Obama's advantage over Hillary from Rasmussen
Sun Mar 16, 2008 at 06:10:45 PM PDT
silver spring recently did an excellent analysis of the SurveyUSA numbers that show Obama has greater coattails than Hillary. This coattail effect was shown last week when Democrats took control of Dennis Hastert's former seat in Illinois. Regardless of who's president, unless the Democrats get a supermajority, expect to see the remaining Senate Republicans filibuster all good progressive legislation, especially under a Democratic president. So the issue of coattails becomes important.
I want to extend what silver spring did by looking at Rasmussen's state numbers. Now, I trust SurveyUSA the most, as they've consistently shown themselves to get very accurate results. In 2004, they did the best job of any polling firm, and their track record's holding this year too. The second best pollster from 2004 is Rasmussen Reports. Let's look at their state numbers. The full numbers after the fold....
First, a little background on the statistics behind this. If you just want the numbers, skip this part.
Why can I say that multiple polls showing the same candidate leading the polls strengthens those results, even if they're all within the margin of error? I mean, one poll showing candidate A leading candidate B within the margin of error doesn't tell us that much. We'd still call it a statistical tie, and we're not really sure if either candidate has the lead. And you'll probably hear those on the losing side of the poll say how you can't trust the polls, or polls don't matter, or something along those lines. Fine. All this is true.
However, if several other polls from reliable polling firms also come out showing A leading B, then we become more certain A really has the lead. Why? Because generally polls taken give a reasonable estimate of their percentages. Some will underestimate, and some will overestimate a given candidate's support. If A and B were really tied, then the polls should go both ways; some would overestimate A's support, others would overestimate B's support. If a bunch of polls come out all showing A leading B, then it becomes highly likely that A really does hold the lead, because it's a more plausible scenario than that of every single poll inflating A's support and/or underestimating B's support.
Still not convinced? Think about tossing a coin a bunch of times. You'd expect to get 50% heads, 50% tails, right? But that's not guaranteed. Say you toss a coin 100 times, you could conceivably get 40 heads, 60 tails. That's 40% heads, and it's still plausible. But toss a coin 1000 times, and 40% heads then (400 H, 600 T) is actually not plausible anymore, thanks to the Law of Large Numbers (it's a stats thing).
If that still completely blows your mind, think about this. Break down the 1000 tosses into ten parts: the first 100 tosses, the second 100 tosses, etc. Sure, you may get 40 H, 60 T on one of them. But would you expect to get those kinds of numbers on all 10 of them? No. Some you'd expect to have more heads than tails, just due to sheer randomness. So getting 400 H, 600 T is much more unreasonable, when you think about what must have happened for every 100 tosses.
Anyway, all this is to say we should look at other polling firms too, though SurveyUSA does seem to hit the nail on the head more often than not. Now, unfortunately, Rasmussen doesn't poll all 50 states for a real comparison with SurveyUSA. Oh well. We'll just have to make do with what we have. Here are all the states Rasmussen's polled. I've bolded which Dem does better against McCain in each state.
| State | Dem | Rep | Dem Margin |
|---|
|
| California | Obama 53% | McCain 38% | Win by 15% |
|---|
| Clinton 46% | McCain 39% | Win by 7% |
|
| Colorado | Obama 46% | McCain 39% | Win by 7% |
|---|
| Clinton 35% | McCain 49% | Lose by 14% |
|
| Connecticut | Obama 50% | McCain 38% | Win by 12% |
|---|
| Clinton 47% | McCain 44% | Win by 3% |
|
| Florida | Obama 43% | McCain 47% | Lose by 4% |
|---|
| Clinton 40% | McCain 47% | Lose by 7% |
|
| Iowa | Obama 44% | McCain 41% | Win by 3% |
|---|
| Clinton 37% | McCain 47% | Lose by 10% |
|
| Michigan | Obama 41% | McCain 44% | Lose by 3% |
|---|
| Clinton 43% | McCain 46% | Lose by 3% |
|
| Minnesota | Obama 53% | McCain 38% | Win by 15% |
|---|
| Clinton 42% | McCain 47% | Lose by 5% |
|
| Missouri | Obama 40% | McCain 42% | Lose by 2% |
|---|
| Clinton 42% | McCain 43% | Lose by 1% |
|
| Nevada | Obama 50% | McCain 38% | Win by 12% |
|---|
| Clinton 40% | McCain 49% | Lose by 9% |
|
| New Hampshire | Obama 49% | McCain 36% | Win by 13% |
|---|
| Clinton 43% | McCain 41% | Win by 2% |
|
| New Jersey | Obama 43% | McCain 45% | Lose by 2% |
|---|
| Clinton 50% | McCain 39% | Win by 11% |
|
| New Mexico | Obama 44% | McCain 44% | Tie |
|---|
| Clinton 38% | McCain 50% | Lose by 12% |
|
| Ohio | Obama 40% | McCain 46% | Lose by 6% |
|---|
| Clinton 40% | McCain 46% | Lose by 6% |
|
| Oregon | Obama 49% | McCain 40% | Win by 9% |
|---|
| Clinton 42% | McCain 45% | Lose by 3% |
|
| Pennsylvania | Obama 43% | McCain 44% | Lose by 1% |
|---|
| Clinton 44% | McCain 46% | Lose by 2% |
|
| South Dakota | Obama 38% | McCain 48% | Lose by 10% |
|---|
| Clinton 38% | McCain 50% | Lose by 12% |
|
| Virginia | Obama 44% | McCain 49% | Lose by 5% |
|---|
| Clinton 41% | McCain 51% | Lose by 10% |
|
| Washington | Obama 44% | McCain 45% | Lose by 1% |
|---|
| Clinton 40% | McCain 48% | Lose by 8% |
|
| Wisconsin | Obama 44% | McCain 43% | Win by 1% |
|---|
| Clinton 38% | McCain 50% | Lose by 12% |
(Hey ct, what's with the "rowspan" attribute not being allowed in the table format? I had wanted the table to look like this.)
And for you visual types, here it is in graphical form, based on the Democratic margin of victory.
In 14 of those states Rasmussen surveyed, Obama outperforms Hillary against McCain. In only 3 states does Hillary do better, and only in one of them does she actually have the lead over McCain, in New Jersey. And no, a Democrat is not going to lose New Jersey, no matter what some Hillary supporters are saying. A "safe" blue state is not suddenly flipping red in 2008.
So let's look at who does better against McCain, starting from Obama's most favorable to Clinton's most favorable. I've also bolded the ones where there's a key Senate race going on.
Colorado, Obama +21
Nevada, Obama +21
Minnesota, Obama +20
Iowa, Obama +13
Wisconsin, Obama +13
New Mexico, Obama +12
Oregon, Obama +12
New Hampshire, Obama +11
Connecticut, Obama +9
California, Obama +8
Washington, Obama +7
Virginia, Obama +5
Florida, Obama +3
South Dakota, Obama +2
Pennsylvania, Obama +1
Michigan, Tie
Ohio, Tie
Missouri, Clinton +1
New Jersey, Clinton +13
Yes, I know, some of you say New Jersey has a key Senate race too. But after Estabrook dropped out, in terms of it being a "key" Senate race, well... it's not. For now.
Now, regarding what I said earlier about the polls agreeing? When do SurveyUSA and Rasmussen agree on who does better against McCain? Well, pretty much, they agree who does better almost every single time.
Colorado, Obama +21 (SUSA: Obama +15)
Nevada, Obama +21 (SUSA: Obama +13)
Minnesota, Obama +20 (SUSA: Obama +3)
Iowa, Obama +13 (SUSA: Obama +14)
Wisconsin, Obama +13 (SUSA: Obama +7)
New Mexico, Obama +12 (SUSA: Obama +7)
Oregon, Obama +12 (SUSA: Obama +13)
New Hampshire, Obama +11 (SUSA: Obama +10)
Connecticut, Obama +9 (SUSA: Obama +11)
California, Obama +8 (SUSA: Obama +1)
Washington, Obama +7 (SUSA: Obama +16)
Virginia, Obama +5 (SUSA: Obama +10)
South Dakota, Obama +2 (SUSA: Obama +8)
Ohio, Tie (SUSA: Tie)
Missouri, Clinton +1 (SUSA: Clinton +2)
New Jersey, Clinton +13 (SUSA: Clinton +5)
Only three states give contradictory data, and with the margins so close, you can't really say that there's contradictory data for Michigan. In the other two, Rasmussen has Obama doing significantly better than Hillary.
Florida, Obama +3 (SUSA: Clinton +11)
Michigan, Tie (SUSA: Obama +1)
Pennsylvania, Obama +1 (SUSA: Clinton +6)
Now, just because Obama or Hillary does better than McCain in those numbers I've listed above doesn't mean they'll actually win those states. Sometimes it's a matter of not losing by as much as the other person.
Strangely enough, both Rasmussen and SurveyUSA show Clinton and Obama performing the same against McCain, though Rasmussen has both of them losing by 6 points to McCain, while SurveyUSA has both of them handily beating him by 10 points! Now that's unexpected, unless you think that Ohioans really don't care which person's the Democratic candidate. But that still doesn't explain going from losing the state by 6 to winning the state by 10.
But for the most part, the two polling firms agree on who does better against McCain, thus making us more sure in stating that that candidate really does indeed have an advantage over the other one. And in terms of possible coattails, it's almost pretty obvious Obama will do a better job. Especially in Colorado and Oregon, where both firms show Obama does considerably better than Clinton against McCain, the coattail factor may be critical in ensuring a Democrat wins those two Senate seats, especially with the polling in those two Senate races not exactly in our favor right now.
Of course, all this just a snapshot in time, and things can change if some revelations come to light in the next couple weeks or months on any of the candidates. But for now, it does seem that the talk about Obama being a map-changer is justified by the polling data, from both SurveyUSA and Rasmussen.
Permalink | 48 comments