Blog post Hope to Nope: Graphics and Politics 2008 - 2018
Make data great again: 24 hours of Donald Trump data
Part of Pulsar’s contribution to the current Design Museum exhibition Hope to Nope: Graphics and Politics 2008-18 includes an interactive search of a day of Donald Trump’s infamous Twitter account.
Trump conducts a surprising amount of the US’s business and announcements from his own Twitter platform, as opposed to @POTUS, the official account of the president. Reports have shown that this can be alarming, or refreshing to observers. Whatever you think, it’s added a different dimension to the traditional way of hearing from politicians.
But what can we learn from looking at what he said over a certain day? The day we dived into was the 27 of February. The day was selected as the most recent day where the US president generated an average number of mentions over a 1 year period and spent the day in Washington DC at the White House. This was the day that Trump tweeted 'WITCH HUNT' in response to the latest revelations from the Mueller investigation which was probing into his business dealings in Russia before the 2016 election, including a visit to the Russian-hosted Miss Universe pageant in 2013.
Diving into the ‘24 Hours of Trump’ search, which Hope to Nope visitors can interact with as part of the exhibition, there are a few things we can find out about the social media response to the US President that day.
Our search simply uses the Pulsar platform to pull through any public mentions of Trump during the 24 hours of midnight - midnight GMT on Tuesday 27 February 2018. The platform then organises the data to let us understand:
•The Trump audience
Which countries and cities are talking most about Trump? What can we learn from their public bio about their gender, job or interests? What times were they posting?
•The Trump network
Does Donald Trump bring people together? The Network Graph on Pulsar shows us how the different accounts mentioning Trump cross over and whose posts were getting the most traction that day.
•Trump content
Which websites are being shared in posts about Trump? Which topics and keywords are coming up time and time again alongside his name?
So what can we find in the 24 Hours of Trump search?
Scanning over the Pulsar search, there are a lot of tabs to dive into. But one fascinating look is at the demographics of the people who talk about Trump on Twitter on the Audience tab on Pulsar.
We can see that there are 757,000 users who mentioned Trump on Twitter on 27 February, with more men than women talking about him (although the genders of 53% of users could not be detected by the public data algorithm). 81% of these users are, unsurprisingly, based in the US, with the UK the next biggest mentioner at 6.4%.
Looking at some of the repeated keywords in the user bios gives a fairly stereotypical view of the majority of people talking about Trump:
We can see they like to mention ‘maga’ or Make America Great Again, and their conservative views, as well as being retired, mothers and wives.
Some of the most repeated topics in posts mentioning Trump included…
The red keywords had negative sentiment attached, the green positive and the grey neutral. We can see the election campaign investigation coming up with mentions of Mueller, Russia, evidence and collusion, as well as a lot of discussion of gun laws, with NRA, and weapons mentioned.
We can also see how the keywords and topics used in posts about Trump were associated with one another - did people who mention Donald Trump also mention Melania Trump, or compare him to Obama?
In this bundle chart below we can see how some of the most used topics were associated - with the connections to the word ‘evidence’ highlighted. People who mentioned ‘evidence’ also talked about ‘collusion’, Fox News and Mueller, as well as Trump’s handle @realdonaldtrump and the fact that he is the president.
Given that it was a fairly bad day for Trump in the news, it’s not surprising that Pulsar’s sentiment analysis module picked up a fairly high level of ‘disgust’ in tweets that mentioned him.
This module analyses the language in posts for the five human emotions - joy, disgust, fear, sadness and anger. Anger was most prominent, followed by disgust and fear on 27 February.
Come down to the museum and see what you can find - it’s time to make data great again.
Related exhibition
The Design Museum has partnered with audience intelligence firm Pulsar in order to create data visualisations which uncover the impact of social media conversations around global political leaders. This blog post was written by Pulsar to mark this partnership and their display inside the Hope to Nope exhibition.