Survey Design Tips
- Define specific goals for the survey.
- Only include questions that directly address those goals.
- Consider options to increase respondent participation, including advance messages, incentives, and reminders.
- Use question types that support the analysis that you will be performing and the kind of results you wish to report.
- When selecting question types, consider the time involved in the analysis stage; for example, coding and evaluating open-ended items.
- Select respondent samples that are representative of the population.and who have the knowledge to answer the questions.
- Pilot-test the survey with a small number of people to identify problems in question wording and instructions; remedy the problems before sending the survey to a large group
- Write an introduction that explains the purpose of the questionnaire, explains confidentiality issues, and includes the due date.
- At the conclusion of the survey, include a thank you and (if appropriate) information about how results can be accessed.
- Place the quickly and easily answered questions at the beginning of the questionnaire. Difficult and/or sensitive questions should be placed toward the end of the questionnaire. Otherwise, potential respondents might assume the entire survey is composed of difficult and/or sensitive questions, which could be a disincentive to participate.
- To encourage a large number of respondents, keep the survey as short and concise as practical.
- Group related questions or questions of a given response type in sections and arrange in a logical order.
- Look for possible order bias (the order in which questions are asked may affect the answers). Choose randomize answers, if appropriate.
- Write questions as clearly as possible. Write for the intended audience (consider their vocabulary and grammar levels and styles). Use simple, everyday language that all respondents will understand - jargon-free, without technical language, slang or culturally specific words. Avoid complex sentence structures.
- Define any terms that you feel may be unclear or not obvious to your audience.
- Avoid asking leading or potentially biased questions.
- Make questions as specific and concrete as possible; i.e., instead of "Do you read regularly?" use "Do you read the Washington Post five or more days per week?"
- Avoid "double-barreled" questions: make sure each question addresses only one issue, attribute, or skill.
- Give respondents the option of "I don't know" or "Other" as a choice, unless you have specific reasons for forcing them to make a choice of responses.
- Consider, for each question, the necessary background information that is required for a thoughtful response. If you use multiple-choice questions, check to see that all possibilities are addressed in multiple choice answers and that each answer is mutually exclusive and neutrally phrased.
- If you use multiple-choice questions, check that the answers are approximately the same length and complexity.
- If you use true/false questions, check that each response option is true or false without exception.
- With all types of questions, avoid determiners "always, never, without a doubt, invariably".
- With all types of questions, avoid negatives and double negatives.
- If you are planning to report or publish your results, document your survey construction, administration, and analysis procedures in enough detail that someone else could replicate them.
- In any report of your data, make sure you are conforming to the level of confidentiality that you have promised to your respondents.