About Us |
Contact Us |
LexisNexis Business Solutions
The race is tight, so coloring in the electoral map has proven to be an arduous task. Media monitoring and social media analysis may help to color in some of those states by providing a glimpse into voter sentiment and enthusiasm that polls might not capture. Take a look at a guest blog post brought to you by the Applied Policy Research Institute from Wright State University.
By James Mainord and Sarah Ammar
As our previous coverage noted -- and recapped in another brief here -- sentiment (and largely polling as well), took a sizable positive turn for Hillary Clinton in early October. This correlates especially with the final debate and with the now infamous audio-tape of Donald Trump and Billy Bush.
However, this momentum was short-lived for Clinton in the polls, especially following last week's letter sent from FBI Director James Comey to Congress regarding the investigation into Clinton's emails and we wondered if Twitter sentiment was echoing that trend.
A look at the general sentiment trend lines from October 27 to today indicate that they do. Clinton's sentiment took a significant dip around the time of the first Comey letter, sent on October 28. However, it appears to have rebounded at the outset of November. Even with that rebound though, if sentiment were a predictor, the overall closeness would be "too close to call."
This could be especially important because, if the race is tightening as indicated by this data from Cognovi and from polls and modeling like that of Nate Silver's FiveThirtyEight, then the battleground states are that much more critical.
Based on polling, early voting patterns, past electoral history and Cognovi Labs Data, this map represents the minimum electoral votes on which both candidates can likely rely on for election day.
That leaves 14 states up for grabs, or 211 electoral votes. That's enough of a margin that either candidate has a mathematical chance at securing enough electoral votes to win the Presidency.
Candidate........Likely Votes........Votes Needed
The remainder of this analysis focuses on the "toss up" states in the map, and examines Cognovi data in these states from October 27-November 4, 2016.
This chart, and corresponding table looks at these states, and what the electoral outcome would be if sentiment were both predictive and the only predictive measure.
In many races where polling and sentiment data are tight, we take a look at one layer deeper, where emotion and tweet volume can reveal additional insights. Below are some examples.
The sentiment graph below illustrates how close sentiment is in Texas. Though Clinton holds a sentiment lead, it's very slight. When sentiment doesn't provide a clear reading, emotion and posting volume data provide an additional layer of insight.
Similar to Texas, Georgia is very close when looking only at sentiment.
Emotion distribution provides insight into the level of enthusiasm. Negative emotion toward Clinton (30%) and positive emotion toward Trump (24%) comprise 54% of the emotions expressed. To further highlight Clinton’s challenge in Georgia, a higher percentage of negative emotion is expressed toward her than her percentage of tweet distribution (which also takes neutral or non-expressive tweets into account).
Cognovi believes that it’s likely that Georgia’s electoral votes go to Trump when taking into account emotion distribution, tightsentiment, Georgia’s voting history, and polling data.
North Carolina appears safer for Clinton than the other two states. She holds a full percentage point sentiment lead and the other metrics look encouraging for her.
Volume distribution in North Carolina deviates from other states as Clinton is getting a larger proportion than in previous states. This also corresponds with a positive emotion expressed toward her (12%) and negative emotion toward Trump (41%).
Though 53% of tweets expressing emotion are either anti-Trump or Pro-Clinton, enthusiasm for Trump is significant at 33%.
Sentiment in Florida is very close. Emotion and volume distribution closely mirrors Texas. 25% of tweets express positive emotion toward Trump, so there is strong enthusiasm. Conversely, negative emotion toward him is almost double at 49%.