In 2012, Facebook conducted a controversial research study on emotional contagion involving manipulating the News Feeds of 689,003 users. The goal was to see if the amount of positive or negative content users saw would affect the emotional tone of their own posts. The study prompted backlash over ethics and privacy concerns. Here are some quick answers about the experiment:
What was the purpose of the experiment? To test if emotions can spread via social networks by reducing exposure to positive or negative content.
How did they manipulate the News Feed? By increasing or decreasing posts with positive and negative words.
How many users were involved? 689,003 users over one week in 2012.
What did they find? Reducing positive content led to less positive posts, and vice versa. Suggesting emotions can spread virally.
Why was it controversial? Over ethics of manipulating emotions without consent and collecting private data.
Background on Facebook and Emotional Contagion Research
Facebook launched in 2004 as a social networking site for college students to connect with each other. It opened to the public in 2006 and quickly grew, reaching 1 billion users by 2012. The platform revolves around users sharing life updates, news, photos, videos, and more on their News Feeds.
With access to massive amounts of user data, Facebook researchers were interested in analyzing behavioral trends. In 2010, they published an analysis of social contagion, finding happiness spreads through social networks while sadness does not.
This sparked interest in experimentally testing emotional contagion by manipulating Facebook content to see if emotions would spread virally to others.
The Social Contagion Hypothesis
The “social contagion hypothesis” proposes that emotions, behaviors, or ideas can spread through populations like infectious diseases. Studies have found evidence for social contagion in things like happiness, voting patterns, and obesity.
Facebook researchers wanted to see if emotional contagion could occur via News Feeds, powered by social influence within the massive Facebook social network.
How the Experiment Was Conducted
In 2012, Facebook data scientists Adam Kramer, Jamie Guillory, and Jeffrey Hancock conducted a week-long experiment on emotional contagion. They manipulated which posts containing positive and negative words appeared in large samples of users’ News Feeds.
Selecting the Experimental Samples
The researchers began with a sample of 155,000 English-speaking Facebook users located across the U.S. who posted at least once per week on average.
They narrowed this down to two equal-sized groups for the experiment:
- The positive emotion manipulation group contained 31,997 users.
- The negative emotion manipulation group contained 31,995 users.
There was also an unmanipulated control group of 689,003 users.
Manipulating the News Feeds
For one week (January 11-18, 2012), News Feed posts containing positive words were reduced for the positive emotion group by about 10% compared to the control.
For the negative emotion group, posts with negative words were reduced by about 10% as well.
This meant each group saw fewer positive or fewer negative posts during that week than normal.
Measuring Emotional Contagion Effects
The researchers analyzed the impact this had on the users’ own Facebook posts made during that same week. They counted positive and negative words using Linguistic Inquiry and Word Count software.
The hypothesis was that being exposed to fewer positive/negative posts would cause users to post less positive/negative content in turn.
Key Findings of the Experiment
In the week following exposure to fewer positive posts, users in the positive manipulation group showed less positive emotion in their own posts compared to the control group. Their number of positive words decreased by about 5-8%.
Users exposed to fewer negative posts conversely showed more positive emotion. Their positive words increased by about 5-10%.
The same asymmetrical effect occurred for negative emotion – exposure to less negative content decreased negative emotion in posts by about 5-10%.
This supported the idea that emotions can spread virally via social networks. Reducing positive exposure reduced positive expressions, while reducing negative exposure increased positive expressions.
Other Findings
The researchers found no significant effects on overall post volume, suggesting emotional contagion occurred independent of amount of usage.
There were also no signs users noticed content being manipulated or changed behavior. The effects appeared entirely subconscious.
Ethical Concerns and Criticisms
Though the study provided evidence for emotional contagion via social networks, it sparked major controversy over ethics and privacy. Critics argued several issues:
Informed Consent
Users were not aware their News Feeds were being manipulated for research purposes. This raised issues of informed consent and transparency.
Harm and Manipulation
Purposely manipulating emotions could negatively impact mental health, especially vulnerable groups. Critics felt it was unethical experimentation.
Privacy Violations
Collecting and analyzing private user data also faced scrutiny over violating privacy boundaries.
Questionable Motives
Some questioned whether Facebook was studying contagion to improve user experience or increase profitability by optimizing News Feeds.
Lack of Oversight
Facebook did not submit the research to an institutional review board or ethics panel for approval, facing criticism for lack of oversight.
Facebook’s Response and Policies Changed
The study was legal under Facebook’s data use policies at the time, but sparked public backlash over ethics. Facebook defended the research as well-intentioned and reacted by updating policies. Changes included:
- Requiring user consent for research participation
- Limiting data collected and time data is held
- Creating an internal review board to oversee research ethics and privacy concerns
Public skepticism remained over how rigorously Facebook would self-police future studies. But the controversy promoted discussion over ethics in big data research and informed consent best practices.
Broader Implications for Social Media Research
The emotional contagion experiment highlighted growing debates as social media platforms aumented research on users:
Informed Consent Difficulties
With billions of users, ensuring informed consent poses challenges. Methodological limitations may prevent consent processes that don’t influence behaviors.
Privacy Expectations Unclear
User privacy expectations regarding data usage for research are often unspecified or vague in platform policies.
Uncharted Research Ethics
Social media allows research methods unforeseen by traditional ethics standards. Guidance for researchers and companies continues developing.
Public Wariness
Experiments like Facebook’s fueled public wariness of research motives and skepticism in platforms policing themselves. But strict external oversight also risks stifling innovation.
Benefits vs. Risks
Ultimately there are often tensions between benefits of large-scale research enabled by social media data and risks from manipulating users or violating expectations. Trade-offs remain debated.
Analysis of the Study Methodology
While controversial ethically, scientifically the experiment revealed insightful findings. We can analyze the study’s methods and results:
Strengths
– Huge sample size (689,003 users) lends high statistical power. Allows detecting smaller effects.
– Naturalistic setting within real-world platform usage has high external validity.
– Longitudinal design tracks changes in behavior over time.
– Compares experimental groups against control group to isolate effects of manipulations.
Limitations
– Relied on correlational data, can’t confirm causation or rule out other variables.
– Short 1-week timeframe limits conclusions on long-term impacts.
– Only reduced positive/negative content, didn’t test increasing positive/negative content.
– Measured limited set of positive/negative words rather than more complex emotions.
– Data based on posts users chose to write, may not reflect other emotional responses.
Future Research
– Explicitly inform users and obtain consent for more ethical replications.
– Conduct longer experiments beyond one week to assess effects over time.
– Compare increasing vs. decreasing positive/negative content.
– Code for broader range of emotions beyond basic positive/negative.
– Collect multidimensional data like face-to-face interactions, physiology, neural activity to measure emotional response.
Conclusion
Facebook’s 2012 study on emotional contagion sparked controversy but provided notable evidence that emotions can spread through online networks by temporarily manipulating exposure to positive and negative content. It highlighted emerging debates around social media research ethics and informed consent requirements that continue evolving to balance innovation, privacy, and public wariness. With appropriate ethics oversight and transparency, online platforms offer promise for rapidly advancing knowledge of virality, networks, and human behavior.