Written by: Josh Rosenberg
Primary Source: Joshua M. Rosenberg – November 10, 2016
How folks were talking about the United States Presidential election forecast (with sentiment analysis) – Joshua M. Rosenberg
Out of an interest in how folks thought and were talking about the United States Presidential election forecast, I used TAGS to track tweets including both the words “forecast” and “election”, “model” and “election”, “predict” and “election”, and tweets mentioning the Twitter account for Five Thirty Eight, and those of a few other forecast-related accounts.
All the data and a file to do the analysis in R are here.
On Tuesday (day of the election) and Wednesday (day after the election), there were more than 200,000 tweets.
I used a new R package for sentiment analysis and looked at the percentage of positive and negative tweets and those possibly reflecting other emotions.
People seemed to be expressing a lot of “trust” – in the forecasts of the election – the morning of the election. Starting around 9:00 pm and peaking at 11:00 pm, folks started to express a lot of “surprise”. Over the next day, both negative and positive tweets seemed to become more common in tweets, until tweets expressing more positive emotions – perhaps coming to terms with the election.
So, pretty interesting, perhaps, but not telling too much without looking more at the tweets.
Latest posts by Josh Rosenberg (see all)
- A Shiny interactive web application to quantify how robust inferences are to potential sources of bias (sensitivity analysis) - January 19, 2018
- Outcomes from a self-generated utility value intervention in science (in IJER) - December 30, 2017
- Review of ‘What’s Worth Teaching: Rethinking Curriculum in the Age of Technology’ - November 7, 2017