It is not currently possible to randomise the order of questions, answer options or pages, or to randomly assign respondents to different versions of a survey. However, depending on the nature of your research, the following suggestions may allow you to work around this:
Known groups of respondents
If you are surveying a known group of people then you can create several different versions of the survey and use something outside of online surveys (names out of a hat, random number tables etc) to randomly assign respondents to surveys. You can then email the different survey web addresses to respondents as appropriate.
Depending on the number and complexity of your questions, you could make use of a combination of routing, pre-population parameters and survey access control to create all of your different pathways within the one survey and direct your respondents through the relevant pathway.
Unknown respondents
If you are surveying an unknown group then it is harder to randomly assign respondents to surveys but can be done in two ways (each with pros and cons):
- Create all the different versions of a survey but only launch one version. When you have sufficient data close the survey, open another version and reuse the URL from the first survey (repeat as required). This assumes that the respondent’s propensity to do a survey is not skewed by another factor and also assumes you can get enough responses for each survey. There’s nothing to stop you switching surveys back and forth at whatever frequency you want.
- If you can host a web page elsewhere you could use a bit of Javascript to present different survey links to people at random. To do this you would create all the surveys you need and then add the web addresses for each survey to the Javascript code. The online surveys team cannot offer further advice on this technique but the Random Link Generator may help.