• Sean Farrell

A Return to Civility: Twitter’s Response to the Final 2020 Presidential Debate

As we outlined in an earlier post, the Twittersphere did not react particularly well to the car crash that was the first US Presidential debate. The tit-for-tat insults and name calling generated very strong emotional responses to both candidates, in particular anger and disgust. The second (or is it third?) debate last Thursday was a completely different ball game, engendering fond nostalgia for the dull and civil presidential campaigns of yesteryear. But were the Twitterati won over by this relative return to normalcy?

As with the first debate, we scraped a tonne of tweets posted during the debate mentioning Trump or Biden to analyse emotional responses using Receptiviti’s cutting-edge emotion models, and again some really interesting patterns jump out at you. Unlike the first debate, our tweet sample has almost twice as many tweets aimed at Trump than at Biden. The volume of tweets aimed at Trump were again incredibly strongly correlated with our emotionality metrics (e.g. ‘good feel’ and ‘bad feel’), which we interpreted previously as a sign of Trump’s polarising nature. This time round though the emotionality expressed in the tweets targeting Biden were also strongly correlated with tweet volume (Figs 1 & 2). Is this an indication that Biden is beginning to energise voters in a similar fashion as Trump?


We see less emotions (both positive and negative) aimed at Biden that at Trump but the correlation with tweet volume is very interesting. There were some big spikes in good feeling for both candidates. For Trump, at his announcement that he’d like to terminate Obamacare and when he exclaimed that nobody has done more for the black community than himself (with the possible exception of Abraham Lincoln) correlated with jumps in positive feeling (Fig1). For Biden, there was a small jump in good feeling when he said he’d pass Obamacare with a public option and call it ‘Bidencare’, and more obvious spikes when he called Trump ‘Abraham Lincoln’ and proclaimed him the most racist president in modern history, and also when he rolled out the lines “we’re going to choose science over fiction; we’re going to choose hope over fear” (which sound like lyrics to some easy listening ballad).

Figure 1 – Good feeling vs time for Trump (left) and Biden (right). The grey bars indicate tweet volume targeted at each candidate, and the coloured lines indicate emotional intensity. Note: the tweet volume is scaled for visualisation clarity.

There were also some significant spikes in negative emotions for Trump, which appear to correlate with particular contentious statements (see Fig 2 and Clips 1 & 2): when he once again claimed that Obama spied on his campaign; when he said only the low IQ illegal immigrants who were released into the community turned up to court; and when he exclaimed he was the least racist person in the room. For Biden, there was a small blip in negative feeling when he sarcastically referred to Trump as Abraham Lincoln, far less than the positive emotion spike at the same time.

Figure 2 – Bad feeling vs time for Trump (left) and Biden (right). The grey bars indicate tweet volume targeted at each candidate, and the coloured lines indicate emotional intensity.


Clip 1 – Good feel, bad feel, joy, love, disgust and anger delivered by Twitter users at both candidates and compared to tweet volume (grey bars) during a discussion around immigration policy.



Clip 2 – Good feel, bad feel, joy, love, disgust and anger delivered by Twitter users at both candidates and compared to tweet volume (grey bars) during a heated exchange about race.


This latest debate generated far less anger than the first one with no obvious spikes that we can link to particular comments or discussion subjects. There was more disgust this time round (slightly more aimed at Trump), with some interesting spikes correlated with volume and around the time of some fiery clashes over immigration and race (some also correlated with spikes in bad feeling, see Fig 3). For Trump, the biggest spike in disgust was when he said he was the least racist person in the room while for Biden there were three spikes: when Trump claimed Biden referred to African-Americans as “superpredators”; when Biden sarcastically called Trump ‘Abraham Lincoln’; and when he stated that he doesn’t oppose fracking.

Figure 3 – Disgust vs time for Trump (left) and Biden (right). The grey bars indicate tweet volume targeted at each candidate, and the coloured lines indicate emotional intensity.

So, lots of negativity aimed at both candidates, just like the first debate. But what about positive emotions? There weren’t any strong correlations for the most part, but some subtle trends. For starters, admiration was pretty low for both candidates but flat for Trump yet slightly trending upwards for Biden during the debate. In contrast, amusement for Biden was pretty flat, yet trended slightly down for Trump. Could this indicate a growing respect for Biden’s performance and a growing exasperation for Trump’s antics throughout the event?

The final emotion that demands comment is love (Fig 4). Although there wasn’t a lot directed at either candidate, there were two spikes in emotion – one for each, and both at the same time – during a discussion around health care. When Trump says he wants to terminate Obamacare and replace it with a “brand new, beautiful healthcare” there’s a slight bump in love. In response, Biden proclaims “what I’m going to do is pass Obamacare with a public option, and become Bidencare”, triggering a massive spike in love. It seems that Twitter may have delivered a ringing endorsement of Joe’s new healthcare slogan.


Figure 4 – Love vs time for Trump (left) and Biden (right). The grey bars indicate tweet volume targeted at each candidate, and the coloured lines indicate emotional intensity.

  • LinkedIn
  • Facebook
  • Twitter

© RECEPTIVITI INC. 2020