7 Midterm: Exercise & Sleep

Data import • Cleaning • Merging • Visualization • t-tests • ANOVA • Post-hoc tests • Recommendation

8 Overview

You are a researcher in sleep expert Matthew Walker’s lab. You have just run an experiment measuring the effect of different exercise methods on sleep habits.

You measured:

Average hours of sleep before the experiment
Average hours of sleep after the experiment
Sleep efficiency at the end of the experiment

Dr. Walker needs you to identify which (if any) exercise method best improves sleep, and to make a clear, evidence-based recommendation.

This midterm assesses your ability to build a fully reproducible analysis from raw data through statistical inference and final reporting.

9 Learning Objectives

By completing this midterm, you will be able to:

Import data from a multi-sheet Excel workbook
Clean messy labels and standardize categorical variables
Merge datasets into a single analysis-ready table
Engineer derived variables and handle missing data responsibly
Produce descriptive statistics and professional tables
Create clear, interpretable visualizations
Conduct and interpret t-tests and ANOVAs
Run and interpret Tukey post-hoc comparisons
Synthesize results into an actionable recommendation

10 Textbook Connection

This assignment builds directly from Chapters 1-5 (and 10) in Reproducible Research Using R.

Students are encouraged to review the chapter before beginning this assignment, as it provides the conceptual foundation and reproducible workflow demonstrated here.

11 Submission Instructions

Submit two files:

Your .qmd file
Your knitted and published .html file

Your report must:

Knit/render without errors
Include short narrative text between code chunks
Use clear section headers matching the numbered tasks below
Suppress unnecessary warnings/messages in the final HTML
Include readable plots and (recommended) formatted tables (e.g., kable())

12 Midterm Tasks

12.1 1. Setup & Data Import

12.1.1 Tasks

Load all packages you use
Import both sheets from midterm_sleep_exercise.xlsx using readxl:
- participant_info_midterm
- sleep_data_midterm
Preview both datasets using appropriate functions (e.g., glimpse() or head())

12.2 2. Merge & Base Cleaning

12.2.1 Tasks

Clean column names so they are consistent and easy to work with
Standardize labels for:
- Exercise_Group (e.g., “cardio”, “CARDIO” → “Cardio”)
- Sex (e.g., “m”, “MALE” → “Male”)
Merge the two datasets into a single dataset

12.2.2 Requirements

Final dataset should be one row per participant
Dataset should include demographics + sleep variables
Include a checkpoint verifying row count and that the merge behaved as expected

12.3 3. Create Derived Variables

12.3.1 Tasks

Create a numeric sleep-change variable:
- Sleep_Difference = post_sleep - pre_sleep_num
Create AgeGroup2 using case_when() with exactly two groups
(example: < 40 vs >= 40)
Identify how many rows are missing Sleep_Difference
Drop rows with missing Sleep_Difference

12.3.2 Requirements

Include a checkpoint showing how many rows were removed and why
Your narrative should briefly justify your missing-data decision

12.4 4. Descriptive Statistics

12.4.1 Tasks

Report descriptive statistics for:

Sleep_Difference (overall)
Sleep_Efficiency (overall)

For each variable, include:

Mean
SD
Min
Max

Then report group-wise summaries by Exercise_Group for:

Sleep_Difference
Sleep_Efficiency

12.4.2 Requirements

Group summaries must be clearly readable
Tables should be formatted professionally using kable() and appropriate captions.

12.5 5. Visualizations (3 plots)

Create three plots:

Boxplot: Sleep_Difference by Exercise_Group
Boxplot: Sleep_Efficiency by Exercise_Group
Scatterplot: Sleep_Difference vs Sleep_Efficiency, with a trend line

12.5.1 Requirements

Each plot must include:

Descriptive title
Clear x- and y-axis labels
A clean theme (e.g., theme_minimal())
Appropriate captions

After each plot, write 1–3 sentences describing what the plot suggests.

12.6 6. t-tests (Two)

Run two t-tests:

Sleep_Difference ~ Sex
Sleep_Difference ~ AgeGroup2

12.6.1 Requirements

For each t-test, report:

Group means
t statistic
df
p-value

Then interpret in plain language:

Is the difference statistically significant?
Is it practically meaningful?

12.7 7. ANOVAs (Two) + Post-hoc Tests

Run two ANOVAs:

ANOVA A: Sleep_Difference ~ Exercise_Group
ANOVA B: Sleep_Efficiency ~ Exercise_Group

12.7.1 Requirements

For each ANOVA, include:

ANOVA table output
F statistic
df
p-value
A brief effect-size comment
PRE from supernova() (required)

Then run Tukey post-hoc tests:

TukeyHSD() for each ANOVA

Interpret results:

Which groups differ significantly?
Which exercise group appears to “win” for each outcome?

12.8 8. Synthesis & Recommendation

Write 4–6 sentences answering:

Considering both outcomes (Sleep_Difference and Sleep_Efficiency) and your post-hoc results, which single exercise regimen would you recommend to improve overall sleep?

12.8.1 Requirements

You may only pick one regimen
Support your recommendation with specific statistical evidence:
- F values and p-values
- Key Tukey contrasts
- Any meaningful patterns from your plots/tables

Your recommendation must be actionable and executive-friendly.

12.9 9. Reflection

Write 3–5 sentences addressing:

What was most challenging?
What did you feel confident about?
What would you do differently next time to improve your analysis or report?

13 Reproducibility Practice

This midterm evaluates not only statistical accuracy, but workflow transparency. Your report must:

Load all packages explicitly
Import both sheets programmatically
Confirm merge success with row counts
Explicitly handle missing values (no silent removal)
Use clearly labeled code chunks
Avoid hard-coding statistical values in written interpretation
Suppress unnecessary warnings/messages
Render cleanly to HTML

14 Publishing Instructions

Render your .qmd file to HTML.
Confirm warnings and messages do not appear in the final document.
Submit both your .qmd file and your published .html report.