How to write a survey
From QualtricsWiki
[edit] How to Write a Survey
Most researchers struggle while attempting to create a great survey. An effective survey begins with an understanding of the research process. The first and most important step is proper formulating of the problem/opportunity to be examined. If the problem analysis is set up incorrectly, the survey questions will result in data that is not optimal — or even worse, not actionable.
The best approach for building a survey is outlined in the following steps:
- Review the basic research objectives of the study
- What core problem/opportunity will this survey data help to resolve?
- What actions will take place as a result of the completed survey data?
- Visualize the desired information outcomes
- What will the output reports look like?
- What charts and graphs will be prepared?
- What information is needed to assure that action is warranted?
- Prepare a written list of the topics in steps 1 and 2 and order them according to their value in solving the research problem
- List the most important topics first.
- Revisit items 1 and 2 to make sure the desired objectives, topics and information are appropriate.
- Try to think of the respondent's point of view
- How easy or difficult is it for the respondent to provide information on each topic? If it is difficult (i.e. the respondent doesn't know, can't remember or won't take the time to answer), try brainstorming other ways to obtain the information — perhaps through a different question or another data collection technique.
- How easy or difficult is it for the respondent to provide information on each topic? If it is difficult (i.e. the respondent doesn't know, can't remember or won't take the time to answer), try brainstorming other ways to obtain the information — perhaps through a different question or another data collection technique.
- Review the sequence of topics to make sure they are unbiased.
- Do questions influence or bias the results of the following questions? Oftentimes, providing too much information or disclosing the purpose of the study can create bias.
At this point, the information needed to build an effective survey has been collected. The focus now shifts to asking valid questions and building the survey instrument.
- Do questions influence or bias the results of the following questions? Oftentimes, providing too much information or disclosing the purpose of the study can create bias.
- Determine the type of question that is best suited for each topic by considering the type of data produced by each question type. Do the question and answer formats provide enough robust information to meet your analysis requirements? Take into account the following criteria when determining what question type to use:
- Question type:
- Open-ended text questions, dichotomous, multiple choice, rank order, multiple choice matrix, Likert or Semantic Differential scales, constant sum, conjoint, or side by side
- Answer Format:
- Categorical, ordered, ranking, and rating
- Type of Analysis to be Conducted:
- Percentages, means and standard deviations, cross tabulations, and statistical tests.
- Percentages, means and standard deviations, cross tabulations, and statistical tests.
- Question type:
- Draft the survey questions. Several versions of each question may be needed when building the survey. Creating both great questions and a great survey can often take five or more drafts — be patient and select the questions that provide the most actionable data.
- Review the question sequence again for bias and logical flow.
- Repeat all of the steps above to find any major holes.
- Will each question make sense to the respondent? (Will they understand what is being asked?)
- Will the answers provide you with data that helps/solves your core problem/opportunity?
- Ensure that more than one set of educated eyes reviews the survey before it is distributed.
- Time the length of the survey. A survey should take less than five to ten minutes for a respondent to complete. At about six questions per minute, and depending on question difficulty, a typical survey is limited to about 30-40 questions. When building a survey, remember that one open-ended text question counts for three multiple choice questions.
- Pretest the complete survey draft to about 20 or more people. Obtain detailed feedback — critically looking at their responses to ensure the respondents understand what the questions are asking and are responding accordingly.
- Did they have questions?
- Did they have trouble understanding what information was wanted from the question?
- Did they take a point of view not covered in the given answer choices?
- Revise your questionnaire and pre-test again or begin data collection.