Sentiment Tracker - Economy, Pandemic and Race Relations by Candidate and State Grouping
CHARTS ARE AUTOMATICALLY UPDATED EVERY 3 DAYS
What we measure
Several reputed polls like the NBC News/WSJ Survey, Pew Research Center survey, NPR/PBS News Hour/Marist Poll and the CNN SSRS Poll have reported that the top three issues on voters’ minds in the upcoming elections are: the coronavirus pandemic, economy and race relations. NMDSI’s Elecurator project is therefore collecting and analyzing posts on major social media sites like Twitter and Reddit as well as on online blogs and forums about these three topics.
How we measure
We measure sentiments using VADER (Valence Aware Dictionary and Sentiment Reasoner). VADER is discussed in detail in: Hutto, C.J. & Gilbert, E.E. (2014). VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Eighth International Conference on Weblogs and Social Media (ICWSM-14). Ann Arbor, MI, June 2014. Given a post, VADER assigns a score between -4 (extremely negative) and +4 (extremely positive) to each word in the post (0 is for neutral words). These scores also take into account the context of the word. For example, the score assigned to the word “helpful” would be different based on whether it appears by itself, as part of the phrase “extremely helpful” or of the phrase “quite helpful”. The sentiment score for a post is based on the scores of its words and is calculated to be between -1 (extremely negative) and +1 (extremely positive) with zero for neutral sentiment. This score is called polarity. More details of these computations and code for the package are available here.
Goals of this analysis
Our main goal is to measure whether the overall sentiments expressed in the posts on the top three issues are positive, negative, or neutral, and how they are changing over time, in each state and the following seven groupings of states:
Strongly Republican: Alabama, Arkansas, Idaho, Kentucky, Louisiana, Mississippi, Nebraska, North Dakota, Oklahoma, South Dakota, Tennessee, West Virginia, Wyoming
Likely Republican: Indiana, Kansas, Missouri, South Carolina, Utah
Leaning Republican: Alaska, Iowa, Montana, Texas
Toss up: Arizona, Florida, Georgia, North Carolina, Ohio
Leaning Democrat: Maine, Michigan, Minnesota, Nevada, New Hampshire, Pennsylvania Wisconsin
Likely Democrat: Colorado, New Mexico, Virginia
Strongly Democrat: California, Connecticut, Delaware, Hawaii, Illinois, Maryland, Massachusetts, New Jersey, New York, Oregon, Rhode Island, Vermont, Washington
Charts you will see for each of the three topics
1. Daily comparison chart
Daily comparison of the average polarity of all the posts on that day about the topic that also reference Joe Biden with the corresponding average polarity for Donald Trump. The comparison starts with June 1, 2020. 2.
2. State grouping chart
This chart compares the average daily polarities of all the posts about each of the three topics that reference Joe Biden with the corresponding average polarities for Donald Trump, across the following seven groups of states:
Strongly Republican: Alabama, Arkansas, Idaho, Kentucky, Louisiana, Mississippi, Nebraska, North Dakota, Oklahoma, South Dakota, Tennessee, West Virginia, Wyoming
Likely Republican: Indiana, Kansas, Missouri, South Carolina, Utah
Leaning Republican: Alaska, Iowa, Montana, Texas
Toss up: Arizona, Florida, Georgia, North Carolina, Ohio
Leaning Democrat: Maine, Michigan, Minnesota, Nevada, New Hampshire, Pennsylvania Wisconsin
Likely Democrat: Colorado, New Mexico, Virginia
Strongly Democrat: California, Connecticut, Delaware, Hawaii, Illinois, Maryland, Massachusetts, New Jersey, New York, Oregon, Rhode Island, Vermont, Washington
Within each group of states the comparison is presented in the form of a Violin plot. The top and bottom of the violin show the most positive and most negative values of sentiment polarity in that group of states beginning with June 1 for each candidate. Joe Biden’s violin will be in blue and Donald Trump’s will be in red. The line going through the middle of each violin is zero (i.e., neutral) polarity. So, if the low end of the violin is further down from zero for a candidate, the more negative the average polarity of sentiment on the topic for the candidate was. The higher the top of the violin from zero, the more positive the average polarity of sentiment on the topic for the candidate.