Presidential Election Debates. Third Debate First Debate Second Debate Figure 5.1: Clinton’s Normalized Positive Sentiment Annotated with Debate Dates To begin this analysis of the impact of the debates on Clinton, look first at figure 5.1, which shows the percent positive sentiment on Twitter from figure 4.5(c) with annotations pointing to the dates of the debates. Clinton receives bumps in positive sentiment as a result of each debate. A review of the news corpus indicates that this aligns with her performance in the debate. As an example, the following headlines are representative of the coverage following the first debate: Debate Takeaways: Xxxxxxx Xxxxxxx Digs In and Prevails Suburban Women Find Little to Like in Xxxxxx Trumps Debate Performance After a Disappointing Debate, Xxxxxx Xxxxx Goes on the Attack Commentators Give Xxxxxxx Xxxxxxx Edge in Debate The debates account for three of five of Clinton’s positive sentiment peaks during the collection period. Moreover, the first debate corresponded to an inflection point in her lead over xxxxx in the popular vote polling. First Debate Second Debate Third Debate Figure 5.2: The Public’s interest in the Presidential Debates Though each of the debates causes Clinton’s positive sentiment to spike, the lasting impact of each is in question. Conversation keywords from section 3.3.2 surrounding the the first debate appear in the top-6 terms for three days following the debate. However, for each of the following debates, related conversation keywords only appear in the top-6 terms the day immediately following the debate. This, along with the fact that the first event corresponds with the global maximum for Clinton’s positive sentiment, suggests that the first debate was far more important to the election than either of the other debates. Figure 5.2 provides a second source of information, specifically results retrieved from Google Trend, to verify this. Third Debate Second Debate First Debate Figure 5.3: Information Regarding Tweets Linked to C1, the Debate Cluster Compare this result to data collected using the debate news cluster, C1. Figure 5.3 displays the data from figure 4.8 alongside annotation of the debate dates. This figure is consistent with the trends in figure 5.1 and figure 5.2. Notice that, again, the first debate is discussed at a higher volume on Twitter than the second or third debates. Furthermore, all three debate data points, and the day following each, have a general low negative sentiment. This is except for the one ...
