COVID-19 IN AMERICA: TWITTER ANALYSIS
Digital Humanities 199 Project
Justin Suarez
DATA
Tweets Collected About the COVID-19 Pandemic
DATA COLLECTION
Throughout the course of the pandemic, I have closely monitored the use of Twitter especially keeping an eye out for trending hashtags about the virus. I decided to focus on hashtags as people are more likely to fall behind mass movements with each other compared to individually looking at terms by themselves. In this regard, I scraped tweets off Twitter through the Twitter Archiving Google Sheet (TAGS) module for each hashtag. When running the TAGS for each of these hashtags, I collected their data when they were trending and have set up their TAGS in order to collect on an hourly basis to have a constantly updated database. Once finished with data collection, we cleaned the data on the Google Sheets to be utilized in visualizations. I also implemented Python code in order to create sentiment analysis for the tweets that come from the TAGS sheet. As for visualizations, I decided to use both Voyant and Tableau to generate my insights as to what the data is conveying. Finally, I put up the data on my Wix site in order to have the research and data available to anyone online.
DATA SELECTION
Throughout the initial months of the pandemic, I chose to monitor seven specific hashtags that people have made trending on Twitter about COVID-19 and that are of worth analyzing in relation to the United States government. These are the hashtags that ensued following certain moments within the pandemic and interaction between the U.S. government and the American people.
MONITORING HASHTAGS
#DONTDRINKBLEACH
This hashtag came as a result of President Trump speaking on using bleach as a remedy for COVID-19 during one of his speeches. Thus, people were trying to advise other to not drink bleach.
#FIREFAUCI
When Dr. Anthony Fauci commented on how earlier measures of lockdowns and social distancing could have flattened the curve much more, Donald Trump retweeted a tweet that called for the resignation of Fauci. Hence, people on Twitter rallied around Fauci to defend him and the scientific facts he was given on the issue.
#HOWMUCHLONGERTILL
People were starting to get wary of the lock downs and social distancing. So, they were commenting on what they were waiting for with this hashtag.
#HOWTOSPENDYOURSTIMULUS
As the federal government rolled out with stimulus checks as unemployment soared, people took to Twitter to talk about how they were planning on spending their checks.
#INJECTDISINFECTANT
As with #DontDrinkBleach, this hashtag is another form of that instance of Trump informing Americans that ingesting or applying disinfectants as a remedy would work. This is a liberal backlash to mobilize people on social media to not fall victim of these false assumptions.
#TRUMPCOVIDFAILS
Throughout the COVID-19 pandemic, President Trump has appeared to fail American society with the amount of misinformation and lack of response to adequately combat the virus and limit its spread.
#WHATTRUMPHASTAUGHTUS
Along the same lines as #TrumpCOVIDFails, this is the continuation of that surge of tweets as people start to imagine what has gone wrong throughout the pandemic in terms of the federal government.
INFORMATION ON THE DATA SET
Data Set
Over the course of 2 months with these seven hashtags, I was able to scrape 5671 unique tweets in order to use for my analysis. This time frame spans from April 5th to June 6th. After the data cleaning, these tweets are all in English. Given that these events are occurring in the United States and are trending for the time being, most of the people who were tweeting are from the United States.
Data Cleaning
I cleaned the data in Google Sheets. First, I made sure to separate the unique tweets from the retweets because I wanted to investigate the words that individuals are using. So, I placed them in respective sheets within Google Sheets. The unique tweets helped me by being able to determine the specific words that people are resonating with. Second, I went through and removed the tweets that were in a different language. I wanted to make sure to constrain the data set to the United States as best as I could. Without geolocation of the tweets, the next best thing would be to make sure we had tweets from Anglophone countries. And, lastly, from the remaining tweets, I took out the links embedded in them and the special characters that would show up as much as I could.
THE TOOLS FOR ANALYSIS
VOYANT TOOLS
I used Voyant to investigate the tweets. Once the data was inputted, I was able to create word clouds of all the hashtags.
SENTIMENT ANALYSIS - PYTHON CODE (VADER SENTIMENT)
Sentiment Analysis would be especially useful to understand the ways in which people are using their tweets in a positive, negative, or neutral fashion. I used sentiment analysis that was created by Vader Sentiment in order to analyze the tweets I scraped.