Semi-Daily Journal

Paul Bruno Is Fair and Balanced

The epistemopolitan has a link to Paul Bruno, who is fair and balanced and explains why the press corps--the press corps that trumpeted a Gallup poll as showing that Arnold Schwarzenegger had 42% support in California--is neither.

A Fortiori: Fair & Balanced: Scatterbrain: When the Field Poll putting Cruz Bustamante ahead of Arnold Schwarzenegger came out on the 16th, it was surprising to many people who had been lead to believe that Schwarzenegger had a sizeable lead on all of his opponents. The source of this confusion seems to be this Gallup poll, released on or around the 11th, which was reported in basically the same fashion by all of the mainstream news outlets. As an example, consider USA Today's coverage of the Gallup poll:

If California's recall election were held today, Democrat Gray Davis would be swept out as governor and Republican political rookie Arnold Schwarzenegger would be swept in.

...

Schwarzenegger muscled into a big early lead on the motley list of candidates to step into Davis' job if he is recalled. A near-majority of the voters surveyed — 42% — say there's a very good or good chance they will go for the Hollywood leading man.

42%? Wow, that's a lot, isn't it? Arnold's bubble must have burst between then and the release of the Field Poll, right? I mean, the Field Poll, taken a mere three or four days later, put Schwarzenegger's support at only 22%. This site, which seems comprehensive in other respects, presents a table compiling the data from the 3 major polls taken so far in the race:

I don't mean to sound as if I'm attacking the people who created that table and maintain what appears to be a useful website, but the fact that they used the Gallup poll to calculate the average is illustrative of the confusion over the Gallup poll and what it means. (In this case, the average is simply the mean of the 3 poll results for each candidate.) The reporting on this recall election has been remarkably poor, and the Gallup poll is but one example of an issue or phenomenon that has been covered in such a way as to confuse the public.

Why does the Gallup poll put Schwarzenegger at 42%, while the other polls put him at 25% or less? This table, from Gallup, presents the results of their poll:

Notice that in the first column of percentages - the column referenced by most mainstream news outlets - the numbers don't add up to 100%, they add up to 172%. Why is that? Well, because the Gallup poll didn't ask people who they were most likely to vote for, it asked people whether there was a "very good" or "good" chance they might vote for a given candidate. That means that one poll respondent could be counted in favor of multiple candidates, because it's possible that someone who stands a "good" chance of voting for Schwarzenegger could also stand a "very good" chance of voting for Bustamante.

Now, there are good reasons to frame the question in that fashion, especially given how early in the race the questions were asked. What mainstream news outlets have been sloppy about, though, is giving readers and viewers a solid understanding of why, in a period of a few days, Schwarzenegger's poll numbers appeared to have been cut almost in half. Given the way the media covered this phenomenon, people would be justified in thinking that Schwarzenegger's bubble had very suddenly burst. In reality, though, it looks more like Schwarzenegger's support was never really as high as the Gallup poll implied. When push comes to shove and respondents are asked to pick just one candidate who they would most likely vote for, many people who might consider voting for Schwarzenegger end up picking somebody else.

The press corps is too unwilling to present relatively subtle issues of context to their viewers. Maybe if they stopped wasting so much time assuring us that they are so incredibly "fair and balanced" the press corps would find the time to present us with both facts and context.

By the way, I disagree: there is never a good reason to phrase any question in such a way that percentages add up to more than 100%. The default assumption of everybody, always, is that percentages add up to 100%. If you're interested in conveying information, you don't phrase questions the way Gallup did. (If you are interested in making a media splash, however...)