« 2008: Harriet Miers, Republican Presidential Hopefulls, and Right-Wing Blogs | Main | TX-22: Tom DeLay Indicted Again »

Monday, October 03, 2005

2006: Zogby Back in the Game

Posted by DavidNYC

Last year, Zogby did a bi-weekly "interactive" (ie, online) presidential poll in (almost) all the swing states. Their scope was bold, their execution, well... not terrific. The following table compares Zogby's last poll in the sixteen states with the actual results:

State Zogby Actual
AR 3.2B 9.8B
FL 1.0K 5.0B
IA 6.0K 0.7B
MI 7.0K 3.4K
MN 2.0K 3.5K
MO 4.2B 7.2B
NH 4.5K 1.4K
NM 0.0 0.8B
NV 5.0B 2.6B
OH 4.0B 2.1B
OR 10.1K 4.16K
PA 5.0K 2.5K
TN 3.3B 14.3B
WA 11.2K 7.2K
WV 4.5B 12.9B
WI 7.0K 0.4K

The states are color-coded based on how they went (or how Zogby thought they'd go) - the K & B are added for those who have difficulty with colored charts.

At first glance, you might think, okay, Zogby only got two states wrong, Iowa and Florida. So did a lot of people, right? But look at those margins. It's pretty ugly.

In only three states did Zogby understate Kerry's strength - Minnesota, Nevada and Ohio. In fully thirteen others, Zogby underestimated Bush's margin, sometimes by laughably big margins. (Look at the huge spreads in Tennesse and West Virginia, for instance.) This Kerry favoritism was something I noticed before the election, but I never bothered doing a full post-mortem afterwards, so I didn't learn the extent of Zogby's failure until now.

How, then, are we to take Zogby's newest effort - monthly polls of 17 senate and 21 gubernatorial races, from now until election day? It's always fun to see this many races polled at once, but is the data any good at all? Will the trendlines even make any sense? I felt that the presidential polls were unusually volatile, so I was hesitant to trust even the trends.

There was also a suggestion last time out that Zogby's Internet polls were as open as your average podunk NewsChannel7 web poll. That was never fully confirmed, but nor did I ever see a good explanation refuting or nullifying that claim. I also haven't seen any polling post-mortems from Zogby himself - usually, firms analyze how they did after each election. I haven't been able to track down any such self-assessments from Zogby, but if you've come across any, please let us know in comments.

Now it's also possible I'm being overly harsh - it's not entirely fair to hammer Zogby for systematically underestimating Bush without knowing whether other pollsters did the same. Unfortunately, I don't think any outfit polled as many states all at once. (The only thing similar that I remember was a one-time 50-state poll by ARG, but that was a couple months before the election.) So anyhow, what do you think of the Zogby Interactive, its methodology, and its reliability?

I suppose I'll indulge in a brief examination of the substance of the polls, or at least, a few of them. On the senate side, there aren't too many surprises. The undeclared Paul Hackett has an eight-point lead over DeWine (R-OH); the semi-vulnerable Stabenow (D-MI) and Cantwell (D-WA) both have decent leads; NJ, worryingly, is, well - as the Magic 8-Ball would say, "Outlook Hazy - Try Again Later"; the bland Herb Kohl (D-WI) has but a one-point lead; and the possibly endangered-but-should-stomp-Katherine-Harris Bill Nelson is up only 4. Frustratingly, they're not polling two of our best pickup chances: Rhode Island and Montana, yet they are polling New York and Texas, which are nobody's idea of close races.

The gubernatorial side shows one major bright spot: Angelides whooping Ahnuld by 8 in Cali. The allegedly in-trouble Granholm (D-MI) has a 10-point lead; Spitzer (D-NY) has a margin that would make a third-world dictator blush (just kidding - it's only 33%); and in the 2005 races, Corzine (D-NJ) is comfortably ahead while Kaine (D-VA) is back just three.

Whew! Like I said, these mega polls can be fun for junkies like myself. But better pull out your Morton's - cuz as they say, when it rains, it pours.

Posted at 05:36 PM in 2006 Elections | Technorati

Trackback Pings

TrackBack URL for this entry:


I could tell after just a couple polling sessions last year that Zogby's Interactive polls were a joke. I remember Tennessee fluctuating 13 points in two weeks once. And in Ohio, he had Kerry running consistently behind, often as far as 11 points, while in Oregon, Kerry was ahead by double digits nearly through the duration of his polling. It's gonna be hard to take seriously his latest round of Senate and Governor polls but I guess it's better than no having no polls at all to analyze.

The biggest flub would seem to be Wisconsin. Herb Kohl consistently wins by colossal margins, and I have a hard time believing he's lost that much popularity unless there's something going on with him that I don't know about. Other than that, what's the deal with New Jersey? If Corzine wins next month, does he get to appoint his own successor for a year or will that Senate seat simply be vacant for a year? Also, I couldn't discern which Democrat Zogby was projecting to run in Tennessee? Is there a front-runner for that seat yet?

Posted by: Mark [TypeKey Profile Page] at October 3, 2005 08:11 PM | Permalink | Edit Comment | Delete Comment

I don't trust Zogby polls one bit. IMHO, Rasmussen is the most reliable.

Posted by: purpleday [TypeKey Profile Page] at October 3, 2005 09:46 PM | Permalink | Edit Comment | Delete Comment