Group 5 Codebook: School Food Assistance Program RCT

Study Overview

This dataset contains results from a randomized controlled trial evaluating the impact of a school food assistance program (breakfast, lunch, and snacks) on student academic performance and behavioural engagement. The study was conducted across schools in the Toronto District School Board.

Important note on randomization: The original study design proposed cluster randomization at the school level, with entire schools assigned to treatment or control conditions. However, to simplify the analysis for this course, the simulated dataset uses individual-level randomization, where each student was independently assigned to treatment or control. You should analyze this data accordingly (i.e., you do not need to account for clustering in your analysis).

Sample

• N = 800 students
• Students span grades 3, 6, 9, and 10
• Schools serve predominantly lower-income communities

Variables

Variable	Type	Range	Description
student_id	Integer	1–800	Unique student identifier
grade	Integer	3, 6, 9, or 10	Student’s grade level
treatment	Binary	0 or 1	Treatment assignment. 1 = received the food assistance program (breakfast, lunch, and snacks provided daily); 0 = control (no food program)
eqao_score	Continuous	0–100	Student’s EQAO standardized test score (percentage). This is the primary academic outcome.
behaviour_term1	Continuous	1.0–5.0	Teacher-rated behavioural engagement score at the beginning of the school year (Term 1). Measured on a Likert-type scale where 1 = very disengaged and 5 = very engaged.
behaviour_term2	Continuous	1.0–5.0	Teacher-rated behavioural engagement score at the end of the school year (Term 2). Same scale as behaviour_term1. Note: some values are missing.
socioeconomic_index	Continuous	1.0–5.0	Composite index of household socioeconomic status. Higher values indicate higher socioeconomic status. Derived from parental income, education, and neighbourhood indicators.
baseline_nourishment	Continuous	1.0–5.0	Self-reported baseline nourishment adequacy at the start of the study, before intervention. Higher values indicate better baseline food security and nutritional intake.
attendance_rate	Continuous	0–100	Percentage of school days attended during the study period.
home_experience	Continuous	1.0–5.0	Composite rating of the student’s home learning environment, including parental involvement, access to learning materials, and study space. Higher values indicate a more supportive home environment.
requested_snack	Binary	0 or 1	Whether the student requested additional snacks beyond the standard program offerings. Only applicable to students in the treatment group (treatment = 1). Values are missing (NA) for control group students, as they were not part of the food program.

Missing Data

• behaviour_term2 has approximately 8% missing values. These represent cases where the end-of-year teacher rating was not completed (e.g., teacher turnover, student transfers mid-year, or incomplete forms).
• requested_snack is NA for all control group students by design, as this variable only applies to participants in the food program.

Measurement Notes

• EQAO scores are based on Ontario’s Education Quality and Accountability Office standardized assessments. Scores are reported as percentages.
• Behavioural engagement ratings were completed by classroom teachers using a validated observation-based rubric at two time points (beginning and end of school year).
• Baseline nourishment was collected via student/parent survey before the intervention began.
• Socioeconomic index was derived from administrative records and census-linked neighbourhood data.

{r} library(readr) data <- read_csv("group5_data.csv") head(data) {r} control_data<-data%>% filter(treatment==0) control_data {r} treatment_data<-data%>% filter(treatment==1) treatment_data