The Washington PostDemocracy Dies in Darkness

Latest Twitter data shows Tom Brady is not viewed as a liar or cheat

New England Patriots quarterback Tom Brady (REUTERS/Charles Krupa)

It is unknown what discipline — if any — Tom Brady will suffer for his probable role in Deflategate but one thing is for sure: the public is starting to turn negative on him.

Manish Tripathi, assistant professor in the practice of marketing at Emory University, provided Twitter data and as you can see, people are upset with Brady starting with the Super Bowl. The software used by Tripathi explains how tweets are coded:

Auto-Sentiment uses pre-defined sentiment categories. It uses a vast set of training posts (over 500,000) that were hand-labelled as positive, negative or neutral. We use these labelled posts to calculate the frequency distribution of each word, negated word, emoticon, etc. present in those documents across the positive, negative and neutral categories. These frequency distributions are then used to construct a model that analyzes each new post and classifies it by sentiment.

Here is an example of what they classify as a negative tweet about Brady:

Here are examples of what they would consider positive:

And here is what they would consider neutral:

“In February, a lot of negativity is from Seattle [Sehawks] fans and also people from New York,” Tripathi explained. “Even in the last two days a lot of negativity is coming from the state of New York.”

“We have seen this before,” said Tripathi. “Most of the emotional tweets, especially negative, you get a lot of that in New York and Boston. In both New York and Boston that negativity can be directed at their own players and team, and after that it is not their own teams it is the team from either New York if you are in Boston, or Boston if you are in New York.”

The word cloud below shows that no matter where the tweet originates from, most stop short of calling Brady a “liar” or say that he “lied” or “cheated.” The larger and darker the font, the more frequent the word usage in tweets on May 7, 2015.