Tweets, Crises and Behavioral Psychology: On Credibility and Information Sharing

How we feel about the content we read on Twitter influences whether we accept and share it—particularly during disasters. My colleague Yasuaki Sakamoto at the Stevens Institute of Technology (SIT) and his PhD students analyzed this dyna-mic more closely in this recent study entitled “Perspective Matters: Sharing of Crisis Information in Social Media”. Using a series behavioral psychology experiments, they examined “how individuals share information related to the 9.0 magnitude earthquake, which hit northeastern Japan on March 11th, 2011.” Their results indicate that individuals were more likely to share crisis infor-mation (1) when they imagined that they were close to the disaster center, (2) when they were thinking about themselves, and (3) when they experienced negative emotions as a result of reading the information.

Yasu and team are particularly interested in “the effects of perspective taking – considering self or other – and location on individuals’ intention to pass on information in a Twitter-like environment.” In other words: does empathy influence information sharing (retweeting) during crises? Does thinking of others in need eliminate the individual differences in perception that arise when thinking of one’s self instead? The authors hypothesize that “individuals’ information sharing decision can be influenced by (1) their imagined proximity, being close to or distant from the disaster center, (2) the perspective that they take, thinking about self or other, and (3) how they feel about the information that they are exposed to in social media, positive, negative or neutral.”

To test these hypotheses, Yasu and company collected one year’s worth of tweets posted by two major news agencies and five individuals following the Japan Earthquake on March 11, 2012. They randomly sampled 100 media tweets and 100 tweets produced by individuals, resulting a combined sample of 200 tweets. Sampling from these two sources (media vs user-generated) enables Yasu and team to test whether people treat the resulting content differently. Next, they recruited 468 volunteers from Amazon’s Mechanical Turk and paid them a nominal fee for their participation in a series of three behavioral psychology experiments.

In the first experiment, the “control” condition, volunteers read through the list of tweets and simply rated the likelihood of sharing a given tweet. The second experiment asked volunteers to read through the list and imagine they were in Fukushima. They were then asked to document their feelings and rate whether they would pass along a given message. Experiment three introduced a hypo-thetical person John based in Fukushima and prompted users to describe how each tweet might make John feel and rate whether they would share the tweet.

The results of these experiments suggest that, “people are more likely to spread crisis information when they think about themselves in the disaster situation. During disasters, then, one recommendation we can give to citizens would be to think about others instead of self, and think about others who are not in the disaster center. Doing so might allow citizens to perceive the information in a different way, and reduce the likelihood of impulsively spreading any seemingly useful but false information.” Yasu and his students also found that “people are more likely to share information associated with negative feelings.” Since rumors tend to evoke negativity,” they spread more quickly. The authors entertain possible ways to manage this problem such as “surrounding negative messages with positive ones,” for example.

In conclusion, Yasu and his students consider the design principles that ought to be considered when designing social media systems to verify and counter rumors. “In practice, designers need to devote significant efforts to understanding the effects of perspective taking and location, as shown in the current work, and develop techniques to mitigate negative influences of unproved information in social media.”

For more on Yasu’s work, see:

• Using Crowdsourcing to Counter False Rumos on Social Media During Crises [Link]