Problem
The researchers tested a program aimed at giving low-income students better access to healthy school meals. They had collected a variety of data in hand-entered, inconsistent Excel files: monthly usage data, teacher surveys, and a one-day snapshot of the items served in the cafeteria.

My Approach
I used a combination of R and manual data cleaning in Excel to merge all of the data files. This included steps like finding and removing extraneous text from the data files, and adding columns to indicate school. To test for differences across schools in the program and control schools, I used mixed effects linear models.

What makes my work unique?
I didn’t just hand off cleaned data and results: I documented all of my cleaning steps in an R Markdown file alongside the code I used to process the data. These files were handed off so that anyone could reproduce my steps.

Results & Impact
The mixed effects models showed increased school meal utilization in the intervention schools. The researchers further analyzed the cleaned tray weights data to understand the impact of the new program on students’ nutrition.