Facebook has been accused of a breach of research ethics because it did not obtain consent for its study into the emotional effects of manipulating user newsfeeds. The overall terms and conditions for Facebook include the statement that information may be used “for internal operations, including troubleshooting, data analysis, testing, research and service improvement” (https://www.facebook.com/about/privacy/your-info#howweuse). However, researchers and legal experts have pointed out that the feed manipulation study went beyond gathering information and involved experimental interventions. The study involved varying the content of users’ feeds to find out if this affected the users’ emotions. For this type of study, university ethics committees typically require explicit informed consent; users need to know that they are taking part in a study, how the study will be conducted and (broadly) what the researchers are looking for. Participants also need to know if there are any potential risks from taking part in the study and should be able to opt out at any time. In this case, the risks was that a Facebook user would experience lowered mood and, however slight this effect might be, it is still unethical intentionally to expose people to this danger (in the cause of research) without their knowledge or consent. In addition, because people did not know that they were participants in a study they did not know about the possibility of opting out.
The Facebook feed case shines an interesting light on the ethics of internet research. For a researcher, the internet is a potential goldmine, full of data. Do you want to know what people think ‘happy’ looks like? A Flickr search will provide you with thousands of images to analyse and count. Do you want to know how people mark, celebrate, struggle with and support each other through the long days of fasting? The Twitter hashtag #Ramadan2014 will tell you. This type of data is ‘wild’; the images and tweets have been posted publicly and are free for anyone to look at and to analyse. However, most people, when posting online, do not think that they are creating research data. They are simply publishing interesting photographs or talking to friends, family or fellow commenters. Therefore, an important question for researchers is whether it is ethical to collect and conduct research on the data. For example, one commenter on the Guardian article linked above cited the MELC (Metaphor in End-of-Life Care) study as an example of unethical internet research. This study uses a large corpus of data, some of which is drawn from internet forum postings. However, although people’s data, in the form of published posts, is used in the MELC study, people are not individually identifiable nor is there any possible risk to the people who created the dataset. An example of Twitter research is Doughty et al (2011) who looked at tweets about the TV show The X-Factor. This paper includes quotations from tweets although the users are not identified. However, the paper does not mention ethics and does not indicate whether the study underwent ethical review. It is unlikely that ethical review was considered unnecessary as the study was conducted with data already in the public domain. Nevertheless, individuals who find their tweets or posts discussed in academic papers may believe that their permission should have been asked especially if they feel that their words have been misrepresented.
Research into language use on the internet is still a relatively new field and some of the ethical questions are still to be worked out. With ‘old’ media, even in digital form, researchers can assume that it is legitimate to analyse published texts and not necessary to ask permission of the writers or publishers. These texts tend to be created by professional writers who understand that when something is published then it becomes, to a certain extent, the property of the public * who may interpret, discuss and use it in ways not intended by the original author. It’s painful for authors to think of their carefully crafted work being placed under cat litter trays but once they have been published, texts are no longer under the control of their authors. Online texts such tweets, posts, comments, images, YouTube videos, etc., are written by diverse authors, many of whom will not have thought about post-publication ownership. These authors may not feel that collation and analysis of their work is legitimate and may not agree that through publication the work has been made available to researchers.
The Facebook feed study clearly should have undergone ethical review and it required informed consent. With other types of internet language/communication research, however, the ethical position is far less clear. On a small forum, informed consent is a possibility but it is not feasible to collect informed consent from large numbers of Twitter users. Nonetheless, research into the ways that people communicate online is essential. Without research it will be very difficult ever to understand incidents such as last year’s Twitter harassment of Caroline Criado-Perez yet a requirement for informed consent would make such research impossible. This would, in its own way, be as unethical as conducting a study into the emotional impact of different newsfeed content without the knowledge or consent of participants.
* I am not talking here about intellectual property rights which remain, of course, with the author.
Doughty, Mark and Rowland, Duncan and Lawson, Shaun (2011) Co-viewing TV with Twitter: more interesting than the shows? CHI2011 Available from http://eprints.lincoln.ac.uk/12540/