Pollster.com

Articles and Analysis

 

POLL: Time Magazine National


Survey of 805 registered voters conducted by ABT-SRBI for Time Magazine, June 19-25, margin of sampling error +/- 3.5% (Time article; SRBI analysis, results).

National
Obama 43%, McCain 38% (w/o leaners)
Obama 47%, McCain 43% (w/ leaners)

"When undecided voters leaning towards Obama and McCain are accounted for, the race narrows to a mere 4 percentage points."

 

Comments
carl29:

Can anybody tell me where to get the demographic breakdown? I don't see it anywhere. According to the McCain camp, the GOP has a 10% disadvantage with regards to party ID. Did you see anywhere in the article the reference of what % of dem. support Obama and what % of rep. support McSame? The only thing they mentioned is independents, which are split between the two.

I honestly don't like coward people. Time should have published the demographic and party breakdown, so we could see whether they picked the wrong people and wrong amount of people. If for some reason they are wrong with the sample, that's fine live with it. I don't feel I can give my two cents with a poll for which the facts are just unknown. Let us see the ingredients!!!!

____________________

Undecided:

What is probably more important is comparing this Time Magazine poll to its last poll as stated in the article:

February 2008 : Obama 48% to McCain 41%, including leaners.

____________________

Undecided:

You might take a read of Tim Rasmussen's article "Why Polls Sometimes Show Different Results" at http://rasmussenreports.com/public_content/politics/election_20082/why_polls_sometimes_show_different_results


____________________

Mike_in_CA:

@ undecided: You need to learn how to read polls, specifically re: margin of error. The biggest problem with polls is that media who reports them sensationalizes them to an extreme extent. Feb it was 48-41, now it is 47-43, both are well within the margin of error of 3.5%.

Margin of error means that Obama's support in the Feb poll could have been 44.5-51.5, and in the June poll 43.5-50.5. In my opinion, those overlap quite sufficiently. Learning the read polls correctly will remove a lot of the sensationalism of their reporting in the media.

A better way to read this poll is that : Not much has changed since February.

____________________

Undecided:

I do not need any lessons in statistics from Mike_in_CA. It is legitimate to compare two polls that used the same methodology for indication of trends. If this were an experiment, the two results could be used to determine a differential with a stated "degree of confidence."

And I also hope Mike will not be offended when others choose to use the same "lesson" directed at me in defending McCain's support level in the polls.

Finally...My comment was neutral and specific as follows: "as stated in the article." Maybe Mike should read the article more carefully.


____________________

Mark Blumenthal:

@carl29: The full results are usually posted at the SRBI web site within 24 hours of the release by Time. We'll add the link to the PDF when it becomes available.

____________________

Mike_in_CA:

@undecided. your condescension becomes you. My comment specifically addressed the notion that you CAN legitimately compare trends between two polls whose margin of errors overlaps. As for McCain's level of support, it's the same thing as Obama's, re: in Feb his support was 37.5-44.5 and in June it is 39.5-46.5.

Maybe you do need a lesson in statistics. I was pointing out that the problem with polls is that MOEs are regularly discarded for purposes of arguing "trends" (especially in the media) and that that is the most ridiculous thing ever. As a statistics major, if I were to ever make that claim I would have probably failed my exam.

Your harassment is disgusting.

____________________

Adam_Il:

Election is close enough, so much that the person who wins the undecideds wins the election. Obama's track record in winning undecideds wasn't great against Clinton. I wonder how will it be against Mccain?

____________________

carl29:

I went to REAL CLEAR POLITICS and found out that in each and every previous poll they had included party ID and demographics. I am left wondering, Why they changed their policy now? I hope someone asks the question. Why was it OK to share that information in every poll but not in this poll?

____________________

Tzal:

I took one statistics course in college. That was more than ten years ago. I have a few embarrassing questions.

Are there different ways to calculate MoE? In particular, should the formula depend or take in to account demographics? Can it? I read a wikipedia article on MoE and it seemed that the assumption of a random sample is important in the confidence of the statistic. But most polls are not truly random in that they are weighted based on demographics or other factors. Is the way pollsters calculate MoE important in determining the value of the poll?

____________________

carl29:

If that had been Time policy since ever, I would mind, but that is not the case:

Feb. 2008:
Republican 31%
Democrat 38%
Indep. 22%
Something else 7%
Don't know 2%

April 2007:
Republican 30%
Democrat 35%
Indep. 26%
Something else 7%
Don't know 2%

March 2007:
Republican 29%
Democrat 35%
Indep. 27%
Something else 8%
Don't know 2%

Feb. 2007:
Republican 29%
Democrat 32%
Indep. 27%
Something else 9%
Don't know 2%

____________________

carl29:

I cut some slack for Time and thought to myself: Um, well, they maybe don't include this information in general elections match-ups. Maybe that is the difference with the other polls in Real Clear Politics since all of them covered the primaries. I was wrong!

I went to the archives for the 2004 general election and found out that TIME DID INCLUDE party ID and demographics. Let's see:

Methodology: The TIME magazine poll was conducted by telephone October19-21, 2004 among a random sample of 1,200 adults throughout America. The random sample includes 1,059 reported registered voters and 803 likely voters. Likely voter figures include leaners and exclude refusals. The margin of error for registered voters is approximately +/-3 percentage points. The margin of error for likely voters is approximately +/- 4% points.

Likely voters reported party identifications are: 35% Democrat, 35% Republican, 23% Independents. Registered voters party affiliations are: 35% Democrat, 33% Republican, 23% Independent.

They had more than one Time poll and each and everyone had tons of demographics breakdown. So, the question remains, May I see the demographics and party ID, please? As it used to be in the "good old days."

____________________

Undecided:

I will point out to Mike_in_CA.--that it is he who first addressed me with the statement "@ undecided: You need to learn how to read polls" along with a "lesson in statistics."

I responded with my own "lesson" in statistics without derision. And I suggested that Mike read the article more carefully because he would have realized that his charge that "I need to read the polls" was unwarranted. I only conceptually repeated a statement from the Time poll article (as clearly stated), thus my ability or inability to read polls was totally irrelevant.

Yet, he now deems to call me "condescending?" (When did it become condescending to discuss "degree of confidence?") Further he accuses me of "harassment." (How does a singular response equate to harassment?) Neither charge against me is true in fact or intent. But after this latest statement of his, maybe he will now proudly wear those monikers of "condescender" and "harasser."

I find it very disturbing that one cannot make factual (and in this case extraordinarily neutral) statements without being personally targeted for belittlement and false accused.

____________________

Undecided:

I am confused by carl29 because the following is posted above from Mark Blumenthal
:
@carl29: The full results are usually posted at the SRBI web site within 24 hours of the release by Time. We'll add the link to the PDF when it becomes available.
Posted on June 27, 2008 11:01 AM

____________________

carl29:

I'm sorry but I just don't see this information at the top of this page. Where is it? This is all I have:

June 27, 2008
POLL: Time Magazine National

Survey of 805 registered voters conducted by ABT-SRBI for Time Magazine, June 19-25, margin of sampling error +/- 3.5% (article).

National
Obama 43%, McCain 38%

"When undecided voters leaning towards Obama and McCain are accounted for, the race narrows to a mere 4 percentage points."


-- Mark Blumenthal

____________________

Mike_in_CA:

@undecided: my original post was moreso directed at the deceitful way that the media reports polls, and I did not intend to give you a "lesson in statistics" (if that's what you'd even call it). As someone very familiar with statistics, I was aiming to provide a more critical, unbiased look at the results, namely in that they weren't all that different between the two polls. If anything I was extrapolating on your original post. I sincerely apologize for any condescending comments. (Though both of your responses were quite biting and a bit ad hominem).

@Tzal: Margin of error can and does change when polls are re-weighted by party. it does not appear that this poll was.

____________________

Undecided:

Mark's response is embedded within the discussion posts. Scroll up.

____________________

carl29:

Oh...It was like a regular blogger. Oh my!! He answered my post! Thank You Mark!!!

@Thank you undecided.

Mark: Do you have any rating for this pollster, I mean Time/ABT-SRBI. I checked the pollster scorecard at Nate's site and he doesn't even have it on the list. Any idea?

P.S: Thak you for your kindness!

____________________

Tzal:

Mike: Right, but what I'd like to know is if you can look at the MoE in a specific poll and compare it to other numbers, such as sample size, to get an idea of the pollster's confidence in his own poll?

If you have a rather large sample, but an MoE greater than what is to be expected given that sample size, would that suggest that the pollster is not confident in his methods? How does a pollster "control" MoE when things like demographics and party ID are such important components of polls?

Like I said earlier, I have very limited experience with statistics. Sorry if I am approaching this from left field, it's not intentional.

____________________

OGLiberal:

I went to the SRBI site and found the party ID breakdown:

Democrats: 36%
Republicans: 28%
Independents: 28%

That sounds about right...or at least close. It's probably more like a 10% gap between Dem and Repub so this poll may have undercounted Dem voters a bit, but not by much. If you factor in MoE, this is pretty close to where I think the race actual is nationally - a 5-7% Obama lead. But Obama's performance in swing states is much better than McCain's at this point, at least from the polls I've been reading.

I'm having a difficult time with their 45-44 split of women voters between Obama and McCain - no way Obama is doing that poorly with women. But then they also have Obama over McCain with men, 49-42. I'd expect it to be tie or McCain with a small lead among men. So even though these both look supsect to me, they may balance each other out.

Would have preferred a larger sample size in this one. The latest Q-Pac state polls included over 1,000 voters each...and they were for individual states. An 800 voter sample size for a national poll seems to me to be too small to get an accurate reading of the electorate.

____________________

carl29:

Now, I'm confused.
McCain gets 82% of Republican
Barack gets 79% of Democrat

McCain gets 13% of democrats
Obama gets 12% of republicans
(I don't know you but I thought that more democrats will vote for McCain than republicans for Obama, but see this poll? I don't know if I can believe it)

____________________

pluck:

This is the wrong thread to post in, but there's no thread here that mentions the recent L.A. Times/Bloomberg poll showing Bush with a 23% approval rating. Seeing as how that particular poll has generally put Bush above the trend line, it's a remarkable result.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR