Do-it-yourself Surveys: Design Tips

Tools like Survey Monkey and Fluid Survey are low or no-cost, easy to use and accessible to anyone with e-mail.   But this also means there are many more surveys out there as well as other competing demands on people’s attention.   It is more important than ever, to design an effective survey that yields data that you can act on.  


Here are some design tips to help:

1. Keep it simple.  Use straight forward wording in the level of language used by the respondent. Also consider a simple design with limited graphics and fonts.


2. Be clear about your intentions. Respondents are more likely to help you if they see something of positive value for them. So take the time to set out why you are asking for this info; what you do with the data; how you will respect privacy and confidentiality.

3. Define the content.   The best place to start is with defining objectives for the survey. What specifically do you want to know from your respondents?  Then develop questions.

4. Refine the question ‘types’.  There are many types of survey question out there like dichotomous choices (yes/no), ranking, multiple choice, and scales.  What you use will depend on the information you seek.  In general, it is good to have both open and close-ended questions in the survey.  When using close-ended questions, the wisdom out there suggests using scales instead of yes/no, to use scales of 5 or 7 for attitude, and to have a middle value.  It is also seems wise to keep the rating scale consistent across questions. 

5. Keep it short.  Look at your survey and edit out what is not essential.  A survey should ideally take less than 15 minutes to complete.  Otherwise, people get bored and drop out.

6. Pre-test your survey. This will help ensure your questions and response options are understandable and all your survey logic works.


Once you have a survey, now what? It is not quite as simple as sending out a link.... 

A couple issues to think about:

• Participation.  To encourage participation, think of timing issues for your audience.  When would be a good or bad time to be asked to participate in a survey?  It can be useful to let people know in advance that you will send them a survey and to provide incentives to participate.

• Sampling.  If you are sending an e-mail with a request to fill in a survey, you will end up with a ‘convenience sample’.   This is useful for learning about the issues; you just can’t state that it represents everyone.   You could also consider ‘quota sampling’ where you make an effort to ensure that you get responses that represent specific sub-groups that you are interested in (e.g. by gender, or age, or type of client, etc.).   If you have a good budget and/or in-house statistical expertise, than you can use probability sampling to make inferences about the total population.  Or try finding a community partner. For instance many academic institutions have programs in which faculty and/or students partner with community organizations to do this. 

blog type: 
Issues & Ideas