Prevent Multiple Submissions by the Same Respondent

Theoretically, in the standard setting an online questionnaire can be completed by anyone as often as desired. In scientific surveys it is usually not very attractive for the participant to answer an extensive questionnaire several times. However, the situation might be different if the survey contains a vote or if there is an incentive for filling in the form.

Note: To restrict who may access the questionnaire at all, use Access Restrictions. This is also where „serial“ and „mailing“ restrictions are set.

The fact that each participant can only complete the questionnaire once can only be reliably ensured if the participants are known in advance. In this case, each participant can be given an individual key or link that allows them to complete the questionnaire only once. SoSci Survey provides the functions Send Mailings and serial number.

If serial e-mails are used and the addressees are specified to remain anonymous, SoSci Survey takes over a trustee function: The assignment of interview and person is known in the database. However, SoSci Survey does not pass them on to the project manager. In this way, you can actually guarantee the participants anonymity and do not have to spend time with personal data (and the obligations associated with it).

However, caution is required if criminal acts are assessed in a survey (e.g., drug use). Because law enforcement authorities could gain access to the data by means of a court order. Most ethics commissions will not approve of this: if participants are assured that they cannot be identified, this would be a serious breach of trust. In such cases it would probably make sense to accept multiple participation. It is also important to deactivate the recording of time stamps in the Privacy Settings. Even if SoSci Survey does not record any IP addresses in the server logs, subscribers could still be identified by means of time stamps if corresponding data is available from other sources. For example, web clients store links that are opened in e-mails.

With the help of SoSci Survey, it is possible to create different participation URLs and then submit them to the participants.

In the control bar on the left, you can select Participation Links in Controls → Send Invitation. Here you can create a list of different links by specifying how many different links are required. In addition, you can also set how often a questionnaire can be completed with this link. So if you want to prevent multiple entries, you have to select the option „Fill in once with the personalized link“ under Multiple entries. The participant then still has the option to pause the questionnaire and continue it later, but can only submit it once.

When assigning the links, make sure that each participant receives their own and individual link, the program will take care of the anonymity of the links. In this way, no e-mail addresses need to be transmitted.

Even if SoSci Survey does not record any IP addresses in the server logs, subscribers could still be identified by means of time stamps if corresponding data is available from other sources. For example, web clients store links that are opened in e-mails.

Cookies are small fragments of data that a website may place in the participant's browser. If the same participant visits the same website again with the same browser, the cookie is transmitted. In principle, this would indicate whether a participant has already completed the questionnaire.

SoSci Survey currently does not offer cookie functionality for the following reasons:

• Cookies are misused by advertisers for so-called tracking. Therefore, they have a bad reputation - especially with regard to data protection. However, it is not in the interest of a survey to be associated with data protection problems. After all, one is dependent on the trust and honesty of the participants.
• Cookies pretend false security, because:
  • Cookies can be easily deleted by the participant. Some browsers also delete cookies automatically or do not accept them in the first place.
  • If a person uses several browsers (Internet Explorer, Firefox, Chrome, Safari, Opera) each browser stores its own cookies. Even if you cannot find the function to delete cookies, you can participate again with another browser.
  • Of course, other devices (e.g., tablets) also allow you to participate again.
• Cookies are computer-specific, not personal-specific. If several authorized persons use the same computer, a participant may be mistakenly excluded from participation.

There are several ways to store cookie-like data fragments elsewhere than in the browser, e.g., in Flash. Flash cookies are also not supported by SoSci Survey, because:

• With the exception of browser binding, they are subject to the same problems as ordinary cookies.
• If a Flash blocker is used or Flash has not been installed on the participant's computer, the technology will not work.

IP is the Internet Protocol. Every computer with internet needs an IP address that is transmitted to a server when a request is made. The server's response (e.g., an HTML website) is sent back to this IP address.

SoSci Survey allows to record IP addresses. However, the IP address is considered as personal data in German jurisdiction. Therefore, the data record must be treated in accordance with the strict provisions of the BDSG (Federal Data Protection Act) if IP addresses are recorded. Also because participants can no longer be assured of complete anonymity. Therefore, the recording of IP addresses must be actively activated by the project manager in the Privacy Settings.

For the following reasons IP addresses are of little help for the control of multiple subscribers:

• The number of IP addresses is limited at least with the common IPv4. Therefore not every device gets its own IP address; IP addresses are shared. E.g., many companies and universities route all Internet traffic via a single IP address.
• Even with private networks it is not much different: If several computers in a household use a router to access the Internet, the computers within the network use different IP addresses, but all use the same IP addresses to the outside world.
• Private Internet users receive a new IP address from the Internet provider's pool each time they dial into the Internet. Even flat rates are often disconnected after 24 hours and reconnected to a different IP address.
• In addition, there are tools such as Tor or JAP that redirect data traffic via other computers, so that the server „sees“ the IP address of another computer not belonging to the participant. Such tools reliably prevent the server from receiving the correct IP address but may also present the transmitted data to interested third parties on a silver tablet (unless the data traffic is encrypted using HTTPS, e.g., on www.soscisurvey.de).

Fingerprinting describes a technique to create a unique fingerprint from the information automatically transmitted by the browser (e.g., browser identification incl. operating system and version, activated plug-ins, screen resolution, etc.). Similar to cookies, this is very attractive for advertisers to display personalized advertisements.

SoSci Survey does not support fingerprinting at the moment, because it is also no reliable way to recognize multiple participants:

• Fingerprints are very specific. But there are still many computers that provide the same fingerprint.
• Fingerprints may also be considered as personal data based on BDSG provisions.
• Using a different browser or device creates a different fingerprint.
• And here too, authorized persons may be excluded if they use the same computer.

If multiple entries are expected in competitions data cleansing might help. It is very unlikely that a participant will fill out the same questionnaire several times with the same motivation. It is much more likely that they complete the questionnaire as quickly as possible.

If one excludes „rusher“ in the course of data cleansing one should already have identified most multiple participants. This could e.g., be done by identifying questionnaires that were answered extraordinarily quickly, as well as cases with more than 20% missing answers (variable MISSREL). See Leiner (2014) for more information on data cleansing.

If you collect the contact data for the competition separately from the other data (Collect Contact Data Separately) you have to make sure that fast participants cannot register by using the PHP function caseTime().