The Correlation Between Twitter Sentiments And Polling Results For The 2016 Presidential Race /

Dev Ramesh

School Name

Spring Valley High School

Grade Level

10th Grade

Presentation Topic

Psychology and Sociology

Presentation Type

Non-Mentored

Oral Presentation Award

1st Place

Abstract

The correlation between Twitter sentiments and polling results for the 2016 presidential race / Dev Ramesh / Behavioral Sciences / / The 2016 presidential race is beginning to attract a large amount of attention as the election draws near. Due to voter curiosity as to which candidates are preferred more by the general populous, an effective tool to gage the rankings of presidential nominees was sought out. The prevalent technique being employed currently is polling, an assessment of public opinion obtained by surveying a population sample. While effective in determining public opinion, this method is not cost efficient and is very time consuming. To replace this unpractical procedure, the implementation of Twitter sentiments to determine the general emotion towards each candidate has been proposed. The objective of this research was to establish the correlation of the sentiments from Twitter to the presidential polling results to suggest it as an alternative. It was hypothesized that Twitter sentiment would have a positive correlation with polling percentages. To collect the Twitter data for each candidate, the plugin NodeXL for Microsoft Excel was utilized, and to calculate the sentiment, the plugin Text2Data also for Microsoft Excel was used. Polling data was collected from realclearpolitics.com. This research found that positive changes in sentiment generally corresponded to positive changes in polling percentage, and negative changes in sentiment generally equated to negative changes in polling percentage. The two variables were strongly correlated, r(68) = .6073, p < 0.001 suggesting that Twitter sentiments could be utilized as a substitute for polls. /

Location

Owens 108

Start Date

4-16-2016 12:00 PM

COinS
 
Apr 16th, 12:00 PM

The Correlation Between Twitter Sentiments And Polling Results For The 2016 Presidential Race /

Owens 108

The correlation between Twitter sentiments and polling results for the 2016 presidential race / Dev Ramesh / Behavioral Sciences / / The 2016 presidential race is beginning to attract a large amount of attention as the election draws near. Due to voter curiosity as to which candidates are preferred more by the general populous, an effective tool to gage the rankings of presidential nominees was sought out. The prevalent technique being employed currently is polling, an assessment of public opinion obtained by surveying a population sample. While effective in determining public opinion, this method is not cost efficient and is very time consuming. To replace this unpractical procedure, the implementation of Twitter sentiments to determine the general emotion towards each candidate has been proposed. The objective of this research was to establish the correlation of the sentiments from Twitter to the presidential polling results to suggest it as an alternative. It was hypothesized that Twitter sentiment would have a positive correlation with polling percentages. To collect the Twitter data for each candidate, the plugin NodeXL for Microsoft Excel was utilized, and to calculate the sentiment, the plugin Text2Data also for Microsoft Excel was used. Polling data was collected from realclearpolitics.com. This research found that positive changes in sentiment generally corresponded to positive changes in polling percentage, and negative changes in sentiment generally equated to negative changes in polling percentage. The two variables were strongly correlated, r(68) = .6073, p < 0.001 suggesting that Twitter sentiments could be utilized as a substitute for polls. /