark in a particular state, they tend to err consistently in the same direction. That implies a "house effect": a feature of their model that's producing systemic error. ARG, by contrast, is all over the map. That suggests something else is at play.

In the other thread, I made the case that we may be seeing what happens when a pollster puts omnibus polls in the field, and doesn't ask the political questions first. I don't know that's what's going on, for sure, but it would explain the wide variation.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR