Guidelines for Surveys
A survey can produce quantitative data that can be easily analyzed. There are many types of questions that can be included in a survey, such as:
- multiple choice questions
- ranking questions
- Scale questions: "I think that coke is: -very good, -good, -not good, -awful."
- ‘Yes’ ‘No’ questions.
- Fill in the blank, or open ended questions, which are difficult to analyze, but can be very useful in eliciting information you hadn’t thought to ask about.
Different types of questions collect different data. This should be thought out before the questions are constructed. The order in which the questions are placed can also be important as you don’t want one question to influence the next. Questions must be easy to answer and must not be biased. Likewise, it is crucial to have multiple streams of evidence so that you are not relying on only one indicator for an important concept in your survey.
- It is important to have an objective. What are you trying to learn, and what are you going to do with the results you get? The ideal objective is narrow in scope and can be clearly stated.
- Each question should have a hypothesis behind it (or- there should be a very good reason to ask that specific question).
- Questions must be worded simply and must be very easy to answer. The respondent should not have to waste any time wondering what a certain question is asking or what it means.
- Do not use double barreled questions (questions that ask 2 things in one question). Example: Do you think that Santa Clara University is prestigious and beautiful? This question will produce unreliable data because someone may agree with the first part, but not with the second part, and therefore will not know how to answer the question.
- Wording must be precise (if the wording is too general, different respondents will read the question in different ways.)
- Open ended questions (questions which don’t provide specific choices) produce qualitative results that are very difficult to analyze.
- Questions must not be biased. Your question should not lead a respondent to any specific answer.
- If a question is too personal, the respondent usually will not answer it.
- A survey should be kept as short as possible. Always ask yourself, "Can I eliminate any of these questions."
- For a survey that is distributed via email…the ‘subject’ line of the email is extremely important and will often determine whether or not a respondent opens the email and takes the survey.
- You will get a better response to your survey if an incentive is offered for filling out the survey, such as a drawing for a prize/gift certificate.
- A survey must be thoroughly tested and proof-read. Ideally, you want to test your survey on a test group from your target audience.
- For an online survey, the maximum time the survey should take to fill out is 5 minutes.