Constructing Questionnaire Items

(Go back to the Research Methods page)

OPEN VERSUS CLOSED-ENDED

Open-ended questions allow the respondent to provide a richer assortment of information because the resonses are not structured ahead of time by the researcher. At the same time, open-ended questions may be harder to score and code.

Closed-ended questions provide response categories for the respondent to select from, such as Agree, Disagree, and Undecided. There is little or no room for the person to tailor responses.

Closed-ended items should have response categories that are exhaustive and mutually exclusive.

WORDING OF QUESTIONS

• Items should be clear and unambiguous: The Census Bureau asked about people working "full time," with 35 hours being the cutoff. People answering the question defined it as 40 hours, leading to some confusion about what the results actually meant.

• Questions should not be double-barreled, involving more than one question embedded in a single questionnaire item.

• Avoid questions that include unwarranted assumptions.

• The respondent must be competent to respond. Sometimes specific knowledge or perspective is necessary; an uninformed person may not provide meaningful responses.

• The question must be posed in a way that people will respond to honestly. Avoid the "spiral of silence:" People may not feel like responding truthfully if they think they are in the minority on some opinion; they may respond that they are "undecided."

• Items should be kept short and simple. People are often not willing to spend time and effort trying to understand a question and all its implications.

• Items should not be worded negatively. "Communists should be prohibited from teaching in our schools" is a troublesome question because it involves the word prohibited, a word that people may ignore. Research has revealed that when people agree with this item, in many cases, they are saying that they believe that such people ought to be able to teach.

• Biased and emotional terms are inappropriate. They lead people in particular directions rather than assessing their actual attitudes.

(Go back to the Research Methods page)

(Updated September 2014