Whether you're in business, politics, or in the academic ivory tower, you will likely use surveys as a tool to achieve your goals. Surveys are designed to draw information from people for several reasons. Sometimes the information will be used to sway a particular group of people, to get them to behave or act in a certain way, such as to buy a new kind of cell phone or to vote for a particular candidate. Very often a company or an organization wants to probe the needs and desires of customers or members so they can better serve them. In academia, surveys are conducted purely for research purposes. No matter what the purpose of the survey, results are only reliable and valid if the questions asked are prepared correctly and carefully.
Keep questions as short as possible. Don't ask, "What is the type of retail store where you or a member of your family are most likely to have bought groceries in the last year?" Instead, ask, "At what type of store do you usually buy groceries? (a) Large general merchandise store (such as Wal-Mart, Target); (b) Supermarket chain store (such as, Safeway, Wegman's); etc.
Make sure everyone who reads the question understands its meaning. Be absolutely clear about the information you want. Don't ask, "How much time do you and your spouse spend watching TV?" Instead, ask, "How much time do you spend watching TV, and how much time does your wife spend watching TV?"
Stick to one topic. Avoid vagueness. The question should point directly to the information that is needed. Don't ask, "What is your favorite breakfast cereal?" Instead, ask, "Which of the following types of breakfast cereal are you likely to purchase for yourself?" (a) Granular cereals; (b) Shredded cereals; etc.
Avoid using sentences that are too confusingly complex. Break up long, complicated sentences into two simpler, shorter sentences. Don't ask, "Do you feel the candidate may have known what he was talking about, but didn't express himself well in his speeches and position papers?" Instead, ask, "Do you feel the candidate's speeches clearly expressed his position? Did his position papers clearly expressed his position?"
Use common words to phrase your questions. Avoid sophisticated or esoteric words that respondents may not understand. Use words every respondent is likely to use in every day speech. Don't ask, "Do you feel the candidate's comprehension of the issue was reflective or superficial?" Instead, ask, "Do you feel the candidate did or did not understand the issue?"
It's easy to introduce bias into questions that will make your survey results less valid and reliable. For example, leading questions point the respondent to one particular answer. ("Do you see a lot of unfairness in President Obama's fiscal policy?") A loaded question, on the other hand, points the respondent to an answer in a way that forces a moral choice rather than provides information. ("Do you believe President Bush's decision to restrict embryonic stem cell research hindered the development of life-saving cures?")