🌍 Analysis of Confidence in Social Organizations

This project analyzes World Values Survey (WVS) Wave 7 data in R to investigate how individual characteristics across countries predict confidence in social institutions. The analysis focuses on Romania and compares its patterns with all other countries combined and with a cluster of socio-economically similar nations.

Dataset

• WVSExtract.csv
• Source: WVS
• Personal Subset (randomly sampled with seed = 33270961):
  • Observations: 50,000
  • Features: 30 variables, including country, demographic attributes, and belief-based factors
  • Targets: 10 institutional confidence measures

Technologies Used

• Language: R
• IDE: RStudio
• Packages: dplyr, ggplot2

Objectives

The main goals of this analysis are to:

1. Perform descriptive analysis of the data and understand its structure.
2. Compare participant responses between Romania and all other countries combined.
3. Identify how well participant attributes predict confidence in social organizations.
4. Cluster countries using external socio-economic indicators to find those similar to Romania and assess how predictors of confidence compare within that group.

Institutions Studied

The following institutions were analyzed, identified by columns in the dataset prefixed with "C":

• Religious institutions
• Governments
• Elections
• Police
• Trade unions
• Courts
• Banks
• Major companies
• Environmental organizations
• Universities

Project Workflow

1. Descriptive Analysis

• Examined dataset dimensions, variable types, value ranges, missing data, and invalid responses.

2. Preprocessing

• Replaced invalid values with NA to ensure accurate analysis.

3. Focus Country (Romania) vs. All Others

• Compared Romania's participant responses against all other countries combined.
• Fitted separate linear regression models for Romania and the pooled group of other countries to predict confidence in institutions using participant attributes.
• Identified the most significant predictors of confidence in institutions.

4. Focus Country vs. Cluster of Similar Countries

• Clustered countries using hierarchical clustering (Euclidean distance, Ward.D2 linkage) based on 11 external indicators:
  • CO₂ emissions
  • Democracy index
  • LGBT equality index
  • Fertility rate
  • GDP per capita
  • Healthcare expenditure
  • Internet usage
  • Life expectancy
  • Religious composition
  • Average years of schooling
  • Unemployment rate
• Fitted linear regression models for the group of countries clustered with Romania (excluding Romania itself).
• Evaluated how well participant attributes within the cluster predicted confidence in social institutions.
• Compared model performance and predictor patterns across Romania, the cluster of similar countries, and the pooled group of other countries.

Key Findings

Romania-Specific Findings

• Confidence in religious institutions, governments, and elections was best predicted by the participant attributes.
• These models had the highest adjusted R², indicating stronger fit and explanatory power.

Strong and Consistent Predictors

• VPolitics (political interest), VReligion (religiosity), and TNeighbourhood (trust in neighbors) were the most significant predictors across Romania, its peer cluster, and the pooled group of other countries. Each showed a positive effect on confidence in social organizations.

Effectiveness of Clustering

• Regression models fitted to Romania produced performance and predictor patterns that more closely matched its peer group than those of all other countries combined.
• This confirms that the clustering approach was effective, enhancing the relevance and interpretability of the analysis by grouping countries with similar socio-economic and cultural contexts.

How to Run

1. Clone the repository or download the ZIP file from GitHub.
2. Open the project folder in RStudio.
3. Run the R script (wvs_confidence_analysis.r) inside the RStudio environment.

Author

Developed by Juan Nathan.