Informed consent, research ethics and compliance

Research ethics and compliance (adherence to legal and institutional rules) have little to do with survey software to begin with. Nevertheless, it is an important aspect of surveys and SoSci Survey can make life easier at one point or another.

What to look out for?

Your first thought with regard to research ethics may be, “Surely my questionnaire is not dangerous?”

Nevertheless, briefly consider whether respondents might suffer any disadvantages as a result of participating in your survey:

• Are you presenting stimulus material?
• Advertisements (including fake ads) can lead to respondent annoyance (“I thought this was a scientific study?!”) or actually influence behavior.
• The presentation of certain content can affect people with a corresponding predisposition (keyword “trigger warning”); what individual respondents feel when receiving seemingly everyday content can vary greatly.
• Deceiving your respondents (false information even when marked as such, fake advertising, …) may not only result in false information sticking with respondents - if respondents feel unnecessarily deceived, it may lead to anger on both sides of the questionnaire.
• The presentation of extreme opinions (e.g., incitement against population groups and other forms of hate speech) can lead to undesirable changes in attitudes.

• Could it be that respondents' answers will later be their undoing?
• What might third parties (supervisors, banks, friends, etc.) do with them if they knew how a person answered in the questionnaire?
• Can you safely exclude that the data can be assigned to individual persons?
• Can you safely exclude that third parties get access to the data? Even if malware is running on your computer?
• Do you ask about behavior that could possibly have criminal consequences? (Disable timestamp)

Informed Consent

When you do or don't need informed consent from respondents and what the information should include can be found in compact form in the [https://www.apa.org/ethics/code?item=11#802]

Much of the information belongs in a good greeting anyway (https://www.soscisurvey.de/help/doku.php/de:create:tips#saubere_knappe_begruessung).

Depending on the research institution or university, there may be fixed guidelines, text templates, or nothing of the sort for informed consent. The templates sometimes differ significantly - and probably more length does not necessarily mean that you actually act more ethically. Who has the time and inclination to read through long texts completely… So the one right template for informed consent does not yet exist.

But if your institution requires you to include a very long text, then look at the following option for display: Display Text Sections Dynamically (Accordion).

If you want to provide detailed information but not necessarily present it, the functions textlink()and buttonToPage() can be helpful. With these functions you can link extensive texts, while the less interested participants will only be confronted with the course version.

If you need or want to ask for explicit consent, you can find the appropriate technical solution here: Obtaining the Participant's Consent

Personal data

If you can find out for individual or all data records afterwards to which (natural) person they belong and if these data say something about the person (e.g. also a personal opinion), then you are dealing with personal data according to DSGVO.

This means a number of legal obligations for you (combined with very severe penalties for non-compliance with the obligations or careless handling of the data), we have compiled a rough overview here: Personal Data

There is a particular tension if the answers in the questionnaire suggest that the person being questioned is at risk. So, for example, if you collect signs of depression or risk factors for suicide. It may then make sense to refer to emergency numbers directly in the questionnaire or - in the case of personal data - to remove the pseudonymization in order to contact the person.

Scale indices can be calculated directly in the questionnaire using valueMean() and valueSum(), so that a suitable text can be displayed via PHP filter when a limit value is exceeded. For removing anonymity (in the case of pseudonymous address entries) there is the function paneldisclose, more about this in the tutorial Privacy sharing and pseudonymization

Clarification / Debriefing

Sometimes it is necessary to inform the respondents after participation about deception or other aspects of the study. For respondents who answer the questionnaire properly until the end, this is of course easy: There, the information can simply be placed on one of the last pages in the questionnaire.

Reaching dropouts is much more challenging. Especially if the persons simply close the questionnaire tab or the browser. Debriefing can then often only be realized via another channel, such as e-mail. But of course, this only works if you know the respondents' email addresses.

If you invite the respondents to the study by e-mail anyway, you can use panelUpdate() to store in the address list which address entries reached a certain page, and who reached the debriefing in the questionnaire later. This way, once the study is complete, you can target and contact via serial mail those who were exposed to the problematic content but did not see a debriefing.

panelUpdate('subgroup', 'stimulus');

Of course, this approach is only possible if the data is collected pseudonymously or on a per-person basis (datenschutz-modus). So here a trade-off between data protection and other protection needs of the respondents may be necessary.

If the e-mail addresses of the respondents are not known in advance, it may be necessary to store them at the beginning by means of an Opt-In for mailing list. In this case, it may be useful to select the single opt-in mode in the question. When using the panelUpdate() function, it may then be necessary to refer to the identifier of the Opt-In question (here as an example “OI01”):

panelUpdate('subgroup', 'stimulus', value('OU01'));