Weekly Polling Report

We’re going to start a new institution here, in which I break down some of the numbers of the past week (it’ll probably happen every Friday or Saturday) and then put it all together for you in nice, easy to understand, macro-form.

Nationally this week, we’ve got 4 major polls to consider (I’m not going to waste my time with tracking polls); Franklin & Marshall, the LA Times-Bloomberg poll, Newsweek, and Time. Now the first thing to keep in mind is that, because of margin of error, there’s no point in paying attention to composite margins of “victory.” A 12% lead may seem like a lot bigger cushion than 4%, but if you account for a margin of error of, say, 3%, a 4% margin could be as high as 10% in reality. So because relatively small margins of error can still cause drastic changes in margins of victory, there’s absolutely no sense to looking at said margins. The important data is in the cross tabs, but on the meta-level, individual compositie scores can be used, so long as margins of error are factored and the data is aggregated with other sources of information.

For example, if we take these 4 polls each with margins of error of +/-3%, and compile both candidate’s composites accordingly, we can get a pretty good feel for where the race stands nationally this week. Obama scores, respectively, 51%, 42%, 49%, and 47%. McCain comes in at 36%, 36%, 37%, and 43%. So if we take a margin of error of 3% and factor that in, we’re left with this. For Obama:

Newsweek: 48-54%

F&M: 39%-45%

Times-Bloomberg: 46%-52%

Time: 44-50%

So if we aggregate that and look for overlaps, we see that 48%, 49%, and 50% would be withing our range in 3 of the 4 polls, meaning that we can pretty safely estimate Obama’s support at about 48-50% this week. Doing the same for McCain:

Newsweek: 33-39%

F&M: 33-39%

Times-Bloomberg: 34-40%

Time: 40-46%

So for McCain we’ve got slightly more agreement between polls, meaning we have a larger degree of overlap. We could reasonably expect McCain’s support to be anywhere from 34-39%, as all of them fall withing the overlap in at least 3 of the 4 polls.

So to sum it up, aggregating the relevant national data this week, we have:

Obama: 48-50%

McCain: 34-39%.

State Level

More important to campaigns though are state level polling data. This data is crucial to determining resource allocation and messaging as the effective battleground comes into focus, and by that account it was a damn good week for Obama. Quinipiac becomes one in a line of pollsters showing Obama with substantial leads in Minnesota and Wisconsin, putting the at the margins of any list of competitive states right now. This not only shrinks the playing field somewhat, letting Obama put more resources into states McCain will probably win but forcing McCain to spread out his defense, more importantly it takes states Kerry won off the table, meaning that a higher share of the year’s battlegrounds are states Bush won in 2004, where McCain has to play on his heels.

The Q poll, along with PPP, also give Obama somewhat cushioned leads in Michigan as well, which has to be comforting after rampant speculation that the Democratic primary’s Michigan issue, and Obama’s lack of campaigning there, would open the blue-tinted swing state to McCain.

SurveyUSA has 2 polls of consequence out this month, one in Ohio and the other in Missouri. For some reason Missouri is not really being polled. Believe it or not, this is the first poll of the state since the end of the Democratic primary. Nevertheless, McCain saw a slight jump since the last SUSA poll a month ago, re-enforcing the notion that, for whatevery reason, Missouri is trending away from Obama. In Ohio, SUSA essentially has a toss-up, with Obama holding a 2% edge. Again, it re-enforces pre-existing notions, that Ohio is a close toss-up state that can go any way right now.

SUSA also has a toss-up in Indiana, another state that’s not being polled.

And that’s your week in polling.