President Trump’s tweeting helps define him. It’s one of the things that makes his supporters love him and his detractors hate him. According to SurveyMonkey data from June 29 – July 3, what people think about Trump’s tweeting (whether it’s positive or negative, or both) predicts their perception of his job performance, even when controlling for age, political party, and education level. Clearly, Trump’s tweeting is part and parcel of his presidency and significantly contributes to how Americans evaluate him.
As president, Donald Trump still independently maintains his own personal Twitter account, @therealDonaldTrump. This is a departure from the precedent established by former President Barack Obama, who only used the official presidential Twitter account, @POTUS, during his tenure as Commander-in-Chief. Trump uses his personal account to confront opponents and to deliver his own unfiltered message to the American people without intermediary news outlets. Despite repeated pleas from his advisors and lawyers, President Trump unabashedly continues to tweet.
Given that there is no end in sight to Trump’s tweets, how do Americans feel about the President’s use of Twitter: do they appreciate his direct and pointed remarks, do they reject it as unpresidential, or are they mixed?
The answer is that most Americans do not like it. A remarkable majority of Americans (64 percent) disapprove of the President’s tweets, a group which includes almost all Democrats (89 percent disapprove) and surprisingly large proportion of Republicans (38 percent disapprove).
We asked how people would describe the general nature of Trump’s tweets, and the top three most frequently mentioned words were all negative: “undignified” (47 percent), “mean” (34 percent), and “dishonest” (28 percent). That being said, 26 percent of Americans found the president’s tweets “entertaining.”
Almost a quarter of respondents went beyond the list of words above and offered their own descriptions of Trump using the “other” response option. We used a machine learning tool—topic modeling with Latent Dirichlet Allocation—on the “other” responses, so we could get the full picture. Topic modeling is a technique that aims to discover themes or “topics” within a large collection of documents: responses, in this case. It is useful when the collection of documents is very large and thus onerous to hand-code, but it also can discover patterns that we, as humans with limited working memory, might miss.
The most frequent topic for “other” descriptions of Trump’s tweets was that the respondent either did not care about Trump’s tweets or did not follow social media. Below are the top 10 topics that people volunteered to describe Trump’s tweets:
Most of the volunteered descriptions of Trump’s tweets were decidedly negative, aside from some people who described the tweets as “honest” or “a way to bypass the biased media.”
These volunteered descriptors align with another question we asked: whether respondents think Trump was doing a better or worse job than they had expected. We used responses to this question to predict the prevalence of these topics in the “other” responses, a technique called structural topic modeling (additional information about structural topic modeling can be found in this excellent vignette for the ‘stm’ R package). The graph below shows a clear division in sentiment about the President’s tweeting based on how well Trump was performing compared to expectations.
Amongst those who thought Trump was doing a worse job than they’d expected, the “Other” topics used to describe his tweeting were, unsurprisingly, all negative. These respondents went beyond the survey-provided options to label the tweets as “undignified,” “mean,” “dishonest,” or “sexist”: they volunteered their own harsh descriptors, like “petty,” “vulgar”, and “stupid.” Trump’s detractors had such strong negative views of his tweeting that they were motivated to write in their own condemnations. Viewing the President’s tweets as “dangerous for the country” was particularly more prevalent amongst those who were disappointed with his performance.
Interestingly, those who thought Trump was doing better than expected were still concerned that the President’s tweeting habit was “undignified,” despite also being more likely to defend the tweets as honest or a way to get around biased media outlets. This implies that even amid those with an optimistic view on Trump’s job performance, there exists a discomfort with the lack of decorum in the President’s tweets.
Significant mixed feelings
Because respondents could select as many words as they liked to describe the President’s tweets, we were curious if people described the tweets as uniformly negative, uniformly positive, or answered with a mix of positive and negative descriptors.
We found that the largest group (46 percent) used all negative words to describe Trump’s tweets, the next largest group (31 percent) used only positive words, yet a sizeable proportion (23 percent) had mixed feelings about the President’s tweets, using both positive and negative descriptors.
There’s a large group of Republicans who somewhat disapprove of the way Trump is using Twitter: 18 percent of Republicans describe his tweets as wholly negative, and 27 percent used both positive and negative language to describe the tweets.
While President Trump’s Twitter use continues to attract scrutiny and attention from the media, these results make it clear that the majority of Americans view his tweeting negatively. Currently, it’s rare for Republicans and Democrats to find as much common ground as they do when they are reading his tweets.
Methodology: This SurveyMonkey Tracking poll was conducted online June 29 through July 3, 2017 among a national sample of 4,965 adults ages 18 and up. Respondents for this survey were selected from the nearly 3 million people who take surveys on the SurveyMonkey platform each day. Data for this week have been weighted for age, race, sex, education, and geography using the Census Bureau’s American Community Survey to reflect the demographic composition of the United States.