Pollster.com

Articles and Analysis

 

Is the Race Narrowing?

Topics: 2008 , Barack Obama , John McCain

Is the presidential race tightening? Our national trend chart shows a two to three point decline in Barack Obama's support to the 44.7% indicated by our trend line this afternoon. Do we see the same pattern in the statewide polling released over the last few weeks?

The following table shows the results from 26 surveys released over the last two weeks where the pollster also did a survey using the same methodology at some point in June. Not surprisingly, the polls show considerable variation, although more show a net improvement of at least one point for McCain (17) than show movement in Obama's direction (7).


07-25 StateChange.png

If there had been no change anywhere in presidential preference between June and July, and if each poll used an identical methodology, then the June to July variation should be entirely random. In that case the pattern of change -- whether Obama or McCain gained -- should be like flipping a coin. About half the surveys would show improvement for one candidate, about half for the other. If we throw out the two ties, we have 24 tosses. The probability of flipping a coin and having it come up "heads" at least 17 times out of 24 by chance alone is about 3%. Pollsters usually consider a difference to be statistically significant when their level of confidence is at least 95%. So we can assume that a two or three point gain for McCain over the last month is probably real.

This week, some voices in the blogosphere have been wondering why Barack Obama's travels to the Middle East and Europe have not produced more of a "bump" in these surveys. As should be obvious from their timing, it is a bit premature to judge the impact of the trip from the state survey results. As for the national tracking polls, given their usual pattern of day-to-day random fluctuation, we are best advised to wait a week or so before coming to firm conclusions (even if today's results may suggest a bounce).

If I had to guess about the slight tightening of the race evident in the national chart and the table above, I would bet on a combination of two sources: The critical news coverage Obama received earlier in the month (the "flip flop" theme) and the McCain paid advertising wave over the last month or so that was reportedly heavier than Obama's.

It is always important to remember that the voters most likely to shift their preferences, especially at this stage of the campaign, tend to be those who pay little attention to news about politics. Political advertising -- which reaches out to those ordinarily not tuned in -- is more likely to reach low information voters.

 

Comments
tdub:

The coin-toss analysis seems flawed. A 97% likelihood that the results are not ENTIRELY from chance doesn't mean a 95% likelihood that all of 2-3% is non-chance-induced.

____________________

Undecided:

I read "statistically significant" and that McCain's 2-3 point gain is "probability" real. The coin toss is a good analogy to help neophytes understand statistics.

Thanks, Mark for the data and learning experience.

____________________

RS:

Mark:
I think the reality is that Obama's primary-race-win-bump (prwb?) is coming back down. I wish I could see a more high-resolution version of the national poll chart, but as is, you can see a rise and fall in Obama's support in the last couple months, while McCain's support actually dips a little.

I did a comparison of the Quinnipac numbers and Gallup's June-July averages:
/blogs/poll_quinnipiac_colorado_michi.html#comment-40412

Something similar to what you see... Of course, loss of the prwb may be due to (as you say) the critical coverage Obama's received (McCain's many gaffes barely make it to the MSM), as well as McCain's attack ads (Obama to blame for rising gas prices!)

But the four national polls in July that included both 2-way and 4-way races seem to show relatively steady support for Obama, with Nader (!?) gaining at McCain's expense.

Fox 7/22-23:
McCain 40, Obama 41 (+1)
McCain 37, Obama 40 (+3), Nader 2, Barr 0

NBC/WSJ 7/18-21:
McCain 41, Obama 47 (+6)
McCain 35, Obama 48 (+13), Nader 5, Barr 2

ABC/Post 7/10-13:
McCain 42, Obama 50 (+8, RVs)
McCain 39, Obama 49 (+10), Nader 5, Barr 2

As an Obama-booster, I am much happier with the more realistic 4-way results :-) [Nader costing McCain his shot at POTUS would be sweet indeed!]

____________________

billmon:

Call me cynical (I would) but I have a sneaking suspicion that over the past month a few percentage points of the exceptionally slow-witted redneck vote finally realized that Obama is, in fact, black.

____________________

Alan Abramowitz:

Agree with Mark that state polls seem to show a slight narrowing of the race. But one contrary piece of evidence is the Gallup tracking poll. This is based on more than 20,000 interviews with RVs in June, more than 18,000 so far in July.

Average Obama Margin in Gallup Tracking Poll by Month

June (27 days) July (23 days) Change
_______________________________________________

OVERALL + 2.5 + 3.5 + 1.0

Male - 5.0 - 5.0 0.0
Female + 9.7 + 11.3 + 1.6

White - 10.7 - 10.7 0.0
Black + 86.7 + 86.3 - 0.4
Hispanic + 31.0 + 32.0 + 1.0

18-29 + 24.0 + 28.7 + 4.7
30-49 + 0.3 + 1.7 + 1.4
50-64 + 2.0 + 3.0 + 1.0
65+ - 4.3 - 9.3 - 5.0

Northeast + 13.7 + 16.0 + 2.3
Midwest + 5.3 + 8.7 + 3.4
South - 10.3 - 10.0 + 0.3
West + 8.3 + 5.0 - 3.3

Red States - 10.7 - 8.7 + 2.0
Purple States + 7.0 + 8.0 + 1.0
Blue States + 17.3 + 15.7 - 1.6
_______________________________________________
Source: Gallup Poll

____________________

Alan Abramowitz:

National non-tracking polls also show little or evidence of any change. Only the Newsweek outlier in June makes it appear that Obama margin has declined slightly. Without Newsweek, there is no change or even a small Obama gain:

Average Obama Margin in National Non-Tracking Polls by Month
_______________________________________________

Poll June July Change
_______________________________________________

Fox + 4 + 1 - 3
NBC/WSJ + 6 + 6 0
CBS/NYT + 6 + 6 0
ABC/WP -1 + 3 + 4
Reuters + 5 + 7 + 2
Quinnipiac + 7 + 9 + 2
IBD/TIPP + 3 + 3 0
Newsweek + 15 + 3 - 12

Average + 5.6 + 4.8 - 0.8

Without
Newsweek + 4.3 + 5.0 + 0.7
_______________________________________________
Source: Pollingreport.com

____________________

jsh1120:

I've been considering this issue for a week or so and would suggest the following hypothesis.

It was often noted during the primaries that when Obama advertised and campaigned, his support increased. That would seem to be a no-brainer, but it isn't. In fact, many politicians (e.g. Giuliani, Thompson) seem to suffer from exposure. The more voters see them, the less they like them. Others seem to have to invest vast resources to move the polls even a bit.

It appears that despite his overall disadvantage in resources McCain invested considerably in advertising in June while Obama spent in terms of infrastructure in a variety of states.

My guess is that voters not already committed to one of the two candidates are likely to "slip back" into indecision or weak support in this relatively "low attention" stage of the general election campaign when the drumbeat of Obama's campaign has been temporarily quieted.

And while I'm skeptical of the "bump" from Obama's foreign travels, I think the initial results may support this hypothesis.

Like a manufacturer of a new product in an established market space, it's up to the Obama campaign to assure that "buy pressure" builds in the last 10 weeks or so before the election. His earlier performances in various state primaries suggest his campaign is up to that task.

____________________



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