Exploratory Data Analysis of the Pre-Election Polls of the European Election 2019 as well as an Evaluation of Their Predictive Power
The relevance of election polls increases steadily since they can be used to forecast the outcome of elections. Forecasting the vote shares of the parties participating in an election can demonstrate shifting voting behavior for instance. Hence, political parties can use this information to react appropriately and adapt their campaign strategies and political programs to the current sentiment of (their) voters. The closer the election day gets, the smaller the lags between polls because of the increasing political and public interest [Huber, 2018].
In order to conduct a poll as representative as possible, different collection methods can be applied. Online panels and phone surveys which also have been used for the collection of the underlying data are just two examples among others. However, the available data is not very robust since the number of state-wide conducted election polls per election for the past European elections was smaller than 30 and many variables, such as the time and the collection methods varied (cp. Europawahl [2019]). This affects the robustness of the distribution as well as the normality which is assumed by many analysis methods that can be used for testing equality in mean and forecasting for instance.
The objective of this project is to present and discuss methods that can be used to transform non-normal distribution. Additionally, a test to evaluate the forecasting power of the polls that were realized during the pre-election period is conducted.
The data used for the analysis was collected by myself from the website Wahlrecht.de. The first observation was recorded on the 25th of October 2018 and the last on the 24th of May 2019.
Organization
Author: Anna Franziska Bothe
Institute: Humboldt University Berlin, Ladislaus von Bortkiewicz Chair of Statistics
Course: Data Analysis I
Semester: SS 2019
Content
.
├── HA-DAI_ABothe.pdf # PDF of final paper
├── HA-DAI_ABothe.Rmd # contains the final code as well as the texts
├── Data # folder contains the data and files that are needed to run the markdown file
├── README.md # this readme file
├── requirements.txt # contains used libraries
├── setup.txt # describes execution of pipeline in detail