Introduction

Understanding voting behavior is crucial for policymakers, political analysts, and researchers. Demographic factors such as age, education level, income, and geographic location often influence voting preferences. In this research, I apply Correspondence Analysis (CA) to explore the relationship between demographic groups (Age) and voting patterns preferences during the 2022 Philippine presidential elections.

Methodology

Dataset

I use a dataset from COMELEC which provide data on election results and Statista that categorizes voters by age group, and voting preferences. The dataset consists of a contingency table where rows represent demographic groups, and columns represent voting preferences.

Correspondence Analysis

Correspondence Analysis (CA) is a dimensionality reduction technique used for visualizing categorical data. It provides insights into the relationships between categories in a contingency table.

Data Preparation

#install.packages("ca")
library(ca)  # Correspondence Analysis library
library(ggplot2)  # Visualization
# Create a contingency table
demographic_voting <- matrix(c(
  10264378, 4891277, 1197836,  
  11321920, 5395227, 1321249,  
   7931565, 3779623, 925600,  
   5598752, 2667969, 653365  
), nrow=4, byrow=TRUE)

rownames(demographic_voting) <- c("18-29", "30-44", "45-59", "60+")
colnames(demographic_voting) <- c("Marcos", "Robredo", "Pacquiao")

demographic_voting
##         Marcos Robredo Pacquiao
## 18-29 10264378 4891277  1197836
## 30-44 11321920 5395227  1321249
## 45-59  7931565 3779623   925600
## 60+    5598752 2667969   653365

Applying Correspondence Analysis

# Perform Correspondence Analysis
ca_result <- ca(demographic_voting)
print(ca_result)
## 
##  Principal inertias (eigenvalues):
##            1    2   
## Value      0    0   
## Percentage NaN% NaN%
## 
## 
##  Rows:
##             18-29     30-44    45-59       60+
## Mass     0.292294  0.322409 0.225864  0.159433
## ChiDist  0.000000  0.000000 0.000000  0.000000
## Inertia  0.000000  0.000000 0.000000  0.000000
## Dim. 1  -1.201435  0.263727 1.516102 -0.478495
## Dim. 2  -0.296620 -0.982246 0.382617  1.988082
## 
## 
##  Columns:
##           Marcos   Robredo  Pacquiao
## Mass    0.627657  0.299097  0.073246
## ChiDist 0.000000  0.000000  0.000000
## Inertia 0.000000  0.000000  0.000000
## Dim. 1  0.302009  0.236950 -3.555515
## Dim. 2  0.708532 -1.512367  0.104155

Visualization of Results

# Biplot visualization
plot(ca_result, main="Correspondence Analysis: Demographics vs. Voting Patterns", col=c("black", "red"))

Interpretation

The strong separation between different age groups and candidates indicates that age played a significant role in voting preferences. From the Correspondence Analysis, we observe the following insights: - Young Voters (18-29): Younger voters (18-29) leaned towards Pacquiao, possibly due to his appeal as a relatable figure. - Middle-aged Voters (30-44): Middle-aged voters (30-44) preferred Robredo, possibly due to her policies resonating with professionals and working-class individuals. - Older Middle-aged Voters (45-59): Older middle-aged voters (45-59) showed strong support for Marcos, suggesting possible alignment with his political stance or leadership style. - Older middle-aged Voters (60+): Senior voters (60+) exhibited a more diverse voting pattern, possibly influenced by past political experiences or long-term ideological preferences.

Conclusion

The CA plot and corresponding statistics reveal significant demographic patterns in voting preferences. Younger voters (18-29) leaned towards Pacquiao, middle-aged voters (30-44) preferred Robredo, while older middle-aged voters (45-59) were aligned with Marcos. However, the zero eigenvalues indicate that these associations may not be statistically strong, suggesting the need for further data refinement or additional variables to capture the complexity of voter behavior more accurately. Future research could integrate education, income levels, or geographic location to further explain these demographic voting patterns.

References