Online Dissent and the Clinton Campaign

Would Hillary Clinton rather be loved, feared, or favorited?

A recent Politico article entitled, “Hillary’s Army of Trolls” by Annie Karni suggests the latter. The article notes that although Clinton has more Twitter followers than any other 2016 hopeful, the responses to Hillary’s tweets are far more polarized than responses to other candidates tweets. For example, even a benign Mother’s Day tweet like, “Hillary wants to call someone’s mom on Mother’s Day. It could be yours,” received as much ire as it did support. One responder wrote, “How about calling the moms of those killed at Benghazi? that’s a great start. Instead, it’ll probably be rich donors. #Witch.”

According to Karni, every Clinton tweet provokes inflammatory responses like this. On the one hand, trolling and negativity come with the territory of having 3.5 million followers. But on the other hand, Karni claims that this level of Twitter trolling is unprecedented: “when it comes to Clinton, the trolling is of a different scale and measure, a clear reflection of the strong feelings she provokes among her legion of detractors.”

But are people really more polarized about Clinton than they are about other politicians? Do people tweet more viciously at Clinton than they do at Obama, for example? Karni’s observations on Clinton’s Twitter detractors invite questions about whether Clinton’s twitter is substantially different from the Twitters of other politicians and why.
First of all, is Karni’s claim that Clinton receives more negative responses on Twitter than other 2016 candidates correct? To figure this out, I would take a sample of Clinton’s tweets and the tweets of every other 2016 candidate from a given timeframe. Then, I would systematically analyze the words in each tweet and the most common hashtags in each tweet. Analyzing each word would allow me to pull out language with negative sentiments. However, this approach takes the words of a tweet out of context. A tweet like “Hillary Clinton is not an evil dictator” might not register the how the word “not” negates the negative sentiment of the word “evil”.  Compiling a list of the most popular hashtags used in response to candidate’s Tweets might be a more reliable approach. Certainly, Hillary would get more derogatory hashtags related to gender like “#witch” or “#ugly”, but could that negativity really compete with that of Twitter demons like Donald Trump?

The second question that Karni’s observation invites is the roll of “trolls”. For the purposes of online political discussion, a troll is a person who interacts negatively with a Twitter campaign not because they are politically motivated, but because they like inciting online arguments. Karni suggests that any Twitter account with enough followers is going to get trolled. But how many followers does a polititican need before the trolls come a knockin’? I suspect that as a Twitter campaign gains followers, a larger and larger percentage of those followers are trolls. Thus, as a Twitter campaign grows, I suspect that it will have a larger and larger amount of negative “noise” that acts as a baseline of negative sentiment. As this “noise” grows, there may be a higher percentage of negative Tweets, but not a higher level of negative sentiment. To quantify this, one could compare the percentage of negative Tweets aimed towards a candidate to a candidate’s approval ratings. The difference between these two values could be attributed to negative “noise” from trolls.

The beauty of this article is that it shows the strengths and weaknesses of social media as a tool for measuring political attitudes. Karni may observe that Hillary Clinton gets more online hate than her fellow candidates, but the polls may not mirror that hate. Whether or not Clinton can get some meaningful data out of her social media may determine her fate in 2016.

To read Karni’s article on Clinton’s Twitter presence, check it out at Politico: http://www.politico.com/story/2015/05/hillary-clintons-twitter-trolls-118079.html