Friday, October 12, 2007

What's wrong with this poll?

OK, there's nothing wrong with the poll itself. At least nothing in the scanty information that's reported about it. But take a look at what The Swamp, a blog maintained by The Chicago Tribune's Washington bureau, does with national survey results about U.S. Sen. Hilary Clinton, D-N.Y., candidate for the Democratic presidential nomination.

But now, just three months from the premier presidential nominating caucuses in Iowa, Clinton has not only solidified her commanding role in the eyes of Democrats nationally, but also started to erase some of the doubts around the edges of people who say she is too polarizing a figure to win the election.

The latest Oct. 4-7 Gallup Poll on the 2008 election shows that 51 percent of Americans surveyed hold a favorable view of Clinton – and 44 percent hold an unfavorable view. That sort of split might be considered fatal for anyone else. But this is someone who, just last month, had split the public right down the middle – 49 percent favorable, 49 percent unfavorable in September.

The latest measure on Clinton could be a fluke. The 51-44 split stands where it did in the early summer of 2006. Or it could be a sign that she is starting to change some minds.
Or it could be a sign of something else.

Your assignment for extra credit. Read the "AP Stylebook" discussion on polls and surveys (pages 193-95 in my edition), and answer the question: What's wrong with the Trib's analysis of this poll? Hint: Read it all, but read paragraph No. 6 on "sampling error" especially carefully. Post your answer to this blog. If you've joined it as an author, post it as a separate message. (BTW, congratulations, Dave, for being the first to post. And "welcome aboard" to Claire, Jill, Alyssa and Lauren.) If you're still trying to get on and collecting little red-and-yellow error messages, post it as a comment to this message.

I guess it could be argued that readers of "The Swamp" are likely to be political junkies who are used to drawing conclusions about political poles, polls, but I still think somebody's reaching a little too hard for a political angle here. The story is about Al Gore's being awarded the Nobel Prize, but that isn't my problem. What I have a problem with is the way The Trib handled the survey research data here.

1 comment:

Z Kirch said...

Polled in what direction?

According to the AP style book, I would assume there are a few things wrong with the poll.

1)There is no margin of error listed anywhere.

2)It did not say how the poll was conducted, for example it didn't say if it was via the internet, phone, etc.

3)It also doesn't have any sub polls, like for ages of the people or if the people are male or female.

4)Also, it doesn't specify who participated in the poll.

5) It didn't say what questions where actually asked in those poll.