Monday, January 23, 2012

Developing Survey Questions

A good many of our faculty have used surveys in their classroom. Each time they do, time must be spent constructing the questions that are asked, and there is, as might be expected, a great deal to take into account when writing those questions. Here are issues to consider when writing survey questions.

1. Wording items: Sometimes the answers to your question depend on whose opinions are under consideration. Nardi notes that “if you ask respondents to agree or disagree with ‘Merit raises should be eliminated for all workers” you will get a different answer that if you asked “I feel that merit raises should be eliminated for all workers.” His point is that the first construction is more about a general belief while the second is much more personal and asks the person to consider for him or herself the answer. This suggests that as you write questions, you would mist in an occasional item asked in a different way and compare its results to other similar questions. You should, however, generally stick to writing either with “I” or “You” but not switch later to more general questions. A second suggestion is to avoid negatives in sentences, because for some people they will not know whether agreeing with a negative sentence means they are disagreeing with it.

2. Statement directions; Mix the direction to your statements so that not all the answers for a specific set of opinions lead to “agree or all lead to “disagree.” You should word some questions sot that people must disagree with some and agree with others. An example might be to ask “Staying up late the night before a test helps a lot,” later followed by “Getting a good night’s rest before a test is helpful.” You could not really agree with both of these. Mixing the direction removes what is known as response bias, which occurs when people simply answer most questions the same way by checking, say, “disagree” for all.

3. Always and never: avoid the use of these words; people e rarely always or never feel something about a statement. It is better to phrase such question using choices such as “most of the time” or “infrequently.”

4. Double-barreled items: This is asking a question that actually measures two things at the same time. Such questions often include the word “and.” Consider this: “Do you like ham and eggs?” How do you answer this if you like one, but not the other. Avoid these.

5. Leading questions: You can consciously or unconsciously allow your own personal biases to creep into your survey questions. If you ask “Do you agree that everyone should undergo drug testing on our campus?” you are leading people in a particular direction and suggesting they should agree with you. You should rewrite this as a statement, “Everyone should undergo drug testing on our campus” and include it is a set of questions that have a range of view points.

This is but a small amount of information on proper question development. Taking these issues into account will help give you richer and more meaningful data when you conduct surveys.

1. Nardi PM. Doing survey research: a guide to qualitative methods. Boston, MA: Pearson Education, Inc., 2003

No comments: