Web Surveys 101

Most surveyors are constantly looking for ways to reduce survey bias that naturally creeps into their questionnaires. One way to help reduce bias is to randomize survey question answer choices. This is true regardless of your survey method: paper, telephone, face-to-face interviews or online. For telephone or face-to-face interviews, respondents often will pick the last options given because it is top-of-mind. This is a specific type of survey bias that can be avoided by using randomized answer choices. More importantly though, like with question order, response order could impact how the respondent answers a question. Randomizing the choices can remove this bias in a reasonable survey sample. Quality online survey tools allow surveyors to select randomize answer options when creating a questionnaire.

Keep in mind when you create surveys, not all questions should have a randomized answers:

1. Demographic questions such as age ranges, education levels completed, etc. When there's a natural, accepted order for questions, you don't need to worry about randomizing the order of your survey answers.
    
2. Scaled survey questions such as those with a likert scale. It doesn't make sense to randomize a 1-5 scale so that 1 and 5 could possibly end up next to each other. It would be confusing for your survey respondent, and the survey data would probably be unreliable.
    
3. Long lists of options. Sometimes it's better to sort long lists in some sort of order, such as alphabetically, to reduce frustration and confusion for the respondent. You wouldn't want to randomize a long list of stores in a mall directory because it makes it difficult to quickly find the stores he or she visits most often.