Skip to navigationSkip to content
#NEVERTWEET

Badly, crazy, weak! The data behind the battle over Donald Trump’s incendiary tweets

Republican Presidential Candidate Donald Trump's children Ivanka Trump and Donald Trump Jr., check on their phones during the second day session of the Republican National Convention in Cleveland, Tuesday, July 19, 2016.
AP Photo/Carolyn Kaster
Daaaaaaaad.
  • Tim Fernholz
By Tim Fernholz

Senior reporter

Published Last updated This article is more than 2 years old.

Donald Trump’s Twitter account is central to his presidential campaign. It is the original platform where he hurled insults and rallied his supporters, and now, thanks to a new data analysis, we can see an internal battle over Trump’s message playing out there.

In a proxy war over the identity of the Republican party, Trump’s staff have fought in vain to professionalize Trump’s campaign since his official party nomination. It’s become a parlor game to guess if Trump is writing tweets or if his staff has taken over. But assembling Donald Trump word salad is not easy—a commonly cited tell: Trump tweets ”Sad!” and his staff tweets “Sad.

While most politicians don’t operate their own feeds, or else they “sign” their tweets when they do, Trump claims to manage his Twitter alone—except when he is blaming “interns” for embarrassing retweets.

To find out who’s really tweeting what, data scientist David Robinson decided to test out one popular hypothesis—that Trump’s tweets come from an Android device (he’s known to use a Samsung Galaxy), while staff tweets come from iPhones.

Robinson pulled the @realDonaldTrump account’s tweets, split them by device, and analyzed their content to make a compelling case that the Android tweets are indeed different from the iPhone tweets. His findings suggest that different people use the account, and that the Android tweets are angrier and more negative.

Here’s his case:

Timing

You can see below that most of the Android-originated tweets were published before 10 am, while the iPhone dominates the account’s daylight hours.

Old-school retweets

Trump, a long-time Twitter aficionado, has long been known to favor an anachronistic method for sharing other users’ posts on the platform—tweeting it in quotation marks. For example, his account retweeted this apparent white supremacist in January:

Robinson searched for tweets starting with quotations, and found that almost all of the “quoted” retweets came from the Android phone—only six came from an iPhone.

The strong association of this unusual Twitter style suggests that Trump is the man behind the Android tweets.

All the feelings

Next, Robinson looked at the words that Trump’s account uses most frequently—here are the top 20:

And then he identified the most common words tweeted by each OS using Trump’s account, and that’s where the real difference can be seen.

The most likely words to come from each OS

1
badly
#makeamericagreatagain
2
crazy
#trump2016
3
weak
join
4
spent
#americafirst
5
joke
#votetrump
6
mails
#imwithyou
7
strong
#crookedhillary
8
talking
#trumppence16
9
ago
7pm
10
brexit
tomorrow
11
dead
video
12
dumb
#inprimary
13
senate
#rncincle
14
correct
#supertuesday
15
divided
#trumptrain
16
funny
#fitn
17
guns
#wiprimary
18
temperament
officers
19
treated
#maga
20
beat
3pm

The iPhone tweets are often about logistics, and frequently use hashtags to try and coordinate their message. The Android tweets, well, sound like Trump. And it’s not just you thinking that. Robinson also used software that analyzes the sentiments behind words, associating them with ten different sentiments: positive, negative, anger, anticipation, disgust, fear, joy, sadness, surprise, and trust.

That analysis found that Trump’s Android tweets use “about 40-80% more words related to disgust, sadness, fear, anger, and other ‘negative’ sentiments than the iPhone account does.”

In other words, yup, that’s Trump.

📬 A periodic dispatch from the annual session of the United Nations General Assembly in NYC.

By providing your email, you agree to the Quartz Privacy Policy.