The horrific school shooting in Uvalde, Texas two weeks ago that took the lives of 19 children and two teachers has, once again, brought to the foreground the policymaking stalement to address the causes of mass shootings. At this point, the political debate is predictable: one side asking for gun control and the other expressing their “thoughts and prayers”, asking to arm teachers and school administrators.

The specific policies being debated are also predictable: gun control, mental health services, and the response and responsibility of law enforcement.

Because so much of our public and political discourse takes place on Twitter, I wondered what Senators were tweeting in the hours after the shooting. I decided to analyze all the tweets from all 100 Senators on May 24, 2022 and May 25, 2022. This observation is not intended as a political statement but simply to show the data of what our elected officials are saying. The results, not surprisingly, differ across the two parties.

Before I show the results of the analysis, a few caveats are in order:

  1. First, the list of Senators’ names and Twitter feeds come from Kelly Smith at the University of California-San Diego, which was last updated in April 2022.
  2. Second, I’m not an expert in Natural Language Processing (NLP) and I didn’t intend to do a thorough audit of these messages. As I’ll show, I’m just doing some simple counts here. If you’d like to explore the data, you can download it directly in Excel format.
  3. Third, the person who helped me pull the tweets did so manually and thus it is possible the dataset is incomplete. I didn’t transcribe text from images or audio from videos—the only thing I’m capturing here are messages in the actual tweets.
  4. Fourth, I distinguish between tweets that were relevant to the school shooting versus those that were not. The shooting started around 11:30 am Central Time, so any tweets before that time were immediately removed from the sample. Thereafter, I went through each tweet and counted it if it appeared to refer to the shooting. Complicating matters, May 25th was the two-year anniversary of George Floyd’s murder, so tweets specific to that anniversary were not included in the sample. Again, you can explore the data yourself if you like.
  5. Finally, I used the WriteWords website to count words and phrases, and Jason Davies’ online Word Cloud tool to create the Word Clouds, which I then edited for color and layout in PowerPoint.

Which Senators tweeted?

On average, Senators tweeted three messages related to the shooting over the two days (the median number of tweets was two). Bob Menendez (D-NJ) and John Cornyn (R-TX) tweeted the most during the two days, with Menendez tweeting 30 times and Cornyn 12 times (though Menendez sent out 9 messages that showed a picture from the James 2:14-17 passage from the Bible). Seven Republican Senators and one Democrat Senator did not tweet at all during the period: Roy Blunt (R-MO), Richard Burr (R-NC), Susan M. Collins (R-ME), Roger Marshall (R-KS), Mitch McConnell (R-KY), Mike Rounds (R-SD), Ben Sasse (R-NE), and Brian Schatz (D-HI).

How much did Senators tweet?

Overall, Democrats (for purposes of this analysis, I include Independents with Democrats) tweeted or retweeted 394 times; Republicans tweeted or retweeted 339 times. Once the manual filtering was applied to pull only those tweets relevant to the Uvalde shooting, I find that Democrats tweeted or retweeted 188 times and Republicans 73 times. The average tweet length was about the same between the two parties, and overall, Democrats used 6,214 words in those tweets and Republicans 2,430 words.

What did Senators tweet?

We see differences in what the two parties were tweeting about when we look at single words and two- and three-word phrases. The most common word from Republicans was “Uvalde” (42 times) while the most common word from Democrats was gun or guns (106 times). Because I’m sure people are going to ask, there are slight differences in the word (and variations on) “pray.” Republicans used the word “pray,” “prayer,” “prayers,” and “praying” 19 times, while Democrats used the words “prayers” and “praying” 15 times.

(An aside on word clouds. I’m including this note only because this is a data visualization blog. Yes, I know word clouds are not the best way to visualize qualitative data but, in this case, I found them useful, especially when paired with the bar chart that shows the frequencies. There is more one could do visually with these data in an interactive data visualization, but I am focused on static images here.)

The two word clouds below (and the associated bar charts) show the 25 most common words tweeted by Senators in the two parties. Words in gray were included in tweets from Senators in both parties; words in color are unique to each. As you can see, the most popular words from Democrats tend to be more active—“act,” “action,” “need,” “must,” “now.” Tweeted words from Republicans appear to be more about the shooting itself—“Uvalde,” “Texas,” “School,” “Elementary.” The word “Alex” appears in the Republican list because it appears in tweets referencing the Luke and Alex School Safety Act.

Word cloud of most frequent words tweeted by Democrat Senators on May 24 and 25, 2022
Word cloud of most frequent words tweeted by Republican Senators on May 24 and 25, 2022

Differences in counts of two- and three-word phrases are also interesting. The most common two-word phrase in tweets from Republican Senators was “in Uvalde” (28 times) while Democrats were much more likely to tweet the two-word phrase “gun violence” (41 times). There are perhaps an even starker difference in three-word phrases: Republican Senators used the phrase “in Uvalde Texas” 11 times, “at robb elementary” 8 times, and “school safety act” (again, the Luke and Alex School Safety Act) 7 times. Democrat Senators used the phrase “we need to” 9 times, and “in uvalde texas” and “gun safety legislation” 8 times. Phrases tweeted by both groups are shown in bold text.

Most common two and three word phrases from Senators

Wrap-Up

I’m sure there’s much more to do with this data—a more thorough NLP analysis; expand the sample to the House of Representatives; expand it to more days; and more. If you would like to explore the data on your own, you can do so by downloading this Excel file.

If you would like to help support the families of the victims and survivors of the shootings, here are some places to consider: