One thing we’ve been asked about a few times is how to have surveys that are anonymous, but where a respondent can be identified across multiple surveys to track progress towards an outcome or changes in opinion over time.
This is challenging, but can be achieved via a couple of methods, depending on the preferences of the survey creator and also the anticipated needs of the respondents.
Two surveys, with an external shared ID.
One use case for this requirement is the idea of an anonymous feedback survey for sort of lesson or training course. The respondent is asked questions at the before the course, and then again afterwards – we want their responses to be linked to each other, but not identifiable as a particular person
For this method, you need some method of randomly assigning an identifier to each respondent – a book of raffle tickets is ideal. These are distributed to the participants at the start of the day, and they enter the number on the ticket (via a survey question) each time they complete a survey.
The survey link can be distributed however you like, both an anonymous contact list or a direct tracking link (or QR code) will work.
The advantages of this system is that it’s very simple to do, and easy for respondents to understand how their anonymity is retained. The two disadvantages are that the responses are collected via separate surveys so cannot be reported on combined, unless the data is exported. The second disadvantage is of course the need to obtain, or create, the ID tickets / tokens.
It can be argued that this method is pseudonymous, rather than anonymous, as the respondent is given a randomised name, so if you're in a very sensitive field, make sure you're aware of this. It's unlikely to be an issue outside surveys that are extremely in-depth.
Single Survey with save-and-continue and a blocking question.
One of the disadvantages of the above method is the separation of the survey into two distinct surveys, which as stated can raise some issues with reporting. By using the “Save and Continue” and Answer validation features, we can instead make a multi-stage survey and separately control respondents’ access to each stage.
This is fairly simple to implement. After ensuring that the survey is set to anonymous, activate the Save-and-continue feature.
This feature does require the respondent to enter an email address, which is connected to the partial response that they’ve entered, but this information is not available to you via any reports (or anywhere else in your account), so anonymity is retained.
To break the survey up into sections, and ensure that respondents can’t proceed any further in the survey than you want, you can use a mandatory free-text question as a way of password-protecting the later stages of the survey.
To do this, create a page on your survey between your two sections.
- On this page, add a “Single Text Box” question.
- Tick “Require an answer to this question”.
- Tick “Validate answer format”
- Chose format as “Number” and in the “from” and “to” boxes, enter a number of your choosing – the same number in both boxes.
- Save the question.
What this means is that this question will now stop respondents progressing in the survey unless they enter the exact number in the validation boxes. Effectively this number acts as a passcode.
So, by separately controlling when this passcode is revealed to respondents, we can control how far the respondents can progress through the survey at any point in time. So, to return to our example of a feedback survey for a course or seminar, then respondents can be distributed the link at the start of the course, they can complete the first stage, get as far as the passcode question, receive a link via save-and-continue, then be given the passcode at the end of the course, then continue to complete the survey.
There is no limit to the number of stages that you can split a survey into using this method, or the time delay between a respondent starting and finishing the survey, so this method could be used for long-term studies, tracking respondents across weeks, months, or longer.
The main disadvantages to this method is that it does require the respondent to enter an email address (some may be reluctant to do so, even with assurances on anonymity) and to retain the email they’re sent with the “return” link in it. But it does remove the need for any external apparatus and opens up some very useful possibilities.