Facebook has recently come under fire for conducting an experiment described as “massive-scale” emotional contagion in the social network. While the idea that Facebook views friends as guinea pigs may be unnerving to some, the controversy has centred on the fact that neither Facebook nor researchers from Cornell University obtained informed consent from the study subjects.
The study results were published in the Proceedings of the National Academy of Sciences despite a lack of institutional review board (IRB) approval (Kramer et al. PNAS USA 2014;111:8788-8790; free full text at www.pnas.org/content/111/24/8788.full.pdf).
Facebook has argued that users agree to the company’s data use policy when they sign up, which constitutes informed consent. Cornell has stated that its IRB ruled that the project was not subject to the university’s Human Research Protection Program. PNAS USA normally adheres to the Health and Human Services Policy for the Protection of Human Research Subjects. Editor-in-Chief Inder Verma acknowledged that as a private company, Facebook was under no obligation to conform to the HHS provisions, so the journal decided that it could proceed with publication. “It is nevertheless a matter of concern that the collection of the data by Facebook may have involved practices that were not fully consistent with the principles of obtaining informed consent and allowing participants to opt out,” Verma added in a published statement of concern.
As for the experiment itself, Facebook manipulated the content received by 689,003 users to test whether this had an impact on their contacts’ subsequent communications. For one group, news feeds expressing positive emotions were filtered out; while those expressing negative postings were filtered out for a second group. A third control group had news feeds randomly filtered without regard to content. The hypothesis was that receiving fewer positive messages would result in fewer positive posts to contacts (and vice versa), which the group defined as emotional contagion.
The large sample size was able to show that the positivity-reduced group expressed fewer positive words (and vice versa), however, this effect was miniscule (effect size d=0.001). Despite this, the authors claimed that the results show that emotional contagion exists in social networking.
Although the study was controversial, it is not the first such study to be produced by Facebook. A previous 3-year study reported that on rainy days, the number of positive posts by Facebook users declined by 1.19% and the number of negative posts increased 1.16% (Coviello et al. PLoS One 2014;9:e90315; free full text at www.ncbi.nlm.nih.gov/pmc/articles/PMC3951248/pdf/pone.0090315.pdf). A positive post yielded an average 1.75 positive posts by friends; a negative post yielded an average 1.29 negative posts. Facebook estimated that when it rains in New York City, users post an additional 1,500 negative posts, which in turn generate 700 negative posts by friends.
This study was approved by the IRB of the University of California, San Diego, which waived the need for participant consent.
A number of studies have explored the concept of social contagion (reviewed by Christakis & Fowler. Stat Med 2013;32:556-577; free full text at www.ncbi.nlm.nih.gov/pmc/articles/PMC3830455/pdf/nihms-524812.pdf). Earlier studies used large databases such as the Framingham Heart Study, to investigate the putative effects of social networks on a range of health behaviours, such as the spread of obesity (Christakis & Fowler. N Engl J Med 2007;357:370-379) and alcohol use (Rosenquist et al. Ann Intern Med 2010;152:426-433), to social phenomena, such as latrine ownership (Shakya et al. Soc Sci Med 2014; epublished March 12, 2014).
A more disturbing trend is to use databases to try to associate genetic factors with qualities such as friendship (Christakis & Fowler. PNAS USA 2014; epublished July 14, 2014), happiness (De Neve et al. J Neurosci Psychol Econ 2012;5), and political ideology (Settle et al. J Polit 2010;72:1189-1198). In the latter study, data from the National Longitudinal Study of Adolescent Health were used to test the hypothesis that individuals with liberal political views are more novelty-seeking, as represented by the 7R variant of the dopamine D4 receptor gene.