Valresultat ur ett regressionsperspektiv
(2025) STAH11 20242Department of Statistics
- Abstract (Swedish)
- The following study explores the economical and political determinants influencing the Swedish parliamentary election outcomes, through the lens of statistical modeling. With analysis of election data from 1970-2022, we employed both multiple regression and ridge regression to explore the following key variables: immigration, taxation, unemployment, inflation, poll performance, incumbency, marginal tax and crime rates.
Our results indicate significant differences in the predictive power of the models across parties with the party specific variables poll performance and incumbency constantly emerging as the most powerful predictors. While more standard regression tactics provided initial good insights, addressing multicollinearity through... (More) - The following study explores the economical and political determinants influencing the Swedish parliamentary election outcomes, through the lens of statistical modeling. With analysis of election data from 1970-2022, we employed both multiple regression and ridge regression to explore the following key variables: immigration, taxation, unemployment, inflation, poll performance, incumbency, marginal tax and crime rates.
Our results indicate significant differences in the predictive power of the models across parties with the party specific variables poll performance and incumbency constantly emerging as the most powerful predictors. While more standard regression tactics provided initial good insights, addressing multicollinearity through variable exclusion and establishing strong simple models, we wanted to explore further, utilizing ridge regression, creating even stronger although more complex models.
Residual analysis revealed challenges with meeting the assumptions for effective linear regression. To mitigate these problems we initiated logarithmic transformations of the independent variables. Through Akaike Information Criterion (AIC) minimization we were guided towards an optimized balende between complexity and predictive accuracy.
The conducted research contributes towards understanding voters behavior dynamics and offers a strong framework for future election analysis. However, the study also acknowledges its limitations, excluding newer political parties and the inability to account for unpredictable external events. Further studies may benefit from additional variable exploration and different regression techniques. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9181367
- author
- Stormby, Alexander LU and Holmin, Hugo LU
- supervisor
-
- Jonas Wallin LU
- organization
- course
- STAH11 20242
- year
- 2025
- type
- M2 - Bachelor Degree
- subject
- keywords
- Regression, Politics, Statistical modeling, election
- language
- Swedish
- id
- 9181367
- date added to LUP
- 2025-02-11 10:31:07
- date last changed
- 2025-02-11 10:31:35
@misc{9181367, abstract = {{The following study explores the economical and political determinants influencing the Swedish parliamentary election outcomes, through the lens of statistical modeling. With analysis of election data from 1970-2022, we employed both multiple regression and ridge regression to explore the following key variables: immigration, taxation, unemployment, inflation, poll performance, incumbency, marginal tax and crime rates. Our results indicate significant differences in the predictive power of the models across parties with the party specific variables poll performance and incumbency constantly emerging as the most powerful predictors. While more standard regression tactics provided initial good insights, addressing multicollinearity through variable exclusion and establishing strong simple models, we wanted to explore further, utilizing ridge regression, creating even stronger although more complex models. Residual analysis revealed challenges with meeting the assumptions for effective linear regression. To mitigate these problems we initiated logarithmic transformations of the independent variables. Through Akaike Information Criterion (AIC) minimization we were guided towards an optimized balende between complexity and predictive accuracy. The conducted research contributes towards understanding voters behavior dynamics and offers a strong framework for future election analysis. However, the study also acknowledges its limitations, excluding newer political parties and the inability to account for unpredictable external events. Further studies may benefit from additional variable exploration and different regression techniques.}}, author = {{Stormby, Alexander and Holmin, Hugo}}, language = {{swe}}, note = {{Student Paper}}, title = {{Valresultat ur ett regressionsperspektiv}}, year = {{2025}}, }