Often a condition for people taking part in your survey is that the information they provide be treated in confidence or that it is provided anonymously. The distinction is subtle but if you want to stand by any assurance you make then you should be aware of the difference and how it applies in the online research world.

If you need to run your survey truly 100% anonymously then any survey invitations sent by email should only use a generic survey link. In other words the link sent out must be the same for everyone in the sample-frame. Be aware that if the services of a panel company are being used its impossible to run a truly anonymous study as the information must exist, albeit in different places, that could theoretically be used to tie the respondent’s identity to the answers they provide. This has to be the case otherwise the panel company would not know who to provide the appropriate rewards to.

This is of course largely academic because when we provide anonymity in terms of research we are really promising that the ultimate users of the research findings will not know who has said what. However, in a court of law (as an extreme if somewhat unlikely example) it could be shown that, strictly speaking, anonymity has not been adhered to and for some companies out there this might be an important consideration. The best and easiest solution in these cases is to recruit people to the research on the basis of ‘confidentiality’ rather than ‘anonymity’; it may dissuade a few respondents but does ensure 100% accuracy and honesty.

If generic survey links are used in the survey invitation emails then be aware that there may be a few issues with the administration of the survey. There will be nothing to stop people clicking on the link and making multiple entries as different respondents and it becomes difficult to provide people links to partially completed surveys if they need them.

There are some techniques that can be used to help deal with multiple submissions. This includes placing anonymous tokens on respondent’s machines in the form of cookies and logging these in the survey data. IP addresses can be logged in the same way. This can help but is not foolproof as people can use different PCs or simply reset their browser cookies.

The ORS service allows surveys to be run in a mode that can tie responses back to email addresses. This makes administration easier and combats multiple submissions. The system can be set to not allow users, (ie. those using the reporting facilities) to be able to access those relationships for particular studies. This level of anonymity is usually acceptable for most surveys.