I began writing this basically as a pre-emptive strike (and also, hopefully--and more importantly-- encouragement) to the folks who are going to get worked up over the Time poll which shows Bush up 50-43 and with 53% job approval--but it turned out to be more of a general post on my sense of where the race is right now.
I'm no saying the Time poll is good news, but as with any single poll, it's not really worth much despair (or exuberance). One of the things that I like to do with a poll is see how consistent the movement in the internals is; that is, did job approval, favorables, move with the horserace numbers. In many scenarios, like after a convention or strong debste performance, you expect to see this. But when you've got a poll like Time, that polls every week, and you've had a week that was lacking in dramatic events, that kind of movement puts up a red flag for me. More below:
In this instance, Bush's job approval went up from 49 to 53 over the course of a week; we've not seen anything like that kind of movement in any of the other polls, and it looks off just based on common sense alone; it wasn't a good week for Bush's performance as president, irrespective of whatever kind of week Kerry had. I'm not saying that folks answer the approval question totally independent of their horserace preferences-but my general rule on polling is that if you see basically everything moving together all at once and there's no apparent cause (like a convention, debate), it generally just means that it's random sampling varation. Now often, this is a result of undersampling, but it doesn't have to be. Even if Dems are properly sampled, it's still perfectly reasonable (and statistically to be expected) that sometimes you just get a group that's more Pro-Bush than the last one.
At this stage, it really is useless to get worried about any individual poll. It's all about the totality and trend (and quite frankly, given how little the pollsters know about the size and shape of the pending electorate, I don't even know if that's worth much) and I think that what can be said with some certainty is the following:
1. Bush is a deeply troubled incument, whose average horserace are significantly below 50 and whose job approval ratings are consistently under 50. A look at the recent polling on job approval:
First, the Outliers:
A) Rasmussen--54
B) Time--53
Borderline:
A) Gallup--51
And the rest:
A) AP--47
B) Dem Corp--48
C) Pew-44
D) WSJ/NBC--49
E) CBS-44
F) Fox-49
G) Newsweek-47
H) Marist--49
- Bush has a very slight lead amongst the folks who have both decided and are being polled--and please note the distinction between that and folks who have both decided but aren't being reached by pollsters--I'm not going to try and quantify the distinction b/c its unknown, but there is a difference.
- The aforementioned slight lead puts Bush's national average horserace numbers well below 50 and hence lands George squarely in no man's land at best and more likely places him in an utterly precarious position that he's unlikely to recover from.
- While the undecideds historically break strongly against the incumbent, we don't want to take anything as a given--but all of the recent polling evidence on the undecideds' leanings (largely done by Zogby and AP/Ispos) suggest that history will hold true. Zogby's most recent measurement, I believe, found that only 16% of the undecideds in his poll believe that Bush deserves another term. And as Charlie Cook always points out, the wrong track numbers amongst undecideds in the AP poll throughout the cycle have been off the charts.
- The battleground, in my view, is basically shaping up as a tiered system (and I haven't forgotten PA--it's safe, and Bush has largely ceded it)
*Primary tier (most likely to decide the election: Ohio, Florida, Wisconsin, Iowa. As I note below, a Kerry victory in Ohio looks increasingly likely--though certainly not to be taken for granted until the votes are counted. If Kerry is able to make the pickup, then the result will almost certainly be determined by Kerry's ability to hold either Iowa or Wisconsin. Of the two, WI is obviously the preferable hold (I think we'll hold both, but that's beside the point for now). In my view, losing WI but keeping IA (as opposed to the opposite) is damaging in only one situation: Kerry takes OH and NH, while Bush takes FL, WI, and manages to swing New Mexico. Then you're looking at 269-269. Obviously, if you switch WI and IA in that scenario, the NM swing wouldn't be decisive.
*Secondary Tier--New Mexico, New Hampshire, Nevada, West Virginia, Maine 1st CD, Colorado; obviously there are tons of scenarios here. Based on the current polling, I can't say its likely that Kerry will swing NH, WV, and NV together; but the effort needs to be made to the extent possible so that there's still a chance of victory if Bush holds OH, FL and CO (and I think he'll hold Colorado, but not by much, and it could swing).
*Not to be ignored--Pennsylvania, Minnesota, Oregon, Virginia, Arkansas. I think that the first three are nearly certainly going Kerry, but they obviously are critical and can't be ignored; the latter two are unlikely Kerry pickups, but reasonable enough that they should be fought for to the extent possible, simply to create alternative and backup winning scenarios.
6. In the four primary battlegrounds-and please be sure to see thirdparty's outstanding diary here:
http://www.dailykos.com/story/2004/10/22/125459/43
A) Bush is in desperate trouble in Ohio. My gut sense is that the campaign is on the verge of ceding it--not publicly of course, but some resources are going to be shifted, most likely to WI, IA and NM, and then the Bushies will hope that maybe they can eek out OH by suppresing enough of the vote. All of the polling points to an undeniable Kerry surge in OH that suggests an extremely likely pick-up. As Thirdparty notes, across the most recent 12 Ohio Polls, Bush reaches 50 in zero and reaches 49 only once. Kerry leads in 7 of the 12 polls, and is even once; of the four polls showing Bush ahead, two are Rasmussen, one is Fox News, and the spread is no greater than 2.
B) Bush is also in desperate trouble in FL, though maybe not quite to the same extent as in Ohio. As Thirdparty notes, of the 12 most recent polls, Bush is above 50 once, and that SUSA poll is the oldest of the group and has since been replaced by a newer one that has Bush back down below 50. Moreover, Bush manages to reach 49 only twice, and 48 only twice. So you're talking 8 of 12 with the incumbent below 48, 10 days before the election. Bush has a "lead" in only 6 of the 12 polls, and of those, he leads by 5 only in the aforementioned SUSA poll and by 4 in a single Mason-Dixon poll; the other 4 show an advantage of Bush by anywhere from 1-3.
C) Wisconsin--Again, thanks to Thirdparty; of the last 9 polls, Bush is above 50 exactly once and the culprit is familiar: the Gallup likely voter model. The additional 8 polls show Bush at 48 or below; additionally, Kerry leads in 5 of the 9 polls.
D) Iowa--Obviously, the two most recent polls are discouaring. But taken in context, when looking at Thirdparty's examination at the 8 most recent polls, Bush is above 50 only once; and Kerry leads in the 3 of the 8.
7. Finally, and this may be more opinion than fact; but I really do believe that Kerry supporters are systemically undersampled by telephone polls for many reasons. I'm not saying dramtically, but I do think, given a high-turnout, that it could be of a couple of (potentially decisive) percentage points. A few quick possible reasons for this:
*Cell phones--I don't know it for a fact, but the cell-only crowd is likely to be Kerry-leaning (probably younger, well-educated, urban, open to lifestyle changes, etc.)
*Minorites--You never know whether potential voters who may not speak English well enough to answer a telephone survey (or who simply don't feel confident enough in their English skills to do so) are not being counted. I know that at least one or two polling outfits do have bilingual interviewers, but I'm not sure how widely this applies. There may also be issues with Hispanic and other immigrant families being more likely to share phones, not yet be established in a permanent residence, etc (and please read no racist intent here--I'm sure what I've just said applies to white familes as well, it's just not something I'm personally aware of).
*Lower income voters--the most obvious, of course, is that these folks are the most likely not to have a phone at all. They're probably also the most likely to have multiple jobs, and hence not be around to be polled, or to have time to answer the questions. There may also be some of the issues above of phone-sharing, no permanent residence, etc. Now, sadly, the turnout is usually quite poor amongst this group--but with such an incredible ground game, if an improved percentage of these folks do vote, it'll make a difference that the polling may be missing.
So all in all, I remain optimstic. It won't be a Kerry landslide--I never thought that. But I do think it'll be a Kerry win both in the electoral and popular votes.