line -->
  • Monday, November 08, 2004


    Damn Lies and Statistics

    For people who like to spelunk through exit poll data, go to Election 2004.

    This poll covers 13,660 respondents. This is where you can find that often ballyhoo'd information that 22% of the people thought "moral values" was the most important issue. However any poll is also an exercise in the pollster's perceptions. A poll or survey is an attempt to quantify certain things. What it also does is smooth out any edges through the artificial means of limited answers.

    Although some of the answers are factual in response such as the income, the most interesting parts (to me) are the ones related to opinions. So notice in that poll question concerning the "Most Important Issue" has only seven answers. It's multiple choice. The first thing I notice under the "which item is most different from the others" view is that "moral values" is amorphous and ambiguous compared to the other six choices. All the other choices are relatively standard political issues that can be legislated and acted on. The others choices were: Taxes (5%), Education (4%), Iraq (15%), Terrorism (19%), Economy/Jobs, (20%), and Health Care (8%). Terrorism, which also rated fairly high, is a slightly less ambiguous issue but can be concretely acted on. Since 9/11/2001, I'm not aware of a single citizen of the US lost to terrorism. I'm not counting Iraq which I see as a separate issue since we, um, thrust ourselves into the situation, so to speak.

    What if your most important issue wasn't on this list? Well, too bad, choose one! Suppose I think stopping corporate crime is most important? Guess I better pick Economy/Jobs, huh? Suppose I think the environment is most important? Out of luck. Better pick your second most important issue and say it's most important anyway. By picking the issues, the poll selects the only legitimate and appropriate answers. I have to admit that the polling organizations have some experience at this; the list is not just pulled out of hat. At least, I don't think so.

    There are plenty of holes in these results that I can see. Yes, they do tell us something about the people at who voted but I suggest reading them critically. Ask yourself the questions but don't confine yourself to the answers they give. Do you fall within the scope of the question? Are you outside of the question? Do any of the answers fit you?

    << Home

    This page is powered by Blogger. Isn't yours?