10 min · Read source and task
Read the relevant weekly R practical and codebook through authorised UCL access. Identify source questions separately from original ferdo exercises.
Core instructional principle
Make the learner construct a solution: define a problem → plan steps → predict output → write code → inspect data → debug → interpret. The instructor asks questions and gives graded hints rather than taking the keyboard.
45–60 minutes of preparation
5 min · Set measurable goals
Specify three actions the learner should perform independently: e.g. join two tables correctly, summarise groups, interpret one output.
15–20 min · Rehearse code
Open RStudio; check working directory, packages, data types, variable names, joins and expected results. Never treat an unrun script as runtime-verified.
10 min · Plan questions and debugging
Write two conceptual prompts, one deliberate code or data mistake and one task using new fictional data.
10 min · Prepare follow-up
Prepare a short independent task, a 10-point formative rubric and an exit ticket. Keep original student files out of the public site.
Quick readiness check
- What object does `x <- c(2, 4, 6)` create? What does `mean(x)` return?
- Why does `mean(c(2, NA, 6))` return `NA`, and what does `na.rm=TRUE` do?
- Before joining two datasets, which keys, labels and missing values would you inspect?
- Can the student independently create and explain one small R expression?
If the first two explanations are weak, teach the full Lesson 0. If they are sound, use a 15–20 minute bridge and move to Week 1. This is an instructional decision, not a diagnosis or official score.
90-minute session plan
| Time | Phase | Evidence |
|---|---|---|
| 0–8 | Diagnostic | Ask the learner to predict `mean(c(2,4,6))` and explain `NA`. |
| 8–20 | Concept | State the research problem in natural language; define inputs, outputs and assumptions. |
| 20–37 | Paired coding | Let learner predict each line before running; explain objects and functions. |
| 37–55 | Guided practice | Learner types and checks each operation, using inspection and error messages. |
| 55–60 | Pause | Break. |
| 60–77 | Independent transfer | New tiny dataset or a variant of a Week 1–4 practical, without copying. |
| 77–86 | Debug and interpret | Find a missing-data or label error; state what the output cannot prove. |
| 86–90 | Exit and homework | Ask a short explanation and give one targeted task. |
Week-by-week priorities
| Stage | Teach | Learner demonstrates |
|---|---|---|
| 0 — Foundations | Input/output, objects, vectors, mean, missingness; use synthetic before/after scores | Explain why a mean is NA; identify four valid matched changes. |
| W1 — Familiarisation | Assignment, functions, indexing, data.frame and inspection | Write and explain an original short R script. |
| W2 — Data Wrangling | Read CSV; inner_join, filter, mutate, group_by, summarise | Audit join keys and labels; prefer codebook over conflicting notes. |
| W3 — Summarising | Pivot data, score AQ-10 screening items, calculate mean and SD | Explain why screening scores do not establish a clinical diagnosis. |
| W4 — Plots | ggplot2 mapping, geoms, scatter, histogram, violin, boxplots | Distinguish median/IQR from mean±SD and confidence intervals. |
Important data-label warnings
In the student-source codebook, occasion=0 pretest, 1 posttest, 2 follow-up, and sex=1 female, 2 male; some instructional hints reverse labels. Do not silently change the dictionary: inspect actual codebook and explain conflicting documentation. AQ-10 is screening, not diagnosis; boxplots show median/quartiles by default, not mean±SD.
Formative rubric — out of 10
| Criterion | Points | Observable evidence |
|---|---|---|
| Problem specification | 2 | Defines inputs, output, assumptions, operations |
| Independent implementation | 3 | Writes the required R expression with appropriate data |
| Debugging and data audit | 2 | Checks missingness, labels and failures |
| Interpretation | 2 | Explains the statistic and avoids unsupported conclusions |
| Reproducibility | 1 | Provides a script that runs from a clean setup |
0–4: focused scaffolding; 5–7: targeted repetition; 8–10: move to transfer. These are suggestions for instructional tracking, not UCL grading bands.
Sample 60-minute mock assessment
| Part | Minutes | Task |
|---|---|---|
| A | 10 | Predict, run and explain a short R snippet, including NA. |
| B | 20 | Transform and summarise a small new synthetic CSV. |
| C | 15 | Find a bad join, incorrect label or missing-data rule. |
| D | 15 | Make an appropriate plot; report the result and one limitation. |
The real UCL exam structure has not been established. Do not advertise this practice as an official or equivalent paper.
After each lesson
Save a short record in your own authorised system; do not publish a real student's performance publicly. Blank template below.
Date / session / source week: Three measurable goals: What the learner completed independently: Bug, data issue or interpretative issue: Hint level used: Formative rubric /10: Student explanation (private record): Next action and homework:
Continue
Public instructor desk · Lesson 0 worked solutions · R syntax guide · PALS Week 1–4 map