ERROR: http://www.dailykos.com/i/user/59419/Daily_Kos_Elections_Polling_Wrap.jpg is not from an approved image host. The approved domains are: imageshack.us, photobucket.com, flickr.com, smugmug.com, allyoucanupload.com, yearlykosconvention.org, picturetrail.com, me.com, webshots.com, editgrid.com, coinkitty.net, actblue.com, secure.actblue.com, narcosphere.narconews.com, tinypic.com, images2.dailykos.com, images.dailykos.com, images1.dailykos.com, skitch.com, mac.com, google.com, googleusercontent.com, twitpic.com, public.tableausoftware.com, twimg.com, dropbox.com, imgur.com, yimg.com, imageshack.com, global.ssl.fastly.net, s3.amazonaws.com/dk-production, images-4.dailykos.com.
Please edit your diary to fix the problem.
So, which is the better route to attaining full electoral consciousness: obsessive poring over state polls, or obsessive poring over national polls?
Here is an argument for a more balanced approach:
The problem with only going by state polls, at least at this early juncture in the cycle, is ... well ... there simply aren't enough of them. One of my biggest personal pet peeves as a political observer is seeing tweets from people on my side of the political ledger arguing Obama is doing just fine in a particular battleground state, and citing a piece of data to buttress that point that is five weeks old. Five weeks? Might as well cite a poll from the 2008 campaign, as the predictive value for both of them is roughly similar.
National polls do allow you to retrofit polls a little bit. If Obama is doing a couple of points better now than he did two weeks ago in national polling, I can feel a little better about, say, Ohio, where the last polls there (2-3 weeks ago) showed a fairly close contest. It also allows us to put state polls in context. Is Obama collapsing in Iowa if PPP had the race at 10 points there a while ago and Marist had it close a few weeks ago? Not necessarily, since the proponderance of national polling back when PPP was in the field looked better for Obama than the polling a few weeks ago.
Now, there is a potential issue with national polling, and it is one that PPP's own Tom Jensen alluded to a while back. His argument, which is an interesting one, is that the national numbers might be skewed because Obama could win several battleground states by modest margins, but get absolutely shellacked in some states that, while normally not amenable to Democrats, will at least give Democrats 40-45 percent of the vote. Arkansas is a state that immediately comes to mind in that regard, as does West Virginia.
Therefore, since those states are not behaving as they "typically" would, it will cost Obama a fraction of his national vote. Whether that skew is a point, or three, it will be notable.
So, in the short term, there is no harm in checking out both sets of data, as long as you remember another cardinal rule: only a fool draws firm conclusions about November based on polls in June. They are worth looking at, still, to scope out the ever-evolving set of battleground states, and to judge any trends or tendencies that are developing. But anyone in the "Obama is safe" or "Obama is doomed" camp, based off current data, is alternately prematurely exuberant or prematurely despondent.
In other polling news:
- We usually don't do specific demographic polls in the wrap, but this one merits mentioning. A new Latino Decisions poll has Barack Obama laying out Mitt Romney in several key states among Latino voters. Check out the margins:
Florida: Obama 53, Romney 37Looking at the 2008 exit polls, Obama led among Florida Latinos by 15, so the margins are similar. Virginia's margin was 31 points, the same as this poll. Nevada was actually marginally worse for Obama this time around: he led by 54 point among Latinos in 2008, and leads in this poll by 49 points. Of course, Obama carried Nevada by a dozen points, and few people see a double-digit win for Obama this time around.
Virginia: Obama 59, Romney 28
Colorado: Obama 70, Romney 22
Nevada: Obama 69, Romney 20
Arizona: Obama 74, Romney 18
But, in the other states of the Southwest, the differences are pretty wide. Colorado Latinos went Obama +23 in 2008. Today? Obama leads by 48. If that's true, it's lights out for Mittens in a key battleground state. Arizona Latinos went Obama +15. Four years and SB 1070 later, Obama carries Arizona Latinos by 56 points. If you were wondering why team Obama is targeting the Grand Canyon State, there you go.
- Looking at those polls in the upper half of the post, it is clear that we could see some genuine intraparty fisticuffs in that Democratic primary in Michigan. In addition to the toplines (where, as one might expect in a House primary several weeks out, "undecided" is the big winner), the campaign for Trevor Thomas also did an informed trial heat that they claimed was a "fairly worded" informed ballot. Except that the last sentence says that his opponent, Steven Pestka, admits that he could "easily be a Republican." Team Pestka immediately, and vociferously, protested, and it looks like they have a legit gripe. According to a local reporter, it looks like the quote attributed in the "fairly worded" informed ballot was part of a two-paragraph quote, where Pestka argued it would be easier for him to be a Republican politically, because of terrain of the area, but that he felt a greater kinship with the Democratic Party. If this bit of polling subterfuge is a sign of things to come, this one could go south in a big hurry.