---
title: "Group N: your question here"
author: "Member A, Member B, Member C"
date: today
format:
  html:
    toc: true
    embed-resources: true
execute:
  echo: false
  warning: false
  message: false
---

```{r}
#| include: false
library(tidyverse)
library(gtsummary)
theme_set(theme_minimal(base_size = 12))

students <- read_csv("data/meded_students.csv")
osce     <- read_csv("data/osce_stations.csv")
```

## Background

Two sentences. What was studied, and why your question matters.

## Methods

Data from 240 students across three government medical colleges (synthetic,
generated for training). Students were allocated to a flipped classroom or
traditional lecture arm and assessed before and after with the same knowledge
test, plus a five-station OSCE.

Post-test scores were unavailable for
`r sum(is.na(students$posttest))` students, who are excluded from all
analyses of knowledge gain. No imputation was performed.

## Results

```{r}
#| label: tbl-main
#| tbl-cap: "Replace with your caption."
students |>
  select(teaching_arm, age, sex, attendance_pct, pretest) |>
  tbl_summary(
    by = teaching_arm,
    label = list(
      age            ~ "Age (years)",
      sex            ~ "Sex",
      attendance_pct ~ "Attendance (%)",
      pretest        ~ "Pre-test score"
    )
  ) |>
  add_overall() |>
  bold_labels()
```

```{r}
#| label: fig-main
#| fig-cap: "Replace with your caption."
students |>
  filter(!is.na(gain)) |>
  ggplot(aes(x = teaching_arm, y = gain, fill = teaching_arm)) +
  geom_boxplot(alpha = 0.75, width = 0.55, outlier.shape = NA) +
  geom_jitter(width = 0.12, alpha = 0.3, size = 1) +
  scale_fill_manual(values = c("Flipped classroom" = "#0f5257",
                               "Traditional lecture" = "#b8860b")) +
  labs(x = NULL, y = "Knowledge gain (post − pre)") +
  theme(legend.position = "none")
```

As shown in @tbl-main and @fig-main, ...

## Limitations

One honest paragraph. What could your design not detect? What would you need
to answer the question properly?

## AI disclosure

[Tool] was used to [specific task]. All output was reviewed, executed and
verified by the authors.

*or*

No generative AI tools were used in this analysis.