Step 1: Questionnaire Design
The design, construction, and testing of any questionnaire is critical to the success of a survey. A well-thought-out questionnaire, which addresses the potential interests of different customer groups can produce very useful information. In addition, poorly worded questions will be difficult, if not impossible, to analyze. Questionnaires should be pretested and revised prior to enumerator training.
A well-designed questionnaire will possess the following characteristics.
Questionnaires should focus only on topics that are necessary for the research the survey is designed to support. While it is certainly tempting to obtain a wide range of information from a survey, it is best to address only topics that are definitely or highly likely of interest to the researcher(s). A technique that can help maintain a tightly focused questionnaire is to work backwards from the outline of the analysis report and draft table shells to the questions required to support the analysis and build the required tables. While analysis results almost invariably lead to the need to construct additional tables to help explain unforeseen features of the data, such an approach provides the discipline necessary to keep the size of the questionnaire in line with the size of the analysis task.
Questionnaires should be of a length and complexity that allows them to be administered in a reasonable amount of time. Obeying the previous rule on questionnaire focus will do the majority of what is required to keep questionnaire length and administration time at reasonable levels. However, even a questionnaire that focuses only on areas that are important to the researcher can be too long. Occasionally difficult choices must be made and questions that are relatively less important must be eliminated. The rules here are rather simple: (1) listen to the respondents during the pretest--they will tell you if the questionnaire is too long; (2) discuss the issue of questionnaire length with enumerators during the enumerator training period--they also will typically not hesitate to say if the questionnaire is too long or too difficult.
Questionnaires should be easy to use. The layout of the questionnaire should be simple, the use of complicated skip patterns should be minimized, the font selected should be readable, it should be clear which responses go with which questions, and questions should be contained within a single page. These principles, while perhaps obvious, are nonetheless important. Questionnaires that are difficult to administer will result in enumerators and respondents that are confused and dissatisfied. This will result in skipped questions, incorrectly entered responses, and, ultimately, data that are of lower quality than is possible with a well-structured, clear questionnaire. Data cleaning will be more difficult, and the results of the data analysis will be of lower quality. Again, listen to respondents and enumerators during training and pretesting.
Questionnaires should contain an opening statement that provides the potential respondent with background information on the survey. Specifically, the opening statement should
- clearly state the purpose and importance of the study,
- explain the importance of respondents and the information that they can provide,
- let the respondent know that his time is valuable and respected,
- address any privacy concerns that potential respondents might naturally have,
- describe the sampling process and how the respondent was selected for participation,
- assure the respondent that there are no “right” or “wrong” responses and that all truthful and accurate responses are acceptable and valuable.
For more information, and more guidelines and ideas on survey questionnaires, refer to the Implementation Manual full text.
