This is a follow-up to today’s post on how nooze is either ignored or manufactured. In the middle of primary season, what could be better than the manufactured nooze of polling?

Tomorrow’s Republican primary in Michigan is a very big deal, because a poor finish will likely end Mitt Romney’s candidacy. It also could validate McCain’s big win in New Hampshire. For these reasons, the press is following the horserace closely. The best way to do that is, as far as anyone can tell, is through polls. What is striking is how the polls look when you put them all together:

Candidate DFP DN MSNBC Zogby ARG Avg
McCain 22 27 22 27 34 26
Romney 27 26 30 24 27 27
Huckabee 16 19 17 15 15 16
Und 16 10 11 9 6
DFP = Detroit Free Press, DN = Detroit News. Links to these polls are here:
http://www.freep.com/apps/pbcs.dll/article?AID=/20080112/NEWS06/80112066
http://www.detnews.com/apps/pbcs.dll/article?AID=/20080112/POLITICS/801120436
http://www.kansascity.com/153/story/442508.html
http://www.zogby.com/news/ReadNews.dbm?ID=1420
http://americanresearchgroup.com/pres08/mirep8-705.html

How can anyone make sense of all of this? If Romney is ahead by 8 in one poll and McCain ahead by 7 in another, what is the point of these polls? That’s why I included an average of all of them – the most common way that media outlets who aggregate all of this information try to report it. That shows a net very tight race with an advantage to Romney.

The problem with taking an average is that some of these polls are rather stupid, which if you follow the links and read the fine print at the end you can see for yourself. The MSNBC poll appears to have been done by dialing random telephone numbers and asking people if they were likely to vote until they had 600 responses; no attempt was made to compare their voting history with the declaration of being a likely voter, or even if they were registered.

The Detroit papers did a bit better, generally calling registered voters until they hit the magic 600 considered kosher for a poll. ARG decided that in this open primary they’d interview 500 Republicans and 100 Independents, nevermind that the results came out different enough that it’s clearly important to discover if their magic 5:1 ratio is the right one to use; they didn’t bother to check that. Which gets me to Zogby.

I’m a follower of John Zogby, which causes both Democrats and Statisticians to consider me either a heretic or an idiot, often both. Zogby asks a bunch of “cross-tab” questions of his respondents so that he can adjust the date he collects to be as representative as possible. In this poll, he adjusted his numbers separating out the 915 respondents by party, age, and gender. If he knows that his pool of respondents has fewer women, for example, than the general population, he weights their responses more heavily.

You’ll note that Zogby has a narrow 3 point lead for McCain, which is inside the margin of error. But I think we have to go with that, for now. The rest of the polls? Rather than call the horserace, they seem to have spent their time on horseshit. I don’t get it. But in the drive to manufacture nooze they had to do something. The result is confusion and nonsense.

In the feast or famine of reporting, the primary season is one of the most incredible feasts of all. Think of it like a giant carnival, a State Fair of greasy food to engorge yourself on. Naturally, like the State Fair, a lot of the feast at hand is far more like to make you sick than provide any decent nutrition. It’s probably best to just stay home and get what you need on your own, just like any other media frenzy.