I am writing an online survey and I am wondering if there are any good techniques for allowing anonymous people to go to the survey and participate and also prevent duplicates.

I have considered the following, but there are drawbacks from each:

• Use cookie in browser
• Record IP address
• Compare answers for similarities along with either/or the first two methods

Of the techniques I have considered, you either prevent multiple people from using the same device, or make it easy for a user to duplicate survey results. Hopefully someone has an excellent way to prevent this :)

• browser
• cookies
• duplicates
• anonymous
• survey

Well I am not sure whether you are thinking of deliberate or accidental duplicates?

• if you think people will want to post a load of results to skew the survey, I cannot add anything because any ID-related question you ask can be falsely answered.
• if you want people to just give their answers without having to go through a login process, how about asking for their initials plus birthdate (ddmmyyyyfl)- that has a pretty good chance of being unique without really compromising their identities or taking too much time.

Was that what you were after?

Ed

I am currently investigating a similar scenario.

Some of the suggestions I found online are:

1. You generate a unique URL - which you can send to their email (this email does not have to be stored), and then you add a checksum to the URL to verify it is valid.
2. Similar to the above mentioned, you provide them with an uniquely generated password, and you validate whether the password has been used before.

The clear limitation is that you require their email, and this is slightly lengthy. However, the email address is not associated with the answer set.

• Meaning, you can validate whether an email address has been used to send a URL/Password to. Prevents same email address from being used over and over.
• Then when the URL/Password is used, you validate whether that unique reference has been used in an answer set before. (The answer set is associated with the Unique Reference, and not the person's email - ensuring anonymity).

The problem with using email, ID numbers, and Birth Date; is that all of these values can be fabricated.
If this approach is used, also do not forget good old CAPTCHA, as a script can be created to run through the combinations and submit answer sets.

I realize this is an old post, but hopefully it helps someone at some point. All of the best.