Statistical Analysis Project – Interlaboratory Proficiency Study

Project Overview

Designed and standardized an inter-laboratory proficiency testing protocol to ensure analytical data quality across Bunge's global laboratory network. The study aimed to evaluate precision, accuracy, and reproducibility of analytical methods used by 13 participating laboratories.

Statistical Methodology

Applied advanced statistical methods to benchmark performance and identify outliers:

Mandel's k and h statistics graph — Figure 1: Mandel's k and h statistics for consistency evaluation

Mandel's h & k Statistics: Used to assess inter-laboratory consistency (h-statistic for between-lab consistency, k-statistic for within-lab consistency).
Grubb's Outlier Test: Implemented to detect statistical outliers in the dataset.

Youden plot graph — Figure 2: Youden plot visualizing systematic vs. random errors

Youden Plots: Visualized to evaluate systematic vs. random errors.
Z-Scores: Calculated to standardize performance metrics across different tests.

Key Outcomes

Conducted data validation and performance benchmarking for 13 laboratories (expanding to 21 laboratories in 2025).
Identified critical sources of variability and recommended targeted corrective actions for underperforming labs.
Authored a comprehensive report and established standardized procedures for ongoing proficiency testing cycles.