This repository contains the R code, data, and analysis for the research project investigating the influence of environmental and individual factors on students’ academic performance, as measured by GPA ratios.
The full study is available on ResearchGate.
This study models GPA ratios (bounded between 0 and 1) using beta regression, which is well-suited for bounded outcomes exhibiting skewness. The analysis reveals the relationship between study environment factors—such as ventilation, lighting, and noise levels—and academic performance.
- Data: Survey responses from 171 university students based in Milan and Rome, capturing demographic, environmental, and behavioral factors.
- Key Findings:
- Ventilated study environments and natural sunlight in heated/cooled spaces are positively associated with GPA ratios.
- The interaction between environmental distractions and ventilation highlights complex effects.
- Unexpected negative correlations, such as those between adequate desk space and GPA ratios, invite further investigation.
- Methods:
- Distribution fitting and exploratory analysis to assess the suitability of beta regression.
- Model selection using BIC to identify parsimonious relationships.
- Residual diagnostics to evaluate model validity.
| Variable | Type | Description |
|---|---|---|
| GPA_ratio | Continuous | GPA ratio (0-1) |
| Gender | Categorical | Gender of the student |
| Major | Categorical | Field of study |
| study_time | Categorical | Preferred study time (e.g., Morning, Night) |
| study_location | Categorical | Study location preference |
| natural_sun_exposure | Binary | Natural sunlight availability (Yes/No) |
| noisy_environment | Binary | Noise level in the study environment |
| heated_cooled | Binary | Adequate temperature control (Yes/No) |
| ventilated | Binary | Ventilation quality (Yes/No) |
| enough_desk_space | Binary | Adequate desk space (Yes/No) |
| often_distracted | Binary | Frequency of distractions (Yes/No) |
| study_in_group | Binary | Group study preference (Yes/No) |
-
Exploratory Data Analysis:
- Normality checks for GPA ratios and key variables.
- Distribution fitting to assess the beta distribution for GPA ratios.
-
Hypothesis Testing:
- Wilcoxon rank-sum tests for non-parametric comparisons.
- Interaction analysis of environmental and individual factors.
-
Regression Models:
- Initial linear regression attempts and diagnostics.
- Transition to beta regression with logit link for bounded outcomes.
-
Model Diagnostics:
- Residual analysis using deviance residuals.
- Investigating heteroscedasticity and model fit.
-
Significant Predictors:
- Ventilation positively influences GPA ratios.
- Natural sunlight in heated/cooled environments has interactive effects.
- Being often distracted in ventilated environments has negative effects.
-
Counterintuitive Findings:
- Adequate desk space shows a negative association with GPA ratios.
- Group study is associated with lower GPA ratios.
-
Limitations:
- Sample size limits generalizability.
- Observational nature precludes causal inference.
To reproduce the results:
-
Clone the repository:
git clone https://github.com/VanniLeonardo/Beta-Regression-Environment cd Beta-Regression-Study-Environment -
Install the required packages (handled by analysis.R).
-
Execute the analysis script:
source("analysis.R")
- analysis.R: Main script containing structured, step-by-step analysis.
If you use this work, please cite: Notaro, A., & Vanni, L. (2024). The Impact of Study Environment on Student GPA Ratios: A Beta Regression Analysis. Bocconi University Mathematical Statistics Project.
The full study is available on ResearchGate.
This project is licensed under the APACHE 2.0 License. See the LICENSE file for more details.
For any questions, feel free to contact:
- Vanni Leonardo at leonardo.vanni@studbocconi.it
- Notaro Anna at anna.notaro@studbocconi.it