Using a dictionary that researchers use to rate words for positivity or negativity, USA TODAY examined a trove of 80,146 Parler posts captured by analysts at the Social Media Analysis Toolkit before Parler went offline. The posts run from 9 a.m., when Trump supporters ramped up their Save America rally in Washington, to 2:30 p.m., when the Capitol was under full siege.
To assess what was driving changes in sentiment, the news organization also examined words and phrases that gained the most as a share of Parler traffic over time.
Data visualizations made the connections clear.
Not only did this project resonate strongly with readers, gaining wide audience interest, it also caught the eyes of lawmakers.
It was cited and shown during President Trump’s impeachment hearings.
The analysis adds weight to reports quoting attorneys for riot suspects saying Trump’s speech inspired the attack on the Capitol. Such interpretations of Trump’s words were at the center of an impeachment trial of the president in the Senate that begins next week.
We used a dictionary that researchers use to rate words for positivity or negativity to examine a 80,146 Parler posts captured by analysts at the Social Media Analysis Toolkit before Parler went offline. To assess what was driving changes in sentiment, we also examined words and phrases that gained the most as a share of Parler traffic over time.
Our text analysis drew on posts and comments from a panel of 4 million accounts that researchers with the Social Media Analysis Toolkit consider representative of the platform’s overall tone.
We used R for analysis and to share rough visualization among the team.
Visualizations were created using D3.
What was the hardest part of this project?
Aside from Aleszu’s impressive analysis, and Mitchell’s successful visualizations, the hardest part of this project was getting everyone onto the same page and ensuring that the data we were using, and the subsequent visualizations were accurate representative of the events happening online as chaos unfolded in Washington.
Mitchell was able to meet Aleszu by using R, and together they made changes to the dataset, methodology and visualizations until all parties were satisfied with our approach and the result. By sharing software, they were able to speak the same ‘language’ and overcome all obstacles by working closely together.
This project should be selected due to its ambition, use of novel technology, its clear, powerful visualizations, and the impact it made with lawmakers.
What can others learn from this project?
When graphics teams work with data teams, it can be extremely helpful for the visual journalist to interview the data reporter so that they can recreate the methodology for themselves.
By doing so, not only are assumptions challenged, but both reporters may discover avenues that they missed, leading to a stronger final product.
Of course, this requires trust and respect between the teams.
Also, R can serve as a great bridge.