Dos and Don’ts of Twitter Credibility [Study]

Twitter seems to have mushroomed overnight. The social network, famous for its mini-blogging format, has grown from a novel way to make it easier for your friends to keep tabs on you to a frequently accessed source of information. As people begin to get more and more information through tweets that are posted by strangers, concerns about credibility have begun to grow, as well.

A new study from Carnegie Mellon took a look at how Twitter users determine whether or not a tweet is credible. They found that as people begin to rely more heavily upon search engines to locate relevant tweets from people they do not usually follow, they begin to be wary of trusting the content they encounter. The researchers culled their findings to come up with ways you can make your tweets more credible.

The investigators began by doing a general study to determine how people use Twitter, what sort of tweets they find credible, and what criteria they use to determine credibility. Then the authors followed up their original findings by conducting two more tests in which they changed key features of tweets to see how the changes affected the posts’ credibility. They found that people find it hard to tell whether a tweet is legitimate just by reading it. As a result, users base their decisions about the credibility of a tweet on things like user names and whether or not the post has been retweeted.

Study participants were told to conduct a Twitter search for information about a political candidate. They were asked to narrate their search, describing what they were doing and why. Assistants encouraged this process by asking about the tweets that were found, including such questions as whether the searcher thought the tweets were written by the candidate or came from legitimate news sources.

Even though participants typically knew about such Twitter features as user biographies that help to establish credibility, they tended to pay attention only to features that were displayed on Twitter’s user interface. These include the content of the tweet itself as well as the author’s name and picture.

To determine how these three tweet features influenced the way users viewed tweets, the study’s authors created tweets with varying levels of authenticity. Then they tried different types of user names and pictures representing common examples found on Twitter. For example, researchers varied user pictures for the faked tweets by inserting Twitter’s default picture, photos, avatars and logos. Their aim was to see what combination of user name type, pictures and content would be viewed as most credible by users.

The researchers suggest that individual users who want to raise their credibility scores choose photos of themselves, rather than using the default picture or an avatar or logo. Topical names were viewed as the most credible, followed by names that appear to be ordinary first- and last-name combinations. Names that read like email addresses were viewed as the least credible.

Keeping all tweets on the topic suggested by the author’s user name enhanced credibility, as did maintaining a strong geographical connection between the writer and the topic. For example, an author writing about the Indy500 who appears to be near the track will generally impress readers as being more credible. Using non-standard grammar and misspellings undermined credibility more than any other single factor. The authors caution, however, that in some industries or groups, non-standard usage may actually increase credibility.

Other suggestions include amassing followers, posting content that stimulates retweets, including the URLs of your sources and retooling your bio to heavily reflect your topic of focus.

From most effective to least effective, here is a list of features that are common to tweets that participants found credible:

• Tweet is actually a retweet from a trusted poster.
• The author appears to be an expert on the subject of the tweet.
• The writer is one the reader follows.
• The reader was able to click through a URL that was included in the tweet.
• The reader has heard of the author.
• The poster’s account has a verification seal
• The writer regularly tweets on the topic.
• The author often posts content that is similar.
• The tweeter uses what appears to be a personal photo as a user image.
• Other posters frequently refer to the author or post retweets of the author’s tweets.

The researchers also noted some tweeting habits that tended to make authors appear less credible. Listed from least damaging to most harmful, here they are:

• The writer uses a logo as a user image.
• The poster follows a large number of authors.
• The tweeter uses a cartoon image or an avatar as a user image.
• The author failed to replace the Twitter default image with a more personalized one.
• The tweet contains improper grammar and punctuation.