Using Data to Investigate Questions and Solve Problems
Do not ask leading questions.
Asking leading questions can influence the response and not capture the respondent’s true feeling.
Example: “Since your car was repaired and now works great, do you trust the mechanics at Repair Zone?”
This question seems to lead the respondent to respond favorably.
Avoid yes/no questions when possible.
Consider rephrasing a yes/no question to a question that allows the respondent to use a Likert scale. This will permit the respondent an ability to rate the strength of his response while also allowing more powerful analyses by the surveyor or pollster.
Example: “Do you believe that the SAT Prep course offered to students by the District helped your child?”
Consider changing it to…
Example: “ Please rate the helpfulness of the District’s SAT Prep course for your child.”
Neither Unhelpful or Helpful
Be mindful of the wording of the questions.
Some questions can be too wordy or use expert language which can lead to confusion for the respondent.
Example: “In 1854, Stephen Douglas put forth a doctrine of territorial sovereignty regarding slavery, which would be legislated as the Kansas-Nebraska Act with political conflict ensuing. What should occur today?”
All state governments should not obviate the use of the Confederate Army flag.
All state governments should obviate the use of the Confederate Army flag.
This question may not provide accurate information of the feelings of the respondents because they may not be familiar with the Kansas-Nebraska Act. They also may not know the meaning of the word obviate.
Keep surveys brief.
Respondents are more likely to take and/or complete a survey when it takes 5-15 minutes.
Avoid asking repetitive questions.
While repetitive questions can help with measuring reliability and consistency, respondents will tire quickly of this practice. By avoiding this practice, it will also help decrease the time it takes respondents to take your survey.
Avoid “double-barreled” questions.
Do not ask about two things in the same question.
Example: “ Please rate the helpfulness of the District’s SAT Prep course and the Math Tutoring Program.”
Neither Unhelpful or Helpful
It may be that a respondent's view is favorable regarding the SAT Prep course but not favorable for the Math Tutoring Program.
When using the Likert scale, be sure to have a neutral category.
For instance, you may want a respondent to describe their level of satisfaction for a product by selecting 1-5 with 1 being the least favorable and 5 being the most favorable. This implies that selecting 3 would describe indifferent.
Neither unfavorable or favorable
The research is mixed when there is not an indifferent option. Some research suggests that this will cause respondents to answer favorably where other research suggests that respondents will answer negatively.
Not having an indifferent option would look like:
When using scales, make sure to keep the direction of your scales the same.
To avoid confusing the respondent and possibly causing inaccurate information, the survey should not have a question with a response option where
1 is least favorable and 5 is most favorable
then have a question with a response option
where 1 is most favorable and 5 is least favorable.
When using scales, it is good practice to use standard scaling.
If you are using 1-5 scale for a response for one question, the other questions should use 1-5 scale, not a 1-10 scale.
Communicate reasoning for survey.
Let the potential audience for your survey know why you are asking them to complete it and why it is important.
If sensitive or personal data is being requested, let the respondents know why and how it will be used.
Have a few people take your survey ahead of time.
You want to identify possible issues that can be corrected before the launch. This will help keep you from losing credibility with your target audience.
Chronology is essential.
Do not design a survey where it alternates between topics. This hurts the flow of the survey and may confuse respondents, which can lead to respondents not finishing.
Include a choice of N/A for questions when applicable.
A survey may have a question about the effectiveness of a math tutoring program but a student did not participate. Thus, this student cannot answer about the effectiveness.
Beware of satisficing and its role in survey methodology.
Satisficing is a portmanteau of satisfy and suffice. It is a decision-making process to lessen the burden of the intense amount of cognition that a person uses when completing a survey. This can cause issues with the results of a survey.