September 30, 2007

Debunking ARG Polls

Note: All poll results taken from Real Clear Politics. GRAPH WARNING

Obviously, it took a poll like the one in South Carolina to cause me to do more than laugh off the results of the latest ARG Poll. So, I decided to do some research on ARG’s numbers in comparison to other polls taken at the same time.

Notice anything strange? Here are the poll results from four seperate polls taken between September 9 and September 19 in Florida:

    Giuliani

  • Rasmussen 29
  • Mason Dixon 24
  • Insider Advantage 24
  • ARG 26
  • Thompson

  • Rasmussen 23
  • Mason Dixon 23
  • Insider Advantage 23
  • ARG 16

    McCain

  • Rasmussen 12
  • Mason Dixon 9
  • Insider Advantage 11
  • ARG 18
  • Romney

  • Rasmussen 11
  • Mason Dixon 13
  • Insider Advantage 12
  • ARG 14

Now, even stranger, notice McCain’s numbers in New Hampshire compared to others taken in June.

    Thompson

  • ARG 3
  • Mason Dixon 12
  • Franklin Pierce 9
  • CNN/WMUR 12
  • Giuliani

  • ARG 21
  • Mason Dixon 16
  • Franklin Pierce 18
  • CNN/WMUR 20
  • McCain

  • ARG 30
  • Mason Dixon 15
  • Franklin Pierce 17
  • CNN/WMUR 20
  • Romney

  • ARG 23
  • Mason Dixon 27
  • Franklin Pierce 27
  • CNN/WMUR 28

There are many more examples of this, that I will post once I can get my excel charts working.

Let’s look at South Carolina by itself in May, comparing two polls, ARG and Insider Advantage:

    Thompson

  • ARG 13
  • Insider Adv 13
  • Giuliani

  • ARG 23
  • Insider Adv 18
  • McCain

  • ARG 32
  • Insider Adv 17
  • Romney

  • ARG 10
  • Insider Adv 8

Now, let’s compare June’s Mason Dixon poll and ARG poll in the state of South Carolina. One must remember that Mason Dixon is considered one of the best when measuring state support:

    Thompson

  • ARG 19
  • Mason Dixon 25
  • Giuliani

  • ARG 22
  • Mason Dixon 21
  • McCain

  • ARG 23
  • Mason Dixon 7
  • Romney

  • ARG 8
  • Mason Dixon 11

I’m starting to see a pattern. How about you? There are many more, and this post could go on all night long.

by @ 8:00 pm. Filed under Poll Watch
Trackback URL for this post:
http://race42012.com/2007/09/30/debunking-arg-polls/trackback/

17 Responses to “Debunking ARG Polls”

  1. Arnie Says:

    think they like Mitt????
    lol

  2. Jeff Fuller Says:

    How did that analysis “debunk” ARG’s results? Are you alleging that they are doing dishonest polling or that they are just off?

    The tone of your post is that of a “conspiracy theory” that ARG is out to get Thompson and wants Romney or McCain elected. Is that what you are alleging?

    I’m not defending ARG’s numbers. Hey, if Romney is really leading in SC (as well as still in IA and NH) then everyone else may as well pack it up. I’d be fine with that and would love these results to be valid . . . but I think it’s hard to argue with the “TRENDS” theory of polling. Even if a polling firm is off in it’s results . . . as long as they employ the same techniques from month to month then we can conclude that the trends they detect are real. Romney’s finally made a push in SC and it’s showing. I’ll buy that trend for sure (as it has been in other places he’s concentrated.

    Buried near the end of the last post I commented:

    Romney did some hard and heavy radio ads in SC starting on Sept 24th highlighting what he did to Khatami (and by extension how he would have responded to Akdimenijab–Ground Zero–UN–Columbia debacle) http://www.mittromney.com/News/Press-Releases/Khatami . . .

    I think he’s been hitting the TV airwaves there recently too (though still not as much as IA and NH)

    Still, I highly doubt that’s enough to give him this much of a swing. DeMint onboard helps and the Mitt Mobile (Romney’s sons) has been working SC too. Still, I don’t expect Mitt to lead in SC and I doubt he’ll win it. If he does, it’s “Game Over” for everyone else.

    Romney does have the best “first impression” of all the cadidates and I’ve found a lot of people whose only exposure to him has been the overtly negative and unfair MSM/DNC reporting and have been big-time impressed when they first heard/saw him. This despite going in with a negative impression of him (that’s why his Fav/Unfav ratios are sooooooo much better in IA and NH . . . where he’s gotten himself some exposure)

  3. Argamenon Says:

    The numbers are not that different. McCain does better on ARG because they poll independents. Thompson does worse for the same reason. Romney’s numbers are in line with the other polls.

  4. Tommy Oliver Says:

    Look at the above polls. Jeff, my graphs didn’t work, so I wasn’t able to get the full context of what I was trying to do. I had a lot more polls, and results from prior elections where ARG was off. However, my graphs didn’t work and I’ll be damned if I’m going to type all of them out.

    Debunking, in the context, means they likely have something off with either their sampling data or their system in some way. I don’t think there is a conspiracy. That would be stupid.

  5. Tommy Oliver Says:

    And answer me this. Huckabee dropping eight points in South Carolina. What caused this?

  6. Adam Says:

    Tommy’s right. There is no reason Huckabee has lost 8 points in SC. Nothing in the race has changed to make his numbers sink. Also I don’t know about any positive trend for Mitt in SC. It’s just not there. No poll has ever shown him getting traction in the state. If there is going to be a surge in support for him it’s not going to happen while his national numbers are tanking, regardless of any ads he may have run. You don’t have to take my word for it. it will be corroborated by other SC state polls, just like after the ARG Big 29 for Mitt was followed by subsequent polling showing him losing or barely ahead in the state.

  7. Adam Says:

    sorry that should have read ‘ARG big 39′…

  8. Adam Says:

    …sigh…one more time…. ‘ARG big 39′ in MI.

  9. Jeff Fuller Says:

    http://americanresearchgroup.com/

    Scrolling through all of ARG’s recent results don’t seem to show any “Pro-Mitt”, “Pro-Mccain”, or “Anti-Fred?” in the national polling (it seems quite in line with most others). Also, the Bush and Congress approval ratings seem in line.

    Also, anyone else see that Fred’s numbers are almost all from men in that national poll. 37% of men wanted Fred and only 7% of women. That’s over a 5:1 ratio! Wonder what the campaign is doing about that . . . ?

    At least in Iowa ARG SHOULDN’T include any independents (as they consistantly do) because Indy’s aren’t allowed to caucus and nearly none of them sign up with a party just to caucus.

    And in reply to #3 above . . . Fred has (unexplicably IMO) done quite well among independents from the get-go.

  10. ACT Blog Says:

    Adam – I don’t think you can attribute Romney’s drop to Thompson’s announcement. Look at Rasmussen – Romney is basically back where he has been. The other polls might not show it for a little while, but you can’t really say Romney is “tanking”.

  11. Tommy Oliver Says:

    Jeff,
    There is no pro anybody, they just show a lack of consistency, when compared monthly or to other polls. Look at the movement month to month. You have people gaining 10 points one month, losing a large number the next. There is no basic trend or outlier.

  12. Jeff Fuller Says:

    Huck could be down because their Aug polling before was right after Ames and he was getting great national exposure . . . but that has worn off a bit and Fred has entered officially to fight more for some of the religious conservatives.

    Conversely, explain to me why Huckabee went UP 6% in the Aug poll . . . AMES right? Now he could be back down where he was before.

    And when you’re talking about single digit percentages you always have to consider the Margin of Error (MOE) which was +/-4% for this SC poll. Actually Huckabee might have been measuring higher than reality in Aug and lower than reality now . . . but still all within the MOE.

    Look, I’m not arguing that these ARG numbers are accurate or will hold up through time . . . but I do think that competitors ignore these trends at their own peril.

  13. Tommy Oliver Says:

    Jeff, also look at the Florida results above. Every one of those polls taken ended on 9/19. All the results are comparable except for ARG. Thompson has 23% in all three polls, but at 16 in ARG. McCain in 3 polls is within 3% of each result, but is at 18% in ARG.

    Look at NH in June, McCain is between 15-20% in three polls, but at 30% in ARG, taken at the same time. Look at Romney in the same period, 27-28 in 3 polls, while 23 in ARG. Thompson between 9-12 in three polls, while a 3% in ARG.

    There is something off here by looking at just those few results by themselves.

  14. Jeff Fuller Says:

    http://americanresearchgroup.com/pres08/screp8-708.html

    Looking at the monthly trends in SC ARG polls there is a precedent for some large shifts . . . most in line and confirmed by other polls.

    Ignoring MOE arguments:

    McCain dropped 9% from May-June and then 13% from June-Aug . . . or a 22% drop in two months.

    Huckabee went up 6% (or tripled his support) from June-Aug.

    Fred jumped 6% from May-June and then another 8% from June-Aug . . . or a 14% jump in two months more than doubling his support.

    However, I do have to admit that none of these previous month’s ARG polls had as much change as this one . . . with Romney going up 17% (nearly tripling his support), Huckabee dropping 8%, and Fred dropping 11% (more than halving his support.).

    I truly doubt it will hold up, but, again, all other things being equal (same questions, methods, etc . . .) the results are what they are and probably (over-) represent something real happening down in SC.

  15. Jason Says:

    Why so much effort to debunk a poll? It seems like borderline over compensation.

    Maybe Romney is trending up, who knows, maybe it’s an outlier. But to go to such great efforts really seems a tad bit of overkill.

  16. Big S Says:

    The difference between ARG and firms like Rasmussen is that they have very different definitions of who is a “likely” primary voter or caucus-goer. Until we know how they screen out the “unlikely” voters, we can’t judge the merits of one against the other, even given past performance (don’t give me the song about how Ras is the “most accurate”; so was Zogby, before 9/11 changed things). The screens are generally based upon the voters’ past (i.e. have they voted in previous primaries and general elections) as well as how tuned in they are about the race. It’s this second part that is interesting, since it is a moving target; voters become more “tuned in” as the election gets closer. Rasmussen’s formula is biased towards the conservative side of the electorate, while ARG’s bias is to let more moderates through.

  17. Sammy Sullivan Says:

    check mine out……

    keep up the good work man…….

State of the Race


Obama Approval


Support R4'12

Meta

Recent Posts

Buy This Book

Categories

Archives

Search

Blogroll

Site Syndication

Main