Trust The Polls, Not the Pollsters…

I figured I would take some time during this election season slowdown (some were hopeful for an Obama VP announcement today, however, this looks to be another successful headfake) I’d get into a meatier subject everyone should consider when digesting the media spew and trying to make sense of this election so far.The bit of civic knowledge which, like bison jerky left one week to long in the dehydrator, is so hard for the American media to bite off and chew? That when it comes to the polls the media rarely have their head and their ass far away from each other.

I’m a fairly serious reader of the polls. I’m not a serious reader of the polls because I’m an obsessive follower of the “American opinion” but because I simply can not find many dependable poll analysises. I don’t go a day without checking the RCP Poll Average, Pollster.com, and Fivethirtyeight’s averages. Just doing this much will give you a clear sense why any pundit, blogger or talking head who cites one poll alone most likely doesn’t know what in the hell they are talking about.

Here’s a primary example: a USA Today/Gallup Poll shows McCain with a +4 lead on Obama. Heaven to Betsy, the sky is falling, a complete flop, the American people have abandoned the candidate of change for old reliable! Not really. A Pew poll released the same day showed Obama at +5. A CNN poll the next day showed Obama at +7. Thats why averages, and especially weighted averages like those used by fivethirtyeight, are particularly valuable. To highlight my point, here’s a screenshot of RCP’s poll averages:

As you can see, outside of one outlier, Obama has maintained a healthy lead in the polls, and the RCP poll tracker has consistently posted a lead between 3 to 6 points ever since the primaries. The instability and unreliability of a single poll can also be demonstrated by how wide a variance one poll will often have over time. Take these charts put together by a Professor at Emory, showing Gallup’s poll results on a daily basis since the primaries ended:

and the poll averages according to ten day groupings:

So, on a daily basis the poll news from Gallup can be either exciting or dismal for Obama supporters, but over the long-term Gallup has also maintained the same overall 4 point lead for Obama. Keep in mind too the “Al Gore” lesson in that national tracking numbers actually mean less that the state by state polls. While in general national tracking numbers are a decent indication of the national mood, two candidates in a close election by the national numbers can be close in the electoral college, or win or lose by a landslide, assuming a few large states is where one candidate picks up most of their popular votes; roughly how Hilary was only able to win around half the number of states as Barack Obama but score 18 million votes. However, this is only one small way polls can be misused.

For example, that poll that a plurality of Americans support offshore drilling as an energy solution at the pump is a little misleading. In order to understand why Americans may have answered that way, you have to look at the question and answers they were given, in this case:

What’s the more important priority for US energy policy today?

(1) Expand exploration, mining/drilling, construction of new power plants

(2) More energy conservation/regulation

(3) Don’t know

Some other possible major priorities that may be missing: alternative energies, regulating oil speculation, anyone? In fact, another poll did take the time to ask whether Americans preferred more drilling/exploration or investment in new technologies, to which an overwhelming 76% chose investing in new technologies. The previous poll is not, then, an indicator of America’s preference for drilling, but just that they prefer drilling to at-home energy conservation solutions.

Polls are only as good as their questions and answers, so without seeing those, throw the poll out the window.

Polls too only exist as snapshots of certain periods of time, usually 4 days or so prior to their being released. So, reality number one is that they don’t show present preference for a candidate, but past preference and they make fairly poor future indicators of future preferences. The best indicator of that is the favorability numbers. For example, even though they all won’t vote for them, a little more than 54% of voters have a favorable view of Barack Obama and 52% do of John McCain. Meanwhile, 34.5% of voters view McCain unfavorably and 34.2% view Obama unfavorably.

What do those numbers mean? Its those numbers, and how they sit now within them, that will have the greater effect on future voting. People who view a candidate unfavorably aren’t generally likely to ever vote them under future conditions, they’ll just stay home. As both candidates enjoy over 50% favorability favorability obviously doesn’t equal votes, but without numbers over 50% the candidate doesn’t have a prayer of winning. Under these numbers too, at least 30% of voters haven’t made up their minds, mostly low info voters who will decide late in the election, and who are mostly unreachable for now.

Another case where the media can’t seem to find their collective sense of worth is anything that may be sensationalizing failure in the polls. We already discussed the problem of “one poll usage,” another popular complaint has been with the “why isn’t Obama winning by more?” argument. Marc Ambinder blows this argument out of the water.

Partisanship. The electorate’s jitterbugging is all confined to a fairly narrow circle, where Democratic zigs are evenly matched by partisan zags. Even when one party hits a dry patch, the other party’s fertility is constrained. If we, in 2000, were a 50.5% Democratic country, in 2002 a 51.% Republican country in terms of the two-party congressional vote, and in 2004, a 50.7% Republican country, and in 2006, a 52% Democratic country in terms of the two-party congressional vote, we would not — should not — expect 2008 to be a year where Democrats capture 54% of the presidential vote. Indeed, given the past few cycles, if Obama recieved roughly 51.5% of the vote to McCain’s 47 or 48%, it would be a post-dated blow-out. Maybe Democrats expect too much.

snip…

Bambiness.  Obama doesn’t counterpunch effectively, and to the extent that baseline opinions about character are being formed now, bruises that were just about to fully heal from the primaries are blushing again. Put it another way, just as the McCain campaign is offended by the rank inexperience and presumptuous of the Obama campaign and aren’t bottling up their feelings, the Obama campaign is offended by the rank immaturity and stupidity of the McCain campaign and aren’t bottling up their feelings.

Gas prices and the Iraq war. McCain strategists credit McCain’s “Dr. No” campaign and the success of the surge as reasons for weakly partisan Republicans starting to get more excited about the McCain campaign.

Information.  Right now, the type of voter who’s paying attention is primed to support John McCain. After the conventions, when younger voters typically tune in — and by younger here, I mean, under 55 or so — then Obama’s margins will widen because these folks are his folks.

You won’t hear about these very real and fair reasons for why Obama isn’t winning by more. There is also the fact that Obama is outperforming John Kerry at this point in nearly every demographic. A case when relatively close numbers are given their real values, which any honest report would say that the numbers go about 48.5 to 51.5 with only a portion of the Obama electorate even polling.

Nor are the polls often used when they could be most helpful. Last week Pew had a poll to gauge the level of interest and favorability for the Obama and McCain cap’s new round of advertising, with a special focus on McCain’s new negative attacks. I, and many other people, had pointed out that the long-term effect of this kind of advertising was negative to the McCain camp, even if in the short term it gains him some ground in the polls by energizing his base. Hence why most negative campaigns come weeks before the election.

Sure enough, Pew backs it up (and I realize this is a one poll situation, however its the only data we have and the numbers spread is fairly far apart):

Most of those who are aware of Obama’s commercials say they are mostly positive messages about the candidate (38%), while fewer (13%) characterize them as negative messages about McCain. The balance of opinion about McCain’s commercials is the opposite – a plurality (31%) sees them as negative messages about his opponent, with fewer (19%) describing them as positive ads.

Which may explain this number:

a slight plurality reports that recently they have come to have a less favorable view of McCain rather than a more favorable view of him – (23% to 18%).

Yet, we spent the whole week hearing about how the Obama camp was “unable” to fight back against the celebrity and “the one” ads. As it turns out, as always with the Obama camp, staying on message and waiting out the short term effects of the negativity was the best they could do.

This is a rather low information post, as the real effect of these examples will wear off in short order, however its just some more examples of how little you can trust the media to give a valuable, and honest opinion. It also shows the reason for my skeptism of the self-righteous one-poll partisans who wave it around like evidence of their inherit superiority. I have something for those people:

Guess what kiddos, the media, despite their billions and millions in paid experts and direct source information, the media has no monopoly on the truth. Nor do they have much interest in it. Theirs is a business and they do what gets you, chicken little Democrats and reactionary Republicans, more hopping and steaming mad about the electoral process in an effort to get you to watch more.

So unplug your damn TV’s, and go to the source of the news and think for yourself.

Good eve’ning.

-Marc-