Following on from last week’s post about the Facebook experiment, it seemed sensible to look at what methods can be used to gain fair consent in internet-based research.
The simplest method, and one most commonly used in surveys and questionnaires, is Implied consent. This method means the requirement of a signed consent document is waived, and the consent is instead embedded in taking part. In a survey, this could be done by only showing the questions to people who ticked an “I agree to take part” button at the very beginning, and redirecting people who ticked “I do not agree to take part” away from the survey. This would be embedded or implied consent, as people would have to say they wanted to take part in order to see the questions.
For simple studies with little to no personally identifiable information, and little to no risk of harm, this method of gaining consent is good enough. However, if this method is done on studied where participant information isn’t stored, then participants have no way of requesting their own data to be withdrawn. Studies using this method can often be a flash in the pan- performed and then forgotten, with often no record of their methods, participants, or findings.
Therefore, for large-scale, complex, or important studies, implied consent isn’t the way to go.
The most sensible for way for a company with a large user base such as Facebook, Google etc, is to get what is known as “generalised informed consent”. This can be obtained by asking people similar to the target demographic if they would give consent to take part in an experiment, after fully informing them about it.
For example, if a researcher wanted to use psychology students at a local university, then the researcher could ask all the psychology students at a different university- if the asked students gave consent, the researcher could assume that the first group of students would also agree. This means the research can be carried out on the first group of students, and the students given an opportunity to be debriefed and to withdraw if they don’t agree with this method of gaining consent.
This method isn’t perfect, as it relies on the second group being similar enough to the first to generalise across. The perfect version of this would use a matched pairs design, asking one of the sets of pairs and testing the other. However, getting this to work would be very inefficient : considering the difficulty involved in setting up a good matched-pairs experiment, using one for a survey study could be seen as a waste of useful matched participants.
Direct, Delayed Consent
Another method is to ask people far enough in advance that they have given informed consent, but the content of the experiment is not fresh in their minds by the time it takes place. This way, they have given consent, and will probably remember that they agreed to take part in research given specific parameters, without the deception content being prominent enough for participants effects to be a problem.
This is a variable method, as obviously different people will forget more easily than others. Also, some people may have changed their minds about whether they want to take part in studies in the interceding time, which may not have been updated. So there is the risk that people will agree to take part in a study, then realise a few months later that they actually don’t want to give any information, causing there to be less participants than expected. Similarly, people may initially disagree if they aren’t sure exactly what the terms mean, and forget to update their actual wishes.
A “Research and Studies” Settings Page
Finally, a website or social network could add a page to their settings dialogue specifically for research inclusion. This could use any or all of the consent methods above, with new questions appearing every so often, so people are reading about it in small amounts. This also provides the options of people being able to change their minds, as they can simply take the survey/answer the questions again with their new preferences.