The aims of this report are to gain analytical insight into the Twitter accounts driving the #TrumpShutdown hashtag as the US enters a historic moment: the longest government shutdown in history.
It is of particular interest for journalists to find out which accounts are driving a hashtag, in order to understand: who the key influencers are, what the secondary issues surrounding the conversation are, and how to integrate these findings into their reporting. Twitter is an invaluable tool for newsgathering and newsmapping, but there are many pitfalls. Countless studies have noted that the ‘echo chamber’ problem – whereby people only interact with people sharing similar political views and ideas – is very real, and affects journalists as much as individuals on Twitter. As such, journalists looking to the social media platform to analyse developing news stories must be wary of the effect that a hashtag shared by an influencer (someone who has many followers, for example) can have in changing the conversation drastically. In turn, this can affect the manner in which a story is perceived, with real implications.
The US is currently experiencing the longest partial government shutdown in its history – a record of 24 days and counting on the 13th of January. Trump has threatened that it could well continue for months. The two political parties are in a deadlock over the funding of the border wall – one of the key themes we shall be looking at in the report. Meanwhile, the New York Times published an explosive story, titled F.B.I. Opened Inquiry Into Whether Trump Was Secretly Working on Behalf of Russia, on the 11th of January. This is another key topic in this report. Finally, this week will see thousands of federal employees forgoing pay for their work since the shutdown. There have been reports of some turning to crowdfunding accounts to raise money for bills, and the Coast Guard published a tip sheet (which has now been removed) aimed at furloughed workers, suggesting ways to make a bit of spare cash – including walking pets and holding garage sales.
For the purpose of this report, we shall look at the primary hashtag #TrumpShutdown between January 12-13, over the course of 24 hours. A combination of Twitter scraping and TAGS was used to collect the dataset. We shall examine the top tweeters, the levels of engagement, the issues surrounding the hashtag, and the real implications of our findings.
Engagement is defined as the strength of the interactions with the topic. Original tweets represent a stronger level of interaction than a retweet, for instance. Out of 76552 tweets, 54779 were retweets – 21773 original tweets, a ratio of 2.56 retweets per original tweet.
This was the most retweeted tweet (2134 times):
I am wondering. Do you guys think that Trump was the kind of landlord that would be gracious enough to work with tenants who couldn’t come up with their rent?— ❤️🧡💛Mia💚💙💜 (@mommamia1217) January 12, 2019
Yeah. Me either.#TrumpShutDown#TrumpTemperTantrum
The highest tweeting frequency was 5963 an hour, during the second hour of the timeframe (mid-afternoon to early evening US time on Jan 12). The median tweet frequency was 2986.
Although the 'location' information on a Twitter account is not reliable, there was more than enough data to form a picture of some of the areas tweeting the most:
Just four accounts were geocoded, out of 76552 tweets. 23867 had no location at all - 31% of the total accounts.
There was a fairly low bot level within the tweets - just 2 out of the top 10 tweeters were clearly bots.
The hashtag #TrumpShutdown includes an opinion – that Trump is responsible for the current shutdown, rather than, for example, the Democrats. As such, many of the tweets will be critical of the shutdown, and anti-Trump. We are interested in the presence of bots, the issues that the top tweeters associate with the primary hashtag, and their influence.
For the timeframe analysed, these were some of the top tweeters. They represent 901 of the tweets in the data set - just 1%.
During the time frame, @openletterbot at 159 tweets. Otherwise known as Resistbot, it’s a service that allows people to compose letters to their elected officials rapidly. It is a bot account – tweeting a small sample of the letters. As an example of a bot, it is at the other end of the spectrum from the malicious bots that are aimed to sow discord on social media. A random sample of its followers using Botometer reveals that most have a very low botometer score – suggesting that it is not connected to a botnet and is used by real people on Twitter. Unfortunately, the letters from constituents published are tweeted as images, so the words can’t be analysed. However, this reveals an important aspect of the top tweeters for this hashtag: social media activism.
During the timeframe @HingleM13834235 tweeted 188 times. Their output was bot-like, and they scored 2.6 on the bot-o-meter – not conclusive. However, they represented a different type of high-volume tweets:
As we can clearly see, this account is taking a specific path with regards to the topic: Russia features heavily through the hashtag #russianasset at 128 times – referring to Trump – and the word Putin comes up 66 times. In comparison to other accounts, there are no mentions of the workers affected by the shutdown, nor the reason for the shutdown itself, the border wall.
These were their top words:
Inflammatory language, imagery and concepts – this account could be a malicious bot account. The amount of photos and videos in very high, especially because the account was created on December 21, 2018. It has only 405 followers. A sample of the followers, however, reveals as low bot-o-meter score for the accounts checked.
@PrincessBravato is an extremely prolific account with 82.2k followers. The username WeThePeople (followed by two US flags) suggests that social media activism will be relevant. The top words tweeted by @PrincessBravata in this time period were:
Of the other key issues are represented in this dataset: ‘wall’ comes in at 7 mentions and ‘border’ at 6. 5 of the 7 tweets containing the word ‘wall’ were original tweets, suggesting that this user has a high level of engagement with the term.
Mention of the employees affected by the shutdown is very low in this dataset too – there are 2 mentions of ‘working’ which directly and indirectly reference the employees, both are retweets rather than originals.
Most prominently on this account, however, was the unfolding Russia story: #trumprussia at 76 mentions, with 53 original tweets containing that hashtag – a huge engagement rate.
Listen you have a Russian asset that has deliberately shut down the government because they are personally investigating him for crimes against United States#TrumpRussia#TrumpShutdown #ImpeachTrump pic.twitter.com/M7i0hJyrOC— WeThePeople🇺🇸🇺🇸 (@PrincessBravato) January 13, 2019
Having looked at some of the top tweeters to get a sense of which issues surrounding the shutdown are being referenced with high frequency, now, in the interests of examining influence, let’s look at the accounts in the dataset with the highest follower count.
Feeding the text of the tweets published from the top follower accounts into the word count, and removing certain words like ‘RT’ (for retweets) and ‘#TrumpShutdown’ (we want to look at the issues surrounding the primary hashtag), we retrieve this:
Still, we find #trumprussia to be at the forefront of the tweets. However, this could be in large part due to one account, @PrincessBravato, that features among the top follower accounts too – several of the accounts don’t mention it at all. An awareness of this is important as a journalist analyzing word clouds – thus far, many elements have indicated to us that the public interest, based on these data sets, is with the Trump-Russia story. And yet, the data is very easily skewed by one of two accounts tweeting at high frequency to large follower accounts.
Rather than digging deeper into individual accounts, let’s look at a generalized word cloud, based off random samples of tweets over the course of the timeframe.These clouds show what the general public is tweeting about #TrumpShutdown – painting, again, a different picture – one that reveals the human side of the shutdown with greater clarity.
‘Families’ is a term that features prominently. Terms such as ‘safety’, ‘risk’, ‘help’, ‘crisis’ and ‘cruel’. There are also active emotions, such as ‘anger’ and ‘blame’. Analyses of these terms could be useful in determining the public mood with regards to this shutdown – with the more plaintive terms suggesting that the public is perhaps expressing hopelessness and sadness, or, on the other hand, expressing strong resistance to the administration behind the shutdown.
Public institutions and figures are mentioned: ‘Coast Guard’, ‘government’, ‘Senator’, ‘Democrats’. There’s a greater focus on the issue behind the shutdown, at least on the surface, with words like ‘wall’ and ‘Mexico’, ‘border’ and ‘illegal’.
‘Putin’ and ‘TrumpRussia’ do come up, but it’s a very, very small component of what these clouds show the public is focused on.
Twitter is a powerful tool that can sway the tone of a story. Research has shown that journalists’ news judgment is impacted by the platform depending on usage, to the point that tweets were deemed equally newsworthy as headlines appearing to be from an AP wire by some of the journalists in the study.
As a story develops, journalists looking to social media must be particularly cautious of allowing high-interest accounts to dictate the angle of the story. In the case of #TrumpShutdown, of course the news of potential collusion with Russia is relevant – but to be cited at such frequency by highly visible accounts does not signify either a public interest or particular relevancy to the developing shutdown story. Twitter accounts are likely to want to get more engagement, thus seeking out topics like Trump’s involvement with Russia which are likely to provoke an extreme reaction in the reader – prompting them to like, retweet or create their own tweets. In this way, and added to this the way Twitter’s algorithms work, ideologically-contentious themes run the risk of being blown out of proportion, detracting from the human stories that should be also made visible. The hashtag #ShutdownStories, by far less popular, offers a good alternative to be read in conjunction with #TrumpShutdown – as it aims to document the harms suffered by citizens during this event.
Today’s #shutdownstories is “can we keep our daughter in preschool? Can we find a way to make the tuition payment? Should we ask to delay our payment?” The shut down hurts families. It’s hurting my family.— Mrs Grondin (@MrsGrondin) January 14, 2019