Error: Unable to create directory /home/demockra/public_html/wp-content/uploads/2010/09. Is its parent directory writable by the server? Tracking Poll Update: Obama +5
by Kevin Van Dyke, Editor
September 26, 2008
For the first time in nine days, the tracking poll average has deviated outside of the Obama +3-4 percentage point window, with Obama taking a +5 composite advantage. Not only is Obama polling well in the two new trackers, but also in the two traditional trackers. In fact, Rasmussen, which has always been the least likely to show swings toward either candidate because of their tight weighting methods, is now showing a five point Obama lead. Today’s trackers will be the last numbers released that include only interviews from before tonight’s debate. Therefore, when looked at in combination with the numbers of the past week, these numbers should give us a good benchmark to assess the effect, if any, of the debates on polling.
Here’s a look at today’s numbers and the trends:
9/26:
Gallup Daily: O+3
Rasmussen: O+5
Diageo/Hotline: O+7
Research 2000: O+5
Simple Average– O+5.0
Adjusted Average– O+5.0
Trends:
Simple Average
9/8:M +2.0
9/9:M +2.0–Height of McCain’s convention bounce
9/10: M +1.4
9/11: M + 1.0
9/12 :M +1.0
9/13: M+1.5
9/14: M +0.25
9/15: Tied—Bounce is naturally fading back to a tied race
9/16: O +1.25
9/17: O+2
9/18: O+3.5—First tracking polls to include financial meltdown numbers
9/19: O+3.25
9/20: O+4
9/21: O+3.25
9/22: O+4—Race appears to be stable for the time being.
9/23: O+3.25
9/24: O+3.75
9/25: O+3.25
9/26: O+5
Obama will have to at least be at +7 on the eve of the election, November 3, to actually win the election with an actual electoral college win of 273 + (otherwise, unfortunately, it will end up either being a draw where both candidates have 269 electoral votes or Obama will actually lose by a slight margin {meaning probably just shy of the supposed loss of Gore’s back in 2000} both in the electoral college and the popular vote mainly because of, unfortunately, the continuing pick up of racism in the polls). There is still time for Obama to reach +7, and it could really be true, especially after the Biden-Palin debate which I expect will be won by Biden and also the fact the next two Presidential debates (on October 7 & 15 respectively) should favor Obama with both focused on the economy. But Obama and Biden also need to make no more mistakes. To make the election a landslide as some Obama supporters seem to hope for Obama needs to at least be at +9 on November 3.
Thanks for commenting. Where do you get this +7 figure from? Has there been a study on this or is this your best guess of what Obama will have to be up by? I assume you’re figuring a +6 swing because of a Bradley effect? I have yet to see any conclusive evidence that indicates this is still an issue. I know it happened in the 80s and early 90s, but there doesn’t seem to be much evidence (that I know of) that it happened in Harold Ford’s Senate race in Tennessee in 2006 (he actually outperformed the polls) for example. Also, there wasn’t much of this in the primary this year, outside of maybe Kentucky and West Virginia.
Also, what about a reverse Bradley effect in many southern states during the primary season (many polls underestimated Obama’s support with black voters)? Could these factors possibly cancel each other out in states like Virginia or North Carolina?
Finally, if we assume the Bradley effect to still be relevant, what states do you think this effect would be the most prominent in?
Thanks again for the comment.
In response to the person who mentioned the Bradley effect. It took me a week of research on political polls to derive at the below stats.
In the polls listed below which have used race, religion and gender as a criterea, in every state where there is a large minority population Obama has polled well above 60% among blacks. This is an averaging over the last two months and does not reflect the Debate.
The Camelot-Thurston Poll 5100 individual phone calls (4831 responses)
The Gandolf Poll 2730 automated phone calls (1932 responses)
The Gallup Poll 1200 phone calls
The Religious Preference Poll (only represents states in deep south)
1. An averaging of polls breaking results down by race National:
white women – 49% white men – 38%
black women – 89% black men- 71%
asian women – 52% asian men – 41%
hispanic women – 74% hispanic men 63%
other women – 61% other men – 57%
Averaging of polls breaking results down by Religion National:
Born Again Conservative Christian white women – 42% Born Again Conservative Christian white men – 31%
Moderate Protestant white women – 59% Moderate Protestant white men 39%
Catholic white women – 43% Catholic white men 38%
Christian black women – 87% Christian black men- 84%
Christian asian women – 54% Christian asian men – 51%
Christian hispanic women – 65% Christian hispanic men 62%
other Christian women – 61% other Christian men – 47%
Other Religions – Other women 89% Other men 79%
Wow, this is great. Thanks for your hard work breaking down this data This should be very useful going forward.
By the way, in case anyone isn’t familiar with the Bradley effect, I’ll try to explain it quickly:
In short, the Bradley effect is where white voters tell a pollster they plan to vote for a black candidate, even though end up not voting for that candidate. It is based on LA mayor Tom Bradley. Mayor Bradley had a large lead in the California governors race in the early 80’s (he was set to replace Jerry Brown). However, on election day he ended up losing. A similar effect happened in Virginia in the early 1990s, Doug Wilder was polling with a large lead in the race for governor. While Wilder still won, it was only barely.
However, recent state elections haven’t shown evidence that this effect is still alive. Carol Mosely Braun outperformed polls in Illinois in 1998 and Harold Ford outperformed polls in Tennessee in 2006.
What the the Bradley effect is NOT:
The Bradley effect is not when someone votes against Obama because he is black. It’s only when they claim they’ll support him and then don’t actually vote for him.
[...] tracking polls in 2008, one of the big differences between Gallup and Rasmussen was that Rasmussen weighted and Gallup did not, leading to more swings in the Gallup tracker.) My concerns here were that a smaller, more [...]