Analyzing the Tweets of Presidential Candidates

By Mass Cognition, Special for  USDR

The frontrunners for the Democratic and Republican nominations for president have, via one medium or another, sent enough words into the public square to fill a small library. But what do all those words reveal about the candidates’ respective messages? Answering questions like that – and posing new ones based on deep insights – is the job of Mass Cognition (, a text analysis firm that has developed a proprietary system to pull actionable data from any collection of text, large or small. The company has recently conducted a pair of studies that look at the emotional content of messages published on the four leading candidates’ Twitter  feeds.

“We wanted to understand each candidate’s core message in terms of the use of emotionally loaded words,” explains Andrew Jeavons, founder of Mass Cognition. “To that end, we looked specifically at tweets sent out by Donald Trump, Ted Cruz, Hilary Clinton and Bernie Sanders during two discrete time frames. Next, we applied our proprietary text analysis system to the overall sample and examined the  results.”

The first study collected 1029 tweets from candidates’ official Twitter accounts from February 23 through midnight on February 29– the week leading up to Super Tuesday. Mass Cognition’s system groups words into eight categories based on their emotional content: fear, anger, sadness, disgust, anticipation, surprise, joy and trust. Naturally, only a small portion of words can be categorized in this manner. Furthermore, the analysis does not attempt to parse the grammar or semantics of tweets – only the emotions expressed by particular  words.

Results from a second study conducted from March 2 through March 8 revealed broadly similar metrics, indicating the consistency and reliability of Mass Cognition’s text analysis process. However it also showed some subtle changes in the emotional words used by the candidates. The system is by no means limited to Twitter or social media; any collection of text can be distilled in this way, from product reviews and customer feedback to emails and  transcripts.

The political tweets study reveals the power of the system to produce data – interpretation of that data depends on a client’s specific requirements. However, the approach already has countless applications, and a multitude of actionable insights can be gleaned from just a single collection of relevant text, as the recent study aptly  demonstrates.

About Mass  Cognition

Mass Cognition was founded by Andrew Jeavons, who is also the developer of the Loquortext management and analysis system Mass Cognition uses. Andrew has worked in the survey software and market research industry for over 20 years in Europe, the US and the Pacific  Rim.

SOURCE Mass  Cognition

All opinions expressed on USDR are those of the author and not necessarily those of US Daily Review.