I got the link from a subreddit promoting Bernie Sanders. They are quite happy that there is now evidence that the "Bernie's Angry Bros" myth has been "disproven". Based on my quotes you might assume you know my view... Read on.
Mr. Spencer wrote an article (Salon, archive) which was an interview with Mr. Winchell. Mr. Winchell ran a sentiment analysis on millions of tweets. But, I want to step back. Mr. Spencer referenced another article to show what this is attempting to disprove.
I ran into Boxer earlier this week and got to talking about a . . . report in The Times. . . . The piece briefly mentions a 2016 incident in which Boxer went to Nevada to try to unify the party after Hillary Clinton defeated Bernie Sanders in the state’s caucus. . . . Someone picked up a chair as if to throw it. “Security whispers in my ear, you’ve got to get off the stage,” Boxer said. “When I left, my heart was pounding. I’d never had that before.”
The most obvious resemblance [between Trump and Sanders supporters] is the adulation they bestow on their respective champions, whom they treat less as normal politicians than as saviors who deserve uncritical and uncompromising support.
A Twitter sentiment analysis won't answer those questions. But, here's the characterization that a Twitter sentiment analysis could analyze.
“When Mr. Sanders’s supporters swarm someone online, they often find multiple access points to that person’s life . . . More commonly, there is a barrage of jabs and threats sometimes framed as jokes. If the target is a woman, and it often is, these insults can veer toward her physical appearance.”
So, the accusation is that Sanders people a) swarm, b) attack ad-hominem, c) make threats, d) frame it as jokes, and e) attack women's appearance. I'm going to be blunt; I think this is wrong. Based on my own personal experience, this isn't the case with most Sanders supporters with whom I've interacted.
Some more details, from earlier context, the accusation is based on minor ideological impurities. And, that they get violent in groups.
But, the question is: does the data show that it's a myth? As a quick aside, I didn't find the actual paper. If someone does please send a link my way.
I downloaded all the followers of the Twitter accounts of the nine most popular Democratic presidential candidates and the president ([around] 100 million Twitter accounts). I then randomly chose followers from them and downloaded all their tweets from 2015 to the present.
I have run two different sentiment analysis algorithms on these tweets. So far, nearly 6.8 million tweets from 280,000 Twitter accounts have been analyzed out of the 100 million-plus tweets I currently have downloaded (I continue downloading more).
Tweets or replies? And, followers of the Twitter user's account? Don't many politicos follow multiple candidates? The alleged theory (without evidence) is that they swarm for minor differences and attack and make threats that are not common among other supporters.
Here's the data:
This seems very, very similar. How were the users selected, again? Followers... Right... But, following isn't indicative of support. It's indicative of interest. It might be that politicos as a group follow multiple people in this list (I know I do). And, that this is really a baseline. Furthermore, this has nothing to do with the idea of swarming, though. Mr. Winchell does present a new theory.
Essentially, his latest theory is that people swarm more because Bernie has more followers. This could make sense. But, once again, following isn't indicative of support. Otherwise, President Trump would be shoe-in against ALL of the candidates.
Therein lies the problem. This data doesn't show it as a "myth" it shows a bad process. Now, anecdotally, I disagree with the accusation. But, there is no data disproving it. There's also no data supporting it. The accusation is up to you to believe. Look at your own experience. You have to figure it out yourself, if you read the news articles, they'll just lie to you.