With the midterm elections among us, politicians use social media platforms to communicate with voters (or to share frustration). With 280 characters on Twitter, they have room to express a lot of emotion in the English language.
To see how certain people use such limited space to share big feelings, a language learning website, Preply, conducted a Machine Learning sentiment analysis of over 100 different Twitter feeds belonging to the most-followed U.S. politicians. From congresspeople to the president, Preply assessed the volume and degree of angry tweets they’ve shared in the past year.
How did this work exactly, you may ask? Preply used a Natural Language Processing mode to analyze each politician’s tweets’ volume and degree of anger. The model scans tweets for words that indicate specific emotions, analyzes their context, and predicts the best-matching emotion based on the text. For example:
- Anger: “I feel irritated and rejected without anyone doing anything or saying anything.”
- Sadness: “I found myself feeling a little discouraged that morning.”
- Joy: “I feel very happy and excited since I learned so many things.”
Preply gathered some key takeaways on which high-profile politicians are angriest online. In the world of social media, word choice, order, and context become even more important for communicating feelings.
The big picture of the study showed Republicans tend to express more anger in their writing than Democrats. Out of the 25 politicians who have shared the angriest tweets on their Twitter accounts in the past year, 19 are Republicans, and 5 are Democrats. Tulsi Gabbard, recently independent, also ranks in the top 25.
Those with the angriest tweets are Chairwoman Ronna McDaniel (R), Representative Louie Gohmert (R), and Representative Marjorie Taylor Green (R) — about 50% of their tweets from the past year are angry.
Some politicians shared many angry tweets on their professional and personal accounts. Though their “angrier” professional accounts are ranked above, these Republican politicians also expressed a high volume of anger on their personal accounts: Senator Ted Cruz (R) and Representatives Matt Gaetz (R), Dan Crenshaw (R) and Madison Cawthorne (R).
Sometimes a single word choice can turn a simple sentence into a strong expression of emotion. After looking at the politicians who shared the angriest tweets in the past year, Preply scored each politician according to their likelihood to share something angry on their timelines.
2 of the 25 angriest politicians — those with the highest “anger scores,” or probabilities of sharing angry messages on Twitter — are Republicans. The three politicians with the angriest Tweets are Republican Chairwoman Ronna McDaniel (mean score of 39.1), Representative Marjorie Taylor Green (39.1), and Senator Ted Cruz (39.0). They use words like “wrong,” “absurd,” “terrifying,” and “deadly” to indicate the intensity of their emotions, as well as simple interjections that make their attitudes clear.
Next, Preply parsed each politician’s angry tweets into three different categories — high, medium, and low anger — to see who is particularly feisty online. It turns out some politicians share high levels of anger in concentrated amounts. We discovered that 49.7% of Representative Adam Schiff’s (D) angry tweets could be classified as “high anger” based on his language; former Senator Tulsi Gabbard (who recently left the Democratic party) also tends to share highly angry tweets, with 48.4% of her angry tweets falling into the “high anger” category.
6 Tidbits of Advice for a Happier, Healthier Life(Opens in a new browser tab)
Finally, some politicians share slightly different sentiments on their professional Twitter accounts than they do on their personal ones. Looking at the Democrats, Preply found party firebrands like Representatives Rashida Tlaib, Ilhan Omar, and Alexandria Ocasio-Cortez tend to share more of their stronger, angrier feelings on their personal accounts.
With all of this in mind, we will see what happens on social media as voters head to the polls this week.