Surveys: some basic rules to avoid bias

Surveys are useful tools to gather information, but when done wrong they can lead to terrible datasets. The video below was played to us at one of our lectures, and is a handy reminder of why surveys need to avoid leading their respondents down a particular path…

With this in mind, I thought I’d follow on from what I learnt from conducting my own survey and put together a list of some basic dos and don’ts to keep in mind when writing a poll or a survey.

Rule 1: Don’t make assumptions about your respondent.

Example: When you are driving, do you ever feel sleepy?

Rule 2: Don’t use one question when you should use two.

Example: Do you think that coffee is stimulating and delicious?

Rule 3: Don’t use technical terms that won’t be understood.

Example:  When you log on to the internet, do you use a fibre-optic or copper wire connection?

Rule 4: Don’t use biased language

Example: Do you think that footballers are greedy and stupid?

Rule 5: Don’t make your questions too vague

Example: Do you like nice things?

Rule 6: Do make your questions clear and easy to understand

Example: Did you enjoy the play? rather than Did you find that the emotional resonances of the play were agreeable?

Rule 7: Do use scales rather than Yes/No when appropriate

Example: How often do you use the library’s wifi? Never/Sometimes/Often

Rule 8: Do give the respondents a range of options and an ‘other’ choice with a text box as well

Example: How useful did you find the talk? Very useful/quite useful/useful/not very useful/not useful

Rule 9: Do make your answer options mutually exclusive

Example: 20-29, 30-39 rather than 20-30, 30-40

Rule 10: Do make sure that your answers match your questions

Example: Should cannabis be legalised? -Yes/No rather than -Always/Often/Sometimes/Never