« »

Search Archives:

Custom Search

Wednesday, June 25, 2008

Superstition, Folk Wisdom, BS, and Polling

I'm going to get hooked on these "two headlines" posts if I'm not careful. After all, I just wrote one yesterday. But this one was too good to pass up. The first headline is "Poll: Race not a big concern in election." Cool!

But then we read the second; "Race, Age Could Be Issues For Voters." Bummer.

Here's the fun part; they're both about the same damned poll! The first headline is from ABC's Chicago affiliate, the second from ABC Evening News. Go ahead and bang your head against your keyboard, I'll wait.

So which one's right? Here's a clue; which one's a definitive statement and which one hedges with "could?" ABC Evening News leaves out a snippet of information that would make their "Big trouble ahead!" headline problematic. The Chicago affiliate includes it:

The ABC News, Washington Post poll shows Democrat Barack Obama leading Republican John McCain 48 to 42 percent among adults expecting to vote.

McCain, however, leads Obama 51 percent to 39 among white voters, but analysts say, those numbers are not an indication of any racial motivations among white voters surveyed.

"The net effect, when you size them and put them together, is essentially nil. Obama's support among white voters, at the end of the day, is about the same as average for Democratic presidential candidates over the last eight elections," said ABC News Polling Director Gary Langer.

That's right, the Evening News story leaves out analysis by their own polling director. That's the problem with that TV machine -- most of the time, the stuff on it is designed to get you to watch the TV machine. They're going to give us a "tight race," even if they have to make one up.

People are always saying you can get a poll to say anything you want. Sure, you can ask people if they'd rather vote for John McCain or Osama bin Laden, tally up the results, and write a headline that reads, "Vast majority prefers McCain." But no one does that, because you can't get away with it any more. Modern polling data is factual data. What you can do is get an analyst or pundit to draw inapt comparisons and throw doubt on the findings. CNN, home of the statistical monstrosity they call their "poll of polls," is expert at this.

In "analyzing" a new LA Times/Bloomberg poll, CNN's polling director warns us not to get too excited about Barack Obama's 12-point national lead over John McCain.

CNN Polling Director Keating Holland notes a substantial lead in June does not always lead to a decisive victory the following November.

“Historically speaking, when June polls show a tight race, the race usually remains tight all the way through November. But when June polls have shown a big lead for one candidate, that lead has often melted," Holland said.

"Bill Clinton was leading Bob Dole by up to 19 points in June, 1996; Clinton won by eight. Michael Dukakis had a 14-point lead over George Bush the elder in June, 1988; Bush won by seven. Jimmy Carter was up nearly 20 points in June, 1976 but in November eked out a two-point win. And Richard Nixon managed an even smaller victory in 1968 even though he had a 16-point margin that June," Holland noted.

See what he did there? Out of the four historical examples he gives us, only one actually lost. These are better examples of why Barack Obama will get the confetti drop in November. Winning -- without the help of the Supreme Court -- is pretty damned decisive.

Of course, not all BS comes out of your teevee. Media Matters points us to a blog post by Pittsburgh Tribune-Review columnist Salena Zito which asserts that a Newsweek poll showing Obama with a 15-point lead means Obama's going to lose. She takes Holland's reasoning to an absurd extreme -- Dukakis had a big lead, Obama has a big lead, therefore Obama will lose.

This isn't so much drawing a historical parallel as it is engaging in crazy electoral superstition. I can do the same thing and come to entirely different conclusions -- it's common electoral folk wisdom that the tallest candidate wins (some pundit always brings this one up). Lanky 6'1" Barack is, therefore, a shoo-in over 5'6" Johnny. Of course, Bush won over taller candidate Kerry. In addition, this reasoning would demand that we'd be talking about Stretch Obama vs. Legs Romney right now. But never mind logic, we've got history.

In fact, the tall candidate superstition has more weight than the Dukakis parallel, because the taller candidate's won more than once. All this new "big lead" superstition has going for it is one occurrence. At least, one that's being cited. This shouldn't strike anyone as an extremely convincing argument. In fact, I can go back to find a fairly big lead that stayed big the whole damned way through.

In July of 1984, Reagan led Mondale by 7 points. It just got bigger, until Reagan took 49 of 50 states in the election. Holland would've said the early lead was bad news for Reagan, while Zito would've argued the Gipper was doomed. And no, I'm not going to waste time trying to find out who was taller.

So it's not the polling that's bad, it's the analysis. Especially when that analysis is motivated by right wing ideology -- as it is in Zito's case -- or as a way to increase ratings by pushing a "tight race" narrative -- as it is for Holland and CNN.

The facts are what the facts are. Barack Obama enjoys a big national lead in most polls and all that means is that Barack Obama enjoys a big national lead in most polls. Don't beware the polls, beware the analysis.


Technorati tags: ; ; ; ; ; ; The tells us will lose -- because the moon is full and he broke a mirror once


Anonymous said...

Sorry to disagree, but your analysis has an obvious flaw or two. First, a lead in the popular vote does not ensure victory, since Presidents are not elected by the popular vote. They never had been. Although the Electoral College was seen as some "vast right wing conspiracy" in 2000 to a many naive folks who had never paid attention to the election process, the Electoral College has always elected the President. Therefore, the double digit lead in support by “likely” voters is far less important than the geographic distribution of support. Since neither candidate has a commanding lead in swing states, it’s premature to dismiss this election as essentially over.
Secondly, you’re making the assumption that nothing will happen between now and November to change people’s minds. While the strongly partisan voters will probably not change their minds on either side, we know from recent elections that Independents will play an important role in deciding this election. Although Independents may be favoring the presumptive democratic nominee at this point, four months is a long time. Their allegiance might or might not change. We need to wait for the various news cycles to play out, for the debates to be held, etc, to see how these Independents actually see things in November. Honestly, although lots of interested people claim that they can predict that, I don’t think anybody really knows.
Thirdly, these polls look at “likely” voters. A lot of Obama’s support is among young voters. Although they seem enthusiastic and involved, historically, they haven’t shown up to vote. They claim this election will be different and it may well be. But we’ll have to wait and see.
Predicting the outcome is November requires predicting what the world looks like in November. Who knows that? Oil could be over $200 a barrel. An Israeli-Iranian conflict may have erupted in a Third World War. God willing, neither of these scenarios will actually happen, but they are not outside the realm of possibility and would each exert a different influence on the outcome of the election.
Sorry, friend, but the game isn’t over until three outs in the ninth. We’re not even at the seventh inning stretch yet. There is a reason they play all nine innings in baseball and there is a reason they don’t end the election before the voters actually vote. If they did end elections based on polling, Guiliani and Clinton would have emerged as the nominees last summer. Pundits and Pollsters can all have their say, including you. Voters make the decision.

Wisco said...

Here's a fun idea; why don't you read the damned post before you comment on it?

I didn't make a single prediction, I didn't analyze any polls, I wrote a post about how stupid assholes use polling data to pull crap out of their asses.

It's a big long comment you left, there. It must've taken you some time. Too bad it's all completely off-topic and you wasted your time reacting to your own incomplete reading.

Next time, take the time to read it before you make such a damned fool of yourself.