Drew Margolin, a professor of communication at Cornell University who studies human dynamics through social media, isn’t just tracking how the electorate is reacting to candidates in one single moment via Twitter, but how they have been reacting since the beginning of the primaries. This historical data gives Drew and his collaborator Yu-Ru Lin, University of Pittsburgh, a unique window into changing sentiment within a party about candidates and topics, as well as across party affiliations.
For example, how are early Trump or Hillary supporters reacting differently from more general Republicans and Democrats who originally favored another candidate? What about those who may have been #NeverTrump or #NeverHillary? And what does it mean for Election 2016?
Margolin found that during the party conventions, twitter users reacted favorably to Clinton’s most memorable, quotable lines, but unfavorably to Trump’s. How did audiences react to ‘zingers’ during this first debate? Margolin’s analysis below.
The make up of the computational focus groups, and general thoughts on using Twitter as a sentiment barometer can be found here.
Negative attention has been the story of this campaign. Both candidates have historically low favorability ratings (negative), both have been household names for a generation (attention), and both draw a lot of criticism from the press (negative attention). Recently, the conventional wisdom has been that whichever candidate dominates the news cycle is going to drop in the polls.
Last night’s debate fits this story almost perfectly. Both candidates drew tremendous attention from Twitter, getting mentioned in 30-50% of tweets by our group members. Most of this attention was negative, as neither candidate drew significantly positive sentiments from any of the 8 groups we tracked, and both drew negative attention from multiple groups.
Negative Sentiment
They did not draw this negative attention equally, however. In particular, while Hillary Clinton did not receive positive sentiments from any group, she came very close with two groups (Defectors to Hillary, Democratic Candidate Avoiders). In addition, only 2 out of the 8 groups directed expressed relatively negative sentiment in tweets that mentioned her. These groups were the Trump Dumpers and Defectors to Trump, both of whom were consistently negative in their tweets about her in our past data.
In strong contrast, Trump received almost universally negative tweets. Our data show that Trump received tweets with relatively negative sentiment in 7 out of the 8 groups we tracked. Only the Trump Dumpers did not express relatively negative sentiment in tweets about Trump. This appeared to be because, as loyal GOP members who are not particularly favorable to Trump, they devoted the bulk of their attention – 47% of their tweets – to Hillary, expressing significant negative sentiment.
Attention
The attention story on its own also looks bad for Trump. Attention tonight was largely negative attention. Each of our 4 Democratic groups devoted substantially more attention to Trump than to Clinton. Similarly, our 4 Republican groups tended to devote more attention to Clinton than to Trump. Furthermore, getting substantially more attention from a group was associated with negative sentiment.
In this context, the problem with Trump’s performance is the fact that he drew almost as much attention as Hillary did from Republican groups. While Democratic groups focused substantially more (50-100%) on Trump, Trump’s mentions were very close to Hillary’s in two GOP groups (Candidate Avoiders, Trump Avoiders). Both of these groups expressed significantly negative sentiment about Trump but were neutral about Hillary.
GOP group members were also quieter, in general, than they had been in the past. Members of our GOP groups were less likely to tweet during the debate than they were during the RNC, while members of our Democratic groups were more likely to tweet than they were during the DNC.GOP group members were quiet early in the debate, and seemed to “head for the exits” early as well, with both the GOP Candidate Avoiders and the Trump Avoiders sending fewer tweets in the last half hour of the debate than they did in the 2nd half hour. All other groups increased in their tweets over each half hour period.
Key Phrases
Hillary’s “prepared to be president” claim drew many cheers from many members of Democratic groups. However, the phrases that drew the most consistent attention across the groups were Trump’s “stop and frisk” and “law and order” promises. Interestingly, these drew many serious comments from both sides that typify the divide in this country. Members of Democratic groups derided “stop and frisk” as racist, ineffective, and unconstitutional. And while they did make many snarky references to “Law and Order” the television show, they also criticized “law and order” as code for police harassment of minorities. Members of GOP groups, however, defended the policy sincerely. They asserted that “stop and frisk” had been effective in New York and that law and order was something the country badly needed.
Overall, these tweets must be interpreted in the context of the race itself. To the extent to which Trump still needs to attract more voters, there was no evidence from our Twitter groups that he was successful in doing so. Whether the negative sentiments expressed in tweets about him actually cost him votes, though, is hard to say.
Note: if you’d like to monitor how are groups are responding throughout the campaign, check out our interactive tool at debatemeter.com.
For media interviews contact:
Rebecca Valli
Office: 607-255-7701
Cell: 607-793-1025
rv234 [at] cornell.edu
Ellen Leventry
Office: 607-255-2722
Cell: 607-793-1441
eel2 [at] cornell.edu
