🏷️ Food Label Readability: Cross-Sectional Analysis

Executive Summary: This repository provides the data and deep statistical modeling used to examine the cross-sectional relationships between perceived multiple traffic light (MTL) legibility and self-reported processed food consumption frequency within the UK.

🌍 Why This Matters

• Nutritional Policy Making: Validates whether the readability of front-of-pack labels (like the MTL system) actually correlates with healthier consumer dietary patterns.
• Health Equity: Evaluates complex stratifications across demographics, addressing whether label readability caters equally to varying levels of household income and education.
• Large-Scale Validation: Utilizes massive multi-year survey datasets to draw broader, national-level dietary conclusions rather than localized experiments.

📊 Dataset

The analysis is driven by robust, multi-wave multi-year data.

• Source: Food and You Survey 2010-2018 (Waves 1-5).
• Features: Label Readability scale, pre-cooked meat consumption, sandwich consumption, dairy intake, and extensive demographic control variables.
• Data File: foodandyou_survey_data.csv

🔬 Methodology Pipeline

Survey Weight Adjustments ➔ Variable Imputation & Recoding ➔ Multicollinearity Validation (VIF) ➔ Proportional Odds Checks (Brant) ➔ Ordinal Logistic Regression

🤖 Models & Analysis Benchmarked

Analytical Method	Purpose in Study
Variance Inflation Factor (VIF)	To detect and remove severe multicollinearity among the demographic and consumption predictors.
Brant Test	To definitively validate the proportional odds assumption (parallel regression) necessary for ordinal models across waves.
Ordinal Logistic Regression	The core model used to calculate the log-odds and marginal effects of readability on food frequency.

🚀 How to Run Locally

1. Prerequisites

Ensure you have R or RStudio alongside a suite of statistical modeling packages:

install.packages(c("tidyverse", "plm", "car", "gplots", "tseries", "lmtest", "readr", "texreg", "ordinal", "VGAM", "survey", "foreign", "MASS", "Hmisc", "reshape2", "brant", "ggplot2", "patchwork"))

2. Execution

1. Clone the repository locally.
2. Open label_readability_analysis.Rmd in your preferred R editor.
3. Adjust the file paths in the initial data code chunk to point to foodandyou_survey_data.csv in your cloned directory.
4. Execute the chunks sequentially, or click Knit to dynamically process the complex variance structures and generate the final publication-ready odds-ratio plots.

📁 Project Structure

The repository is purposefully flat to house the interwoven pipeline:

• label_readability_analysis.Rmd — The main analysis script detailing the full methodological pipeline.
• foodandyou_survey_data.csv — The combined mult-wave survey baseline data.

📄 Citation

If you use this code or dataset in your research, please cite the original paper:

Avalos Valdebenito, C., Shryane, N., & Wang, Y. (2026). Food Label Readability and Consumption Frequency: Isolating Content-Specific Effects via a Non-Equivalent Dependent Variable Design. Nutrients, 18(2), 197. https://doi.org/10.3390/nu18020197

---

👤 Author: Constanza Avalos-Valdebenito