Poll Vault

[gdnp]

USA Today/Gallup has a stunner:

Clearly an outlier for now; if there's really been a shift it will show up in other polls.

Nonsensical, IMHO. From everything I saw and read McCain had a bad week and Obama a good one.

Sympathy vote?

 

[chipmunk stew]

Nonsensical, IMHO. From everything I saw and read McCain had a bad week and Obama a good one.

Sympathy vote?

The Surge^TM?

 

[David Wong]

For all of you who say polling is meaningless, we have:

Gallup USA Today poll, Jul 25-27: McCain by 4

Gallup daily tracking poll, Jul 25-27: Obama by 8.


Uh, Gallup? Hello?

 

[gdnp]

I guess when half the country is on vacation...

 

[aerosolben]

For all of you who say polling is meaningless, we have:

Gallup USA Today poll, Jul 25-27: McCain by 4

Gallup daily tracking poll, Jul 25-27: Obama by 8.


Uh, Gallup? Hello?

The first is likely voters, the second is registered voters. Don't worry, you're not the first to notice.

 

[Brainster]

Like I said, it's an outlier. But it seems possible given what we know from other polls. Young voters (assumed to support Obama) have been self-reporting much lower probabilities of voting. In March 66% of those 18-29 reported themselves as "certain" to vote; today only 46% of those young voters are positive. That's a pretty dramatic change, especially when you consider how well Obama does among those voters. Obama's turn towards the center may be turning those younger folks off.

 

[Pookster]

Interesting article on accuracy of Gallup polls 100 days out from the election ...


http://www.cnn.com/2008/POLITICS/07/29/poll.prediction/index.html


They've only been wrong about who would win twice since 1960. What it doesn't say is if the polls were of registered or just likely voters.

They were wrong in 1960 (Nixon) and 1988 (Dukakis).

The margin of victory is often way off though for many of the elections. So take it all for what it is -- a poll taken 100 days before an election. A lot can happen in 100 days.

 

[Brainster]

They've only been wrong about who would win twice since 1960. What it doesn't say is if the polls were of registered or just likely voters.

They were wrong in 1960 (Nixon) and 1988 (Dukakis).

And arguably (since their poll just measures national voting), 2000 (Bush).

 

[Brainster]

Looks like the current trendline is towards McCain. In the Gallup tracking poll (not the same as the Gallup/USA Today poll), Obama has dropped four points in the last two days.

The percentage of voters favoring Obama for president swelled from 45% in July 21-23 tracking, conducted at the outset of Obama's weeklong visit to Europe and the Middle East, to 49% in July 24-26 interviewing conducted at the height of publicity surrounding the tour. At the same time, McCain's support ebbed from 43% to 40%, and the percentage of undecided voters fell from 7% to 4%.

With Obama now back on U.S. soil and filling less of the nightly news, voter preferences have reverted to their pre-trip levels.

 

[gdnp]

Looks like the current trendline is towards McCain. In the Gallup tracking poll (not the same as the Gallup/USA Today poll), Obama has dropped four points in the last two days.

Hmm. Peaked at 49, now at 46...that would be how many percent?

I am still not convinced we are seeing much more than random statistical fluctuations. Still, if Obama really did bump on his world tour, it bodes well for him as people start paying more attention in the fall.

 

[chipmunk stew]

Hmm. Peaked at 49, now at 46...that would be how many percent?

I am still not convinced we are seeing much more than random statistical fluctuations. Still, if Obama really did bump on his world tour, it bodes well for him as people start paying more attention in the fall.

Yeah, I don't see how you can construe that graph as a "trend" one way or the other. If anything, it just looks like the numbers are settling back to the level they were at pre-travel.

I think you're right, gdnp. The only thing this graph seems to suggest is that exposure is good for Obama.

Something I find strange about the RCP Average is that, while it also appears to show a little "bump" for Obama during his trip abroad, it shows a big "bump" for McCain:
http://www.realclearpolitics.com/epolls/2008/president/us/general_election_mccain_vs_obama-225.html

Again, this could be noise, or it could be simply that when there's high election coverage of any kind, people are much more likely to state their preference (as opposed to stating Undecided) than they are during slow election coverage, and it makes sense on an intuitive level that there are more McCain leaners among the Undecided.

 

[Pookster]

Looks like the current trendline is towards McCain. In the Gallup tracking poll (not the same as the Gallup/USA Today poll), Obama has dropped four points in the last two days.

They're both within their range of daily polling values for the two week period (McCain 40-44%, Obama 46-49%). When they begin to deviate from those, then you may have a change in the trend. Otherwise, both are basically scoring the same with the usual noise that can happen.

 

[Brainster]

Further evidence on recent movement for or against Obama can be found at Intrade:



The last two weeks have seen Obama's shares slide by about 5 cents, while McCain's shares have risen by a similar amount:



Obama still remains the clear favorite over at InTrade. But the direction is very much against him, and barring a significant change in the polls soon, we should see the stock prices of Obama and McCain start to converge rapidly.

Although these are not "polls" per se, I hope the mods will accept them as reactions to polling data and thus on-topic to the thread.

 

[gdnp]

Although these are not "polls" per se, I hope the mods will accept them as reactions to polling data and thus on-topic to the thread.

Clearly on topic, IMHO, as would be oddsmaker's spreads in Britain, for example. Another expression of consensus opinion.

 

[Pookster]

From CNN/Opinion Research Corp. poll ...

http://www.cnn.com/2008/POLITICS/07/30/campaign.poll/index.html

It was a weeklong trip that drew blanket media coverage and sharp criticism from Republicans, but Sen. Barack Obama's recent tour abroad did little to alter his standing in the presidential race, according to a new CNN/Opinion Research Corp. poll.

According to the survey -- the first national poll conducted entirely after the Democratic presidential candidate's trip to Iraq, Afghanistan, the Middle East and Europe -- the race for the White House has remained virtually unchanged since late June, with Obama holding a 51 percent to 44 percent edge over presumed Republican nominee Sen. John McCain.

In a similar poll conducted one month ago, Obama held a 5 point lead over McCain, 50 percent to 45 percent.

The things McCain has to worry about the most ...

McCain falls behind Obama in every single economic and domestic issue tested by the poll, especially when it comes to jobs and health care. Obama polls ahead in both those categories by double-digit margins.

...

The Arizona senator also polls lower on the issue of high gas prices, a problem McCain has repeatedly said he's better able to alleviate than Obama.

The poll, conducted July 27-29, surveyed 914 registered voters and carries a margin of error of plus or minus 3 percentage points

 

[Pookster]

Rasmussen matchup with leaners over the past two weeks ...

http://www.rasmussenreports.com/pub...al_election/general_election_match_up_history

                 July 16     July 30
McCain            45%         46%
Obama             48%         48%

 

[Pookster]

And arguably (since their poll just measures national voting), 2000 (Bush).

I agree. Gallup was close to the final result 100 days out only three times since 1960. They may be accurate (which may just be pure luck) with most of their victors, but they're way off on how the margins play out.

 

[Brainster]

I agree. Gallup was close to the final result 100 days out of only three times since 1960. They may be accurate (which may just be pure luck) with most of their victors, but they're way off on how the margins play out.

In fairness they may have accurately measured late July sentiment and been made to look bad by the fickle public. On the election they were most wrong about, (Dukakis/Bush in 1988), most observers at the time agreed that Dukakis seemed to be in the lead in mid-summer and got crushed.

 

[Pookster]

In fairness they may have accurately measured late July sentiment and been made to look bad by the fickle public. On the election they were most wrong about, (Dukakis/Bush in 1988), most observers at the time agreed that Dukakis seemed to be in the lead in mid-summer and got crushed.

I mostly agree. Most people need a significant reason to change their minds. Bush's campaign was quite effective in giving them that reason after July of 1988.

 

[Brainster]

Gallup daily tracking from Friday shows the race tied:

This is not much different from the results reported in Gallup Poll Daily tracking on Thursday, when the two were nearly tied with 45% for Obama and 44% for McCain. However, it is a substantial turnaround from earlier this week when Obama held a statistically significant lead coming off his high-profile trip to Europe, Afghanistan, and the Middle East.

Three-day rolling average, with 2680 registered voters and error margin of +/- 2%.