Assessment Brief: Individual Policy Report

Overview

Weighting 30% of module grade
Format Individual policy report, 1,500–2,000 words, in Markdown
Topic Based on your group’s sustainable energy investigation (Phase 2)
Submission report.md in your report repo (geol-skills-report-<username>), tagged final
Deadline

Monday 11 January 2027, 12:00 (UK time)

What you are writing

A policy report answering your group’s research question. Imagine your reader is a government adviser: busy, not a specialist, and needs to make a decision. Your report should tell them what you found, how confident they should be, and what it means.

This is individual work. Your group collaborates on data collection and analysis; you write up independently. Your report should reflect your own interpretation, your own figures, and your own conclusions — even if you used shared data and code.

From start to submission

  1. Accept the assignment (Week 7). Open the report assignment and accept it. This creates your own private repo, geol-skills-report-<username>. Clone it with GitHub Desktop. You need to be on the module roster to accept; if you can’t, ask a member of staff.
  2. Write in report.md. Your repo already contains report.md, with a heading for each section below. Write your report in that file, and keep its name and place at the top of the repo.
  3. Make each figure with a script in figures/, and embed the PNG it saves in report.md (see Format guidelines).
  4. Commit and push as you go. At the end of the Week 9 application session, your group repo’s history is copied into your report repo, so your group’s data and code sit alongside your report. Your own files are never overwritten. After that, the group repo is read-only.
  5. Submit by tagging your final commit final and pushing the tag, by the deadline. In GitHub Desktop: History → right-click the commit → Create Tag… → final, then Push. Pushing commits without the tag does not submit anything. Check that final appears under Tags on your repo’s GitHub page. Once you tag it as final, you can’t undo!

Structure

Your report should be 1,500–2,000 words, including figure captions and excluding code. Use the structure below:

1. Summary

Summary length: 200–300 words.

The main finding, the key evidence, and your recommendation. A reader who stops here should still know the answer.

2. Introduction

Introduction length: 200–300 words.

The question you investigated and why it matters. What motivated the investigation? What is the policy context?

3. Methods

Methods length: 250–400 words.

Your data sources, analytical approach, and the statistical tests you used. Enough detail that someone could reproduce your analysis from your GitHub repo.

4. Results

Results length: 500–750 words.

Your key figures (2–3), statistical test results, effect sizes, and confidence intervals. Interpret each result in plain language. Every figure should be referenced in the text and captioned with a “so what” — not just a description of the axes.

5. Discussion and conclusions

Discussion and conclusions length: about 500 words.

What the results mean and what they don’t. Acknowledge limitations, assumptions, and potential confounders. State your policy recommendation and how confident you are in it. Be honest about uncertainty.

Format guidelines

  • Length: 1,500–2,000 words, including figure captions and excluding code.
  • File: write in Markdown, as report.md at the top of your report repo. GitHub displays Markdown with its figures, so that is exactly what your marker will read.
  • Figures: 2–3 well-chosen figures, each captioned and referenced in the text. Each figure is made by its own R script in the figures/ folder, named to match: figures/figure-costs.R creates figures/figure-costs.png. Scripts run from the top of your repo, so they read data from data/...; call set.seed() if a script uses randomness. Each script ends by saving its figure with ggsave(), e.g. ggsave("figures/figure-costs.png", width = 6, height = 4, dpi = 150) (see figures/figure-example.R in your repo). Embed the PNG with ![Alt text describing the figure](figures/figure-costs.png). Markers will re-run your scripts and check that each figure comes out the same.
  • Citations: Cite your data sources clearly. A simple inline citation is fine (e.g., “Source: DUKES 2026 Table 5.6B”). You do not need a formal reference list.

What we are looking for

This report is your opportunity to demonstrate your ability to use scientific and statistical methods to design, implement and evaluate an effective experiment, and to convincingly communicate your findings. You will use these same approaches as you undertake your dissertation research project next year.

Your overall mark will reflect your knowledge and understanding of the course content, and the degree to which you have effectively applied the skills taught on the module.

A good report will satisfy of these desiderata:

Criterion What this means
Clear research question The report addresses a specific, answerable question
Appropriate methods Statistical tests match the data and question; assumptions are checked
Honest presentation Figures are not misleading; numbers have context (“Is that a big number?”)
Effect sizes and uncertainty Results include confidence intervals or effect sizes, not just p-values
Limitations acknowledged The report says what the analysis cannot tell us
Clear communication A non-specialist could follow the argument from question to conclusion
Reproducibility Methods are described clearly; code is committed and runs

Commit history matters

Your GitHub commit history shows how your work developed over time. We expect to see commits from multiple weeks — not a single commit the night before the deadline. The commit history:

  • Demonstrates your individual contribution
  • Provides evidence of genuine engagement (not just AI output)
  • Shows iterative development — real analysis evolves

A report with no commit history, or a single large commit, will raise questions about how it was produced.

AI use

You may use AI tools to help with coding, structure, and editing. But the assessment is designed so that generic AI output is insufficient.

Whether or not you use AI tools in your report, you must:

  1. Understand everything in your report. If asked, you should be able to explain any figure, test, or conclusion.
  2. Take full responsibility for the content of your report. AI makes confident errors and writes specious text. We expect that you will take intellectual ownership for the work done and its presentation. Code or text generated by AI tools may be a useful starting point, but the final code and text must be substantively your own.
  3. Not use AI to generate data or fabricate results.
  4. Commit an AI reflections document (from Week 9) alongside your report, describing how you used AI and outlining what you accepted, rejected, and modified.

Checklist before submission