Monday, September 14, 2009


You're doing a story on public reaction to some local policy initiative, say a new park planned for the neighborhood. Normally, you go out and interview a few people, ask their views, and compile it into a story. But this time you want to get a more representative sample of public opinion than the views expressed by a few passersby. You do a Web or library search, and discover, alas, that no surveys have been done on this particular issue (call it the Park issue).

That means you have to do your own survey. But you don't have a background in statistics or market research. And you were never really good in those subjects anyway. No problem; it's a local story, so no need to hire a professional polling firm.

Your company allows you to buy, from a list broker, a list of randomly selected neighborhoood people, with their postal or e-mail addresses. (The random selection just means that the list is statistically representative of the community, so you're in good shape.) Your task now is to formulate the written questions. This is the most important part of your research.

Here's the wrong way to write it: "Do you support or oppose the new park planned for this neighborhood?"

Here's the right way: "Please indicate your position on the new park planned for this neighborhood:"

1) I support the park
2) I oppose the park
3) I'm not sure

Here's the wrong way for the next question: "Why do you support or oppose it?" Or: "Why aren't you sure?"

Here's the right way:
"If you support the park, please indicate why. Select as many reasons as apply:"

1) A park adds a much needed open-air resource to our congested neighborhood.
2) We need a safe place for children to play.
3) [Reason 3]
4) [Reason 4, etc.]
5) Other (explain): ______________________________

"If you oppose the park, please indicate why. Select as many reasons as apply:"
1) The park will raises taxes, which we cannot afford.
2) The park will require destroying our already small supply of affordable housing.
3) [Reason 3]
4) [Reason 4, etc.]
5) Other (explain): ______________________________

For those who are not sure, you can call or interview some of them later to discuss their hesitation.

Of course, there are other things to consider in conducting a survey. But the important point here is that written questions, unlike verbal questions, must be very tightly forumulated, with multiple choice options given to the respondents. That's what makes it possible to tabulate. It's what enables you to quantify public opinion: "X percent opposes the park, Y percent supports the park, Z percent are not sure."

And after you get the statistics, you can go back to interview some respondents to elaborate on their views.

No comments: