An open letter to PNAS about their publication of the Facebook study
(I submitted the letter below to the Proceedings of the National Academy of Sciences. It was not published on the basis that they don’t publish letters alleging misconduct, but I’m actually not trying to allege here that the study authors necessarily did anything wrong in conducting the experiment – I’m suggesting that the journal made the wrong decision in choosing to publish the article based on what we, the readers, could see of it. We simply can’t tell whether this research met scientific standards for ethics and editors, reviewers, and readers should be given enough information to make that determination in cases where there may be doubt.)
To the editor:
In the June 17, 2014 issue of PNAS, Kramer, et al. (1) report “a massive (N = 689,003) experiment on Facebook” in which they manipulated the emotional content of views of unwitting users’ News Feeds and studied the impact on users’ subsequent posts. As the editor of a scientific journal that reports studies on the relationships among emotions and choices, I am deeply disturbed by PNAS’s decision to publish this study as written.
The authors do not describe any IRB review of their protocol. They contend that because the intervention was implemented by algorithm and “no text was seen by the researchers”, the study was “consistent with Facebook’s Data Use Policy, to which all users agree prior to creating an account on Facebook, constituting informed consent for this research.” This rationale fails to satisfy for three reasons:
First, the Facebook Data Use Policy specifies Facebook may use user data “for internal operations, including … research”.(2) Although the word “research” appears, this statement clearly refers to internal studies. This study, however, meets the U.S. Federal definition of research as “a systematic investigation … designed to develop or contribute to generalizable knowledge … whether or not … conducted or supported under a program which is considered research for other purposes.” (2)
Moreover, 45 CFR 46 defines a human subject as “a living individual about whom an investigator … conducting research obtains (1) Data through intervention or interaction with the individual…” (2) Although raw data was never viewed by the investigators, it was collected through an intervention with set in motion by the investigators and the study constitutes human subjects research.
Second, even if users understood “research” in the Data Use Policy to refer to human subjects research, it is questionable how agreeing to “research” in general, rather than a specific protocol, could possibly constitute informed consent. None of the elements of informed consent – information, comprehension and voluntariness (3) – are present here.
Third, as the authors report that users were selected at random, it appears that children were not excluded. Based on population demographics and Americans’ social networking use (4,5), approximately 12% of the study’s sample could be expected to be aged 12-17. Human subjects research involving children may be exempt from IRB oversight when it includes only observation of public behavior, but here users were being systematically exposed to higher or lower levels of positive and negative affect which could have significant consequences for their health, including, for example, suicidality. Kramer et al. themselves allude to the potential public health relevance of their findings and the relationship between emotion and well-being.
In short, this study is likely non-exempt human subjects research, should have received IRB review, and should have followed established principles of research ethics designed to maintain public trust in science and scientists. If it did, PNAS should clarify this in a corrigendum; if it did not, it should not have been published by PNAS – or any scientific journal – and should be retracted.
Sincerely yours,
Alan Schwartz, PhD, Editor-in-Chief, Medical Decision Making
(My title included for identification purposes. This letter represents my personal viewpoint, and not the official viewpoint of the journal or its publisher)
References
1. Kramer ADI, Guillory JE, Hancock JT (2014) Experimental evidence of massive-scale emotional contagion through social networks. PNAS 111(24):8788-8790.
2. U.S. Department of Health and Human Services (2009) Code of Federal Regulations – Title 45 Public Welfare CFR 46. http://www.hhs.gov/ohrp/humansubjects/guidance/45cfr46.html. Retrieved 2014-06-28.
3. Belmont Report (1979). The Belmont Report: Ethical principles and guidelines for the protection of human subjects of research. http://www.hhs.gov/ohrp/humansubjects/guidance/belmont.html. Retrieved 2014-06-28.
4. Pew Research Center (2014). 6 new facts about Facebook. http://www.pewresearch.org/fact-tank/2014/02/03/6-new-facts-about-facebook/. Retrieved 2014-06-28.
5. U.S. Census Bureau, Population Division (2014). Annual Estimates of the Resident Population by Single Year of Age and Sex for the United States: April 1, 2010 to July 1, 2013.
