Questionnaire design
From ICE Primer: A Tobacco Control Research Methodology Primer
Ideally, the process of questionnaire design begins with the logic model and the analytic plan. Each of the constructs and variables in the plan must be measured by carefully designed questions.
Concomitant aims of questionnaire design are to maintain high response rates and high engagement of the respondent to encourage thoughtful responses. These aims require that the questionnaire be clear and logical, appealing and interesting, and not unduly long. Memory aids should be incorporated whenever possible and consistent with the research aims.
The use of questionnaires and interviews has inherent limitations. Some variables and constructs, such as future use of a facility, or attitude ten years previously, might be too difficult to measure through questioning. Other considerations which limit what might be asked have to do with ethical concerns, respondent burden, language and literacy.
Research questionnaires are typically designed by a team of subject matter specialists, each with specific aims, and methodology experts. The construction of a questionnaire may require several meetings and teleconferences, supplemented with email exchanges. It is helpful to have a site for posting the most recent version and a history of changes.
Often, suitable sets questions with known validity and reliability have already been developed by other studies. The use of these, adapted if necessary, can increase the relevance and comparability of the survey results.
(Link to data base of questions in common use in tobacco surveys.)
Special attention should be paid to the introduction or introductory script, especially where this is the first approach to the respondent. The introduction will inform the respondent about the survey, determine eligibility (in the screener), and perhaps route the respondent to appropriate parts of the questionnaire.
It is frequently the case that once a questionnaire is assembled to incorporate the aims of all members of the team, it is necessary to shorten it to make it of reasonable length for the respondents.
Questionnaires may be self-administered, on paper or on the web. Alternatively, they may be interviewer-administered, through face-to-face or telephone interviews. In computer assisted data collection, the questionnaire is programmed using questionnaire-authoring software, so that answers can be entered as questions appear on a screen, and collected automatically in a data file. Common acronyms are CATI (computer assisted telephone interviewing), CAPI (computer assisted personal interviewing), and CASI (computer assisted xxx interviewing).
Programming of questionnaires for CAPI, CATI or CASI administration allows the possibility of smart questions, which can be tailored to the respondent's previous answers in the same survey, or possibly in a previous wave.
Questionnaires should be tested carefully for logic and clarity. This can be done systematically by team members responding under an exhaustive set of scenarios, so that complex routing can be verified.
Questionnaires should be pre-tested using a cognitive interviewing process, with participants who are like the eventual respondents in terms of interests and understanding.
Translation of the questionnaires into multiple languages presents special challenges....
