Institutional Repository of Vilnius University
Rinkimų rezultatų prognozavimas, analizuojant socialinių tinklų turinį: 2019 m. Vilniaus miesto mero rinkimai /
Abstract
dc:descriptionThis Bachelor Thesis, tilted as “Predicting election results by analysing the content of social media: 2019 Election of Vilnius City Mayor” studies the application of social media content research method to the forecasting of Lithuanian election results. The main reason to write this paper was that Lithuanian municipality elections are still relatively unexplored. Although most predictions of election results are based on traditional polls, they often are inaccurate. Fragmentation of Lithuanian party system and high electoral volatility complicates election prediction by polls. Due to this reason, the paper suggests the alternative social media approach, which would allow us to measure the public opinion on candidates during the final days before the elections. This would result in a more accurate forecast. Using the “wisdom of crowds” theory and its main idea that a crowd can make a better decision than individuals, this paper states that elections can be predicted based on public opinion on candidates in social media. Public opinion is understood as separate opinions of independent individuals aggregated altogether. The method of quantitative content analysis was employed to analyse the public opinion of the candidates for Vilnius city mayor in social networks. With regard to this method, the analysis encompassed 387 posts and comments about the candidates, which were published in Facebook on 1st-3rd March 2019, and 408 posts and comments, which were published in October. With the help of linear regression, we tested the relationship between positive comments on candidates in social media and their vote share in elections. After conducting this analysis, it was observed that if the analysis is performed on the last days before the election, even the number of mentions correlate with the candidate’s vote share. If we change independent variable from number of mentions to the positive comments share of a candidate, we get an even more accurate forecast. If we try to perform the analysis based on social media 6 months before the elections, it is completely impossible to predict the elections. The thesis also includes the comparison between the accuracy of forecasts based on social media and the accuracy of traditional survey predictions. Long period before the elections determines, that traditional surveys are more efficient way to predict elections, they may not predict the candidate’s vote share, yet they may forecast the winner of elections. However, the advantage of social media content analysis is that it can be conducted on the last days before the election, which leads to much greater accuracy than traditional surveys.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2019
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Vaicekauskaitė, Dalia,
- Contributors dc:contributor
-
- Jastramskis, Mažvydas
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD88375357&prefLang=en_US