Mr. Zogby says Bush gains 3%, but still trails by 2%

[The Vorlon]

The ever entertaining Mr. Zogby has a new poll out.

This poll presented for entertainment purposes only, no warranty expressed or implied!

http://www.zogby.com/news/ReadNews.dbm?ID=832

This is actualy tied for the Best result Mr. Zogby has ever given Bush this year in the Head to Head.

Big jump in undecideds.

Bush job approval up 4% to 46% (Not comparable to other Job approval polls, as Zogby asks a different question BTW, but still up 4% which is good news for Mr. Bush)

Right track/wrong track showed a strong improvement for Bush as well, going from -14 to -6 - a gain of 8%

Huge change in "Was it worth it" re war in iraq, went from -3 to +13 (!)

As always, here are the key words in any Zogby poll:

Slight weights were added to region, party, age, race, religion, gender and party ID to more accurately reflect the voting population. Margins of error are higher in sub-groups.

Zogby's own words were:

Pollster John Zogby: “This has been a good week for the President. Voters like to see their President on the world stage, larger than Washington, DC. The photo ops display a world leader meeting with US allies. With the death of President Reagan, Americans have seen the more compassionate side of George W. Bush. But even with the absence of the bad news from the previous few weeks, the President has not gained in the polls. While it is difficult for Senator Kerry to break through a President on the world stage and a nation in mourning, his numbers have actually gone down. The real movement is among undecideds who numbered only 5% in March and now are 11%. The first jump was due to some Republicans and conservatives leaving the Bush column; the last month has seen some Democrats abandoning Kerry. In this highly charged atmosphere, both candidates will have to get back their base to win.”

Details to follow.

[Reds4]

Looks like a poll that is somewhat good news for Bush now at least. A couple of sentences that makes me question this polls findings though "  Kerry is preferred over Bush among men (44%- 41%) and statistically tied with the President at 43% among women. "

Kerry cannot be ahead by 3% among men and even among women.

[agcatter]

Yeah, he's got the gender thing totally messed up.

[The Vorlon]

Looks like a poll that is somewhat good news for Bush now at least. A couple of sentences that makes me question this polls findings though "  Kerry is preferred over Bush among men (44%- 41%) and statistically tied with the President at 43% among women. "

Kerry cannot be ahead by 3% among men and even among women.

Other than being in direct contradiction of every poll taken in the past quarter century it seems to make sense.

ITS A ZOGBY.  THEY NEVER MAKE SENSE.

[Reds4]

Caps lock? Must be emphasising the point. But yeah, this makes me wonder about the sample, who knows maybe he just thinks up some numbers. lol.

[The Vorlon]

Caps lock? Must be emphasising the point. But yeah, this makes me wonder about the sample, who knows maybe he just thinks up some numbers. lol.

In 8 of his last 14 polls he has missed by 6% or more versus the actual election results

Before he "thinks up some numbers" he should thing again..

Zogby's Polls versus Actual Results

[Lunar]

Do you just calculate the error by adding the missed percentages for the two candidates?  None of them add up to 100, so obviously there are third parties at work too.

[The Vorlon]

Do you just calculate the error by adding the missed percentages for the two candidates?  None of them add up to 100, so obviously there are third parties at work too.

Yes, if a poll says X will win by 10 and the actual margin is 4, the error would be 6.

There are a zillion ways you "could" do it.  That is the method the National Council on Public Opinion polls uses, and while not perfect, it is reasonable enough.

[Lunar]

Does Zogby ever defend himself from the obvious proof of some structural errors?

[agcatter]

Being from Texas I well remember the 2002 Zogby polling in the Senate campaign in Texas.  Three days before the election our friend Zogby had Democrat Kirk down just one.  Scared the hell out of me.  For what?  Kirk lost by 12!  (Although Zogby had it a tad closer at 4 the day before the election).  He still missed it a mile.

The Cornyn polling according to someone I know in the campaign told me later never during the campaign was it ever closer than 6 and they had Cornyn ahead by 10 on the eve of the election.  Don't know for sure if that's true or not but given the final margin I can believe it.  How do you miss that badly?

[emergingDmajority1]

it felt like Bush had a better few weeks, and it was reflected in the polls a little. But it's still and uphill battle for him, and I feel as though Kerry has not yet begun to fight.

[The Vorlon]

Does Zogby ever defend himself from the obvious proof of some structural errors?

Zogby reminds people of the times he is right, and hopes they forget the times he was wrong.

In 2000 he did indeed have Gore up 2% at the end, which was an excellent result compared to the actual result - in fact he was tied for being the 6th most accurate pollster among those who released a result just before the 2000 election.

He fails to remind people that in the same election he had:

Bush & Gore TIED in California (Error of 12%) and that he missed New York, Pensylvania, and Illinois by 6%, 5% & 5% respectively.

Oh yes, and that 2000 New York Senate race - he missed that by mere 9%, and the Pennsylvania race by 10%...

[The Vorlon]

I can believe it.  How do you miss that badly?

Zogby, to use a baseball anology, always swings for the fences.

He hopes if he hits 1 home run he hopes people remember the home run and forget the 7 times he struck out.

The big thing he does is he ASSUMES he does indeed know the correct distribution of Democrats, Republicans, and Independants in the population.

To the degree he guesses right, his polls get more accurate, to the degree he guesses wrong, things blow up on him.

A lot of it is just money too.  It is alot cheaper to make a bunch of assumptions than it is to do a poll dead straight by the book.

[Lunar]

People obviously pay him money to keep doing his polls, so it seems strange that he doesn't find and eliminate the error in his work.  The most accurate polling firms get the most money, no?  I suppose cheap quickie pollsters might be an exception, no clue.

[The Vorlon]

The most accurate polling firms get the most money, no?

Unfortunately, in the media arena, no.

When the media is paying for a poll, they are, essentially, buying a front page story.

X leads Y by 7% says the headline....

To put it bluntly, they really do not care  (in many cases) if the poll is right, they just want to buy it cheap.

This is a broad generalization, and there are certainly exceptions.  Recently the Plain Dealer newspaper in Ohio forked over $38,000 to Mason Dixon to do a 1500 sample size poll, but by and large, in media polling, cheap will win over good far too often.

Many of really good firms NEVER get hired by the media because they are just too expensive.

A good example is a firm called Public Opinion Strategies (R)  

These guys are just deadly, their accuracy track record makes a guy like Zogby look just plain sick.  

They have as clients probably a dozen GOP senators, about a dozen Governors and maybe a third of the GOP members of the House, they might do 5 public polls a year, if that.

In a campiagn you care about one thing only from your pollster - how accurate is the information you are getting?

In a senate race when you spend $10 million, the 200K you drop on polling is a drop in the bucket and vital to spending the other 9.8 million correctly.

POA wants about $50 for a state snapshot poll, a price most newspapers and media are just not willing to pay, so the hire the XYZ Telemarketing to do a "poll" for $4K - they still get a headline that says "X leads Y"

Sad, but true.

[struct310]

Well if the Zogby poll shows Bush moving up then that means Bush will be ahead of Kerry by 4 percent in all the other national polls.

[The Vorlon]

Well if the Zogby poll shows Bush moving up then that means Bush will be ahead of Kerry by 4 percent in all the other national polls.

And a new Gallup shows Kerry up by 6 (!)

I am confused...

[MODU]

Vorlon

http://www.zogby.com/news/ReadNews.dbm?ID=836

"The poll, conducted June 14-19, shows Mr. Bush would win 285 Electoral College votes, compared to 253 for Mr. Kerry."

"It marks the first time in the three polling periods conducted so far that Mr. Bush has taken the lead in the race, gaining 43 votes in the last two weeks and 67 in the last month. But races in many states are still so close as to leave the result unclear."

Did not locate the details behind this poll he is running.  Interesting article though, discussing the Ronald Reagan effect and all.  Worth the 2 minutes to read . . . even for a Zogby poll.

[The Vorlon]

Vorlon

http://www.zogby.com/news/ReadNews.dbm?ID=836

"The poll, conducted June 14-19, shows Mr. Bush would win 285 Electoral College votes, compared to 253 for Mr. Kerry."

"It marks the first time in the three polling periods conducted so far that Mr. Bush has taken the lead in the race, gaining 43 votes in the last two weeks and 67 in the last month. But races in many states are still so close as to leave the result unclear."

Did not locate the details behind this poll he is running.  Interesting article though, discussing the Ronald Reagan effect and all.  Worth the 2 minutes to read . . . even for a Zogby poll.

I have always said Zogby is a smart man.

His polls are assumptions, built on assupmtions, built on assuptions and methodologically they are held together with buttlegum, duct tape and chicken wire,....  but he is still a smart man.

I always read what he has to say.... I may not believe it, but I read it.

http://www.uselectionatlas.org/USPRESIDENT/GENERAL/CAMPAIGN/2004/IMAGES/zogby.pdf

This is a FULL breakout of his June 21st set of internet polls, for your amusement.

No warranty.. expressed or implied

[pieman]

I agree Zogby is smart.

He knows that polls that don't move, don't make good headlines. His polls always make headlines and are remembered. Not because they are right, but because they show different results.

He has found a way to create a "reputable" poll (at least reputable enough so that the newspapers buy and print it) that creates outlier results that sell papers.    

I think his internet poll is brilliant marketing. First of all these polls create great headlines because there is always at least one outlier that shakes people up. second, they are cheap.

What more would his best customers, the newspapers, want?

[cwelsch]

Caps lock? Must be emphasising the point. But yeah, this makes me wonder about the sample, who knows maybe he just thinks up some numbers. lol.

In 8 of his last 14 polls he has missed by 6% or more versus the actual election results

Before he "thinks up some numbers" he should thing again..

Zogby's Polls versus Actual Results

You added number twice.  If he predicts 53-47 and the result is 51-49, then he missed the margin by 2% but you would count the difference on both sides and say 4%.

[The Vorlon]

Caps lock? Must be emphasising the point. But yeah, this makes me wonder about the sample, who knows maybe he just thinks up some numbers. lol.

In 8 of his last 14 polls he has missed by 6% or more versus the actual election results

Before he "thinks up some numbers" he should thing again..

Zogby's Polls versus Actual Results

You added number twice.  If he predicts 53-47 and the result is 51-49, then he missed the margin by 2% but you would count the difference on both sides and say 4%.

You are correct, the margin of error goes both ways.. ie +/- 3% means a "lead" error of 6%

My graphic shows deviation from lead.

So if he says X will win by 10, and actually wins by 3, I count that as an error of 7%.

I do it the same for all the firms, so it is valid, just different...