Battleground State Polls- Zogby

[The Vorlon]

Here is the link
http://online.wsj.com/public/resources/documents/info-battleground04.html

 The polls are clearly a good sign for Kerry, but most do believe Zogby polls have a slight democratic bias. These state polls are very interesting. Bush ahead by 5% in Iowa and behind 3.3% in Missouri? I would think Bush is clearly ahead in Missouri. Also, if Zogby is slightly democratic as I believe, then Bush is likely slightly ahead in Florida. The polling from PA, and WI are particularly bad looking for Bush.  rkansas looks safer Bush after this, but many states are leaning to Kerry at the time being.

Would this be the ONLINE poll (ie Internet based) that Zogby is trying to sort out?

Why yes...  yes it is...

[The Vorlon]

Well then nevermind, the results mean basically nothing in my opinion. I wonder if this is the only kind of state polling Zogby is doing this year? I remember in 2000 he had state polls he conducted daily for about the last month or so before the election that were just his regular polls, I guess I just assumed this was something like that. I don't even know how he expects to get any poll of any importance from online.

Actually in 2000 Harris Interactive (Ie the Harris poll people) ran an internet based poll where they screened people pretty heavy before the included folks in a sample and then weighted the ^&&^^$ out of the result and they NAILED it, there final call was Gore +.6% - which differed from the actual result by only 0.09% (!)

Now they may have just gotten totally lucky, or there may be something to it...

30 years ago when the first started doing telephone polling versus face to face, Telephone polls were considered voodoo too...

Internet polls are clearly not sorted out yet, but I do think they are the future.

You get 100,000 people, you screen the %$%% out of them to determine their EXACT demographic profile, weight accordingly, and you should be damn close...

In theory....

unless all the online people have lied to you about who they really are demographically/politically...

[The Vorlon]

People have said of Zogby in the past..."Live by Zogby, Die by Zogby".

I want to point out as well that Zogby nailed the 2000 election, the only major polling operation that tracked the late surge by Gore the weekend of Bush's DWI revelation, calling Gore the winner by .05%. Most other polls had Bush 2% to 6% ahead of Gore the weekend before. Zogby's Democratic Primary polls this year were pretty close to actual results as well.

My own view is...I would tend to rely on his polls more than I would discount them.

That being the case, this latest polling effort by him is very good news for Kerry!

Actually a number of firms got 2000 right.

The pollsters had an excelent year in 2000.

Zogby has Gore +2 which was a great call,

But Harris (Both phone and Internet) both called it a dead heat, as did FoxNews.

TIPP also clearly tracked the Gore rise at the end, they ended up at Bush +2 in their 3 day tracking poll, but their certainly nailed the trend going from Bush +8 to Bush +2 over the last 3 days.

Even CBS (which I love to hate) got lucky and called Gore +1.

To give you an idea how good a job the polls did in 2000, Zogby's 2% error, which many years would have "won" the honors actually had him tied for 5th!

Zogby had a brutal year in 2002 however.

He polled 17 Senate/Governor races and got only 12 right! and had the highest average margin of error of any major polling firm.
(According to the national Council on Public Polling)

Zogby is a wildcard - he has been spectacularly right and wrong so many times it is hard to know what to think of him.

[The Vorlon]

You get 100,000 people, you screen the %$%% out of them to determine their EXACT demographic profile, weight accordingly, and you should be damn close...

Vorlon,

I'm no poll expert like yourself (I'm reasonably well versed, have an education heavy on math and science, and am lucky enough to have a few connections in some of the polling organizations... but I'm not The Vorlon!).  In fact, I've referenced you a couple of times already in my brief existence on this board.  So, I was surprised to see your response to this... especially since Zogby's numbers look a bit askew in a few of these states (maybe that's because of his specific weighting).  But, the argument makes sense (with the caveat you point out... that there is not tremendous lying going on)...

As long as you know the existing demographic, you can screen the data to come up with a reasonable result... even though the raw internet poll data is probably wretchedly skewed.

I know you're not giving the Vorlon thumbs up to internet polling... especially one by Zogby.  But it's interesting that you don't dismiss it out of hand (and your reasoning makes perfect sense).

Hope you're quoting me accurately..

According to a lot of posts I've read I have  said a lot of things I don't every recall actually saying..  

I could see an internet survey working something like this;

You sign up 100,000 volunteers, you then telephone them back and screen like crazy to nail down exactly who they are - you get names, phone numbers, SS numbers, etc... and verify like crazy...

You then "build" a sample of maybe 10,000 that looks exactly like America based on this data, amd then give them all unique ID numbers to enter when you poll them...

The huge key is how valid you data is from the internet.

How does that joke go....

"Lesbian Internet romance ends badly when both guys meet..."

[The Vorlon]

in fact, I do actually disagree with you a bit on the Fox/OD reliability... but that's for another thread

I don't follow anyone on blind faith, but I do consider the source.  

You at least seem to be a straight shooter.

Opinion Dynamics is a pretty darn good firm  There President John Gorman is an old polling combat vetern from a zillion years ago. - A Tad "old school" for my tastes, but still good.

It is also nice to have him in my "top tier" pollsters for optics - I have Democracy Corps in there, and didn't have a 'GOP' firm in there to balance it out.

I'd actually like to put Public Opinion Strategies polls up, but their stuff is all copyrighted and private.

<<I don't follow anyone on blind faith, but I do consider the source. >>

Good advice, today, tomorrow, anytime, any place..

<<You at least seem to be a straight shooter.>>

All I wanna do is call all 50 states correctly in 2004
(Got 48/50 in 2000)

[The Vorlon]

LOL, you are pretty damn slick there Vorlon man. What about D.C.? you gonna call that right? .

I have, at this early date, gone out on a limb and Called DC for Kerry, plus Utah and Wyoming for Bush.

I am waiting for more data before I call Rhode Island and Massachusetts for Kerry.

It's early, but I have Idaho and Alaska looking good for Bush.

[The Vorlon]

What states did you call wrongly Vorlon?

Iowa & Wisconsin. (%$#@ing cheeseheads)

Looking at 2000 versus 2004 I think one factor that nobody (except me of course ) has factored in is how good a campaign manager Donna Brazille was from a pure logistical get out the vote point of view.

I am reminded of this because Brazille "stole" Wisconsin, Oregon, Iowa, New Mexico, maybe even Minnesota for Gore,

Gore  was down 2 or 3% in all of those states and she just plain kicked some GOP butt in voter turnout.

Brazille's "Get out the Vote" was really truly outstanding, she just blew away all the turnout projection models in a lot of the big states: Michigan, Pennsylvania, Florida, even Ohio to a lesser degree.

Based on what I have seen so far, the Kerry folks are just not even in the same league as Brazille.

Your ground game is worth a difference of maybe 4% or so ultimately, and in 2000 the "Team Brazille" clearly beat "Team Rove" at the ground game, it wasn't even close. - They got the whole 4%

The GOP has done a very good job on the ground so far in 2004, and Kerry better close that gap, and soon...

This to me is the unreported story so far in 2004.

[The Vorlon]

What states did you call wrongly Vorlon?

Iowa & Wisconsin. (%$#@ing cheeseheads)

Sorry!

Ah... present company excepted... of course