Oklahoma State University graduate student Siddharth Grover has always been interested in politics, so the idea to analyze the Twitter buzz around the US presidential candidates came naturally to him. He did a similar analysis of the General Election in India, while getting his MBA there in 2014.
For this project, Grover analyzed trending topics, keywords and sentiment of the tweets sent from both candidates’ Twitter accounts @realdonaldtrump and @hillaryclinton plus trending hashtags like #trump2016 and #clinton2016, from April to June 2016.
He also collected information on the number of followers, retweets, and “favorited” tweets for both candidates. As we all know during campaign season, political candidates are prolific, so the volume amounted to about 200,000 tweets. In three months.
Using SAS® Enterprise Miner and SAS® Sentiment Analysis Studio, Grover analyzed the contents of every tweet, examining each candidate’s views on the top issues (immigration, taxes, gun control, etc.) and identifying patterns of sentiment and how they shifted across time and geographic regions. He presented his findings at the Analytics Experience 2016 conference. The results were fascinating.
For Hillary Clinton, there was a strong relationship among the terms “status,” “supporter” and “love," which, according to Grover, might imply her affection for her supporters. Terms like “campaign,” “money” and “voter” might imply asking voters to vote or donate money for her campaign.
Grover found that Donald Trump was much more focused on the media. He frequently tweets at media handles like @CNN, @FoxNews and @foxandfriends, whereas Clinton’s top mentions were to @POTUS, @billclinton, and @realDonaldTrump. For Trump, Grover saw a strong relationship among terms “foxnews,” “cnn” and “trump2016.” Trump also frequently used terms like “American,” “gun” and “potus” (President of the United States).
Grover also examined the frequency of policy-focused keywords used by both candidates and it’s an interesting blink-like view of both candidates’ rhetoric. Trump liked to talk about “terror,” “immigration” and “jobs.” Clinton’s favorites were “guns,” “health” and “taxes.” It’s worth noting that the Orlando mass shooting happened on June 12, providing a tragic backdrop for both candidates to project their reactions through their own political lens.
Sentiment analysis provided an interesting macro view of the tone and tenor of each candidate’s campaign. Grover extracted two random samples of 5,000 tweets each and used SAS Sentiment Analysis Studio to perform the analysis. He found that overall sentiment expressed by Clinton was positive (37%) and neutral (35%), with a smaller percentage being categorized as negative (28%). Trump came in more clearly in the negative camp (47%), with the remainder split pretty evenly between positive (27%) and neutral (25%).
Grover’s succinct conclusion states: “We discovered that Donald Trump frequently tweeted at and mentioned media handles in his tweets and the tweets had an overall negative sentiment. Hillary Clinton, on the other hand, had more policy focused keywords and frequently tweeted more to the political establishment. Clinton fell short on engagement, measured through re‐tweets, but generated more overall positive sentiment.”
You can find Grover’s full paper here: Clinton vs. Trump 2016: Analyzing and Visualizing Sentiments towards Hillary Clinton and Donald Trump’s Policies. We’ll also check back with him at SAS Global Forum 2017, when he’ll present the complete findings of his research – he plans to carry the analysis through to the election in November. Any guesses as to what the top terms will be then?