An article was posted today on Yahoo news by blogger Adriana Diaz claiming that "Study Says Twitter Can Gauge Moods." The evidence for this claim came from researchers at Cornell University who examined 500 million tweets from more than 2 million people in 84 countries. The article indicated that between the hours of 6 am and 9 am people are very positive, that moods decline throughout the day and at around 4 pm moods pick up, weekends included. They determined this claim through the words people used in their tweets such as "awesome," "agree," "annoyed," and "afraid," swear words as well.
This claim however cannot be proven fact because it does not follow all principles of scientific thinking that are crucial when determining how reliable a claim is. The first principle it defies is ruling out rival hypothesis. Researchers do not know whether or not another variable could have caused these emotions; whether it is that they drank caffeine in the morning that boosted their mood, or whether the morning and nighttime is the only time of day they were presented with an opportunity to tweet. For that, we are not able to reject other probable explanations. Secondly, this claim goes against the principle of correlation isn't causation. The argument that the words people use in their tweets determine the mood they are in is not necessary true for many reasons. First, Tweeter is a social media network that connects individuals over the Internet, which means anything can be said behind closed doors whether or not it is truthful. Secondly, because certain words are used at certain times of the day does not accurately display one's mood. There is no operational definition or factually for what words represent what moods; words can have many different meanings and the tone of these words are unknown. Thus, the connection between moods and words over an Internet site cannot be truly demonstrated.
One alternate explanation for the findings could be that individuals tweet what they think their followers would like to hear. Harvard psychologist, Dan Gildbert, states that "Tweets may tell us more about what the tweeter thinks the follower wants to hear than about what the tweeter is actually feeling" (The New York Times). The principle that is most useful for evaluating this claim is falsiability. This claim does not try and explain every tweet but instead can be used as a testable hypothesis.