Making the case

Research Skills — Week 4

Recap

Recap

• Writing for a policy audience
• Write a summary
• How to lie with statistics
• “What’s missing?”
• Effective figures for a briefing
• The briefing assignment
• Wrap-up

The journey so far

• Week 1: Does biomass help reach net zero? (Curiosity)
• Week 2: The numbers are interesting (Discovery)
• Week 3: The answer depends on the assumptions (Doubt)

Week 4: Write it up. Make the case. And learn to spot when someone else isn’t making it honestly.

w01-ppdac-ring.R

• [11:05–11:10] Frame the arc. Students have gone from curiosity to evidence to uncertainty. Now: communication and judgement.
• Point at the ring: where this week sits in PPDAC (Week 1, “The cycle”).

Questions?

Submit questions:

PollEv.com/geol

text geol to 07480 781235

• Open the Poll Everywhere Q&A activity before the session and leave it active throughout.
• Questions appear in your Poll Everywhere presenter view; upvoted questions rise to the top.
• Address during transitions or at the end of concept blocks, not mid-explanation.
• Reset (archive) the activity between sessions so questions don’t carry over from the previous week.

Writing for a policy audience

• Recap

Writing for a policy audience

🎓 Concept block 1

• Write a summary
• How to lie with statistics
• “What’s missing?”
• Effective figures for a briefing
• The briefing assignment
• Wrap-up

Who is your reader?

A policy-maker: busy, not a specialist, needs to make a decision.

What they need:

• A clear bottom line
• Evidence they can trust
• Honest acknowledgement of uncertainty

• [11:10–11:30] Writing for non-specialists. This is the communication strand coming to a head.

Structure of a policy briefing

Section	Purpose
Summary	The answer — first. Reader could stop here.
Evidence	Figures, data, analysis
Caveats	What’s uncertain, what depends on assumptions
Conclusions	Recommendation, with caveats attached

Compare to an academic paper: background first, answer last.

A briefing is the opposite — answer first, evidence after.

• Key principle: “Your reader should be able to stop after the summary and still know the main message.”
• The next slide is a real briefing in exactly this shape. Open it live — preview-links: auto means the link opens in-frame.

A real parliamentary briefing

Governance of solar radiation modification research — POSTnote 773, Parliamentary Office of Science and Technology, 13 July 2026.

https://post.parliament.uk/research-briefings/post-pn-0773/ · DOI 10.58248/PN773

POST’s sections	Our structure
Overview and key points	Summary — the answer, first
How the research is conducted; funding figures	Evidence
Geopolitics; ethics and legislation	Caveats
What has been tried already	Conclusions

• The point of this slide is that the four-part shape (summary → evidence → caveats → conclusions) is general.
• POSTnotes are peer reviewed and list their contributors at the end of the briefing. That is what “evidence they can trust” looks like in practice.

The summary test

If your reader stops after the first paragraph, do they know:

1. What you investigated?
2. What you found?
3. What you recommend?

• This is the standard for a good summary. If any of the three are missing, the summary needs work.
• Keep the three questions live for the rest of the session. The two exemplar reports are built so that this test separates them: the faithful summary answers all three, the traitor’s answers only the middle one — loudly. Students who apply the test catch it; students who read for tone do not.

Write a summary

• Recap
• Writing for a policy audience

Write a summary

💬✏️ Exercise 1

• How to lie with statistics
• “What’s missing?”
• Effective figures for a briefing
• The briefing assignment
• Wrap-up

10 minutes

Write a 3–4 sentence summary of your biomass findings, aimed at a government minister.

Then we’ll share a few.

Is the message clear? Is the uncertainty honest? Would a minister know what to do?

• [11:30–11:45] Students write, then 2–3 volunteers share. Class gives quick feedback using the three questions above.
• Have them type it under Summary in the briefing.md they started in Week 3 (github.com → pencil icon → Commit changes), so it’s waiting for them in the application session.

How to lie with statistics

• Recap
• Writing for a policy audience
• Write a summary

How to lie with statistics

🎓 Concept block 2

• “What’s missing?”
• Effective figures for a briefing
• The briefing assignment
• Wrap-up

• To avoid statistical mistakes or sleights of hand, we must recognize them.
• One way to recognize them is to make them - deliberately!

Same data – two conclusions

Two reports present the same data in different ways.

Both are technically accurate.

One is honest. The other is not.

As a desk, annotate: - What does each report do well? - Where is it dishonest?

Links: Pro-AI | Anti-AI

• [11:45–12:10] This is the exemplar reveal. Use the faithful and traitor AI environmental impact reports.
• Take your time. Let students react. The moment they realize the traitor report “feels” more convincing is pedagogically powerful.

Two exemplar reports

Faithful report

Balanced evidence. Honest caveats. Uncertain where the data are uncertain.

Traitor report

Selective evidence. Misleading framing. Confident where it shouldn’t be.

Ardross Castle, where The Traitors is filmed.
Derek Spence / Geograph, CC BY-SA 2.0

Let’s look at the techniques.

• Show both reports (or excerpts) side by side. Students read briefly, then you walk through the techniques.

Technique 1: Cherry-picking

Choose the number that tells your story.

Example: AI energy per query — 0.3 Wh (a Google search) vs 18.9 Wh (a complex AI task). Same topic, 60× difference, depending on which you cite.

• The traitor report cited the smaller number to downplay AI’s energy cost. The faithful report cited the range.

Technique 2: Invalid chart type

A pie chart of things that don’t form a whole.

w02-double-counted-pie.R

Or a bar chart where the y-axis doesn’t start at zero.

w02-truncated-axis.R

The chart is “technically correct” but visually lies.

• Both of these are the Week 2 “spot the lie” charts, unchanged. Say so: the point of a callback is that they diagnosed these themselves two weeks ago and are now being asked to imagine drawing one on purpose.

Technique 3: Cross-scale conflation

Compare a small thing to a big thing without adjusting the scale.

Example: “Global AI uses less electricity than UK households.”

Both are true numbers — but the comparison is meaningless without matching the scales (global vs national).

Technique 4: Omission

What you don’t show matters as much as what you do.

Example: Omit the supply chain, the payback period, or the alternative scenario. The remaining evidence looks cleaner — and more convincing.

• Don’t reveal all the techniques in the teaching notes — save some for the application session reveal, when students have had a chance to spot them during peer review.

The lesson

The most dangerous misinformation is technically sourced.

You can’t just fact-check the numbers — you have to audit the choices.

• This connects everything: Week 1 (falsifiability), Week 2 (“Compared to what?”), Week 3 (“What assumptions?”).
• The skills for spotting deception are the same skills for avoiding it accidentally.

“What’s missing?”

• Recap
• Writing for a policy audience
• Write a summary
• How to lie with statistics

“What’s missing?”

💬 Exercise 2

• Effective figures for a briefing
• The briefing assignment
• Wrap-up

Spot the omission

w03-emission-factors.R

What’s been left out? How does the omission change the impression?

• [12:10–12:20] Small-group discussion. Every number on this chart is correct and sourced. What is missing is the supply chain and the payback period — the two columns of emission_factors.csv this view does not touch.
• The sting: this is their own Week 2 chart, and they met the fuller picture in Week 3. The omission they are diagnosing is one they made.

Attendance

Show the QR from https://attendance.durham.ac.uk/mytimetable, then ← back to the task.

Effective figures for a briefing

• Recap
• Writing for a policy audience
• Write a summary
• How to lie with statistics
• “What’s missing?”

Effective figures for a briefing

🎓 Concept block 3

• The briefing assignment
• Wrap-up

Principles (revisited)

1. Every figure needs a “so what” — what should the reader take from it?
2. Don’t show everything. Show the thing that matters.
3. Captions tell the story, not just the axes.
4. Label directly rather than relying on legends.

• [12:20–12:40] Practical guidance for the figures going into the briefing. Show before/after examples.

Before

w04-figure-before.R

1 So what? 2 Show what matters 3 Captions tell the story 4 Label directly

• Show this one on its own and say nothing about it yet. Let them look.
• Every number on it is correct. That is the whole point.
• The four principles pinned along the bottom are the same ones from two slides ago, in the same order and the same spot — point at the number, not just the words, so it reads identically on the After slide.

After

1 So what? 2 Show what matters 3 Captions tell the story 4 Label directly

w04-figure-after.R

• NEXT SLIDE invites students to discuss what changed.

Before and after

	Before	After

Same data, same file. What changed — and which change did the most work?

Series	All six fuels	The four largest by mean over 2000–2025
Legend order	Alphabetical by column name — nothing to do with size	Largest mean first
Y axis	twh	Generation (TWh)
Words	None	A subtitle saying what to look at; a named source
Panel	Default grey	White, faint gridlines

w04-figure-before.R · w04-figure-after.R

• Spend the time on the first row. Ranking the fuels by their mean over the whole period rather than by the latest year is not a cosmetic choice. Rank on 2025 and coal drops out — 0 TWh, sixth place — and the headline becomes “wind has nearly caught gas”, a 7 TWh gap.
• Rank on the mean and coal is back in second place, and you can see the thing that actually happened: coal and wind swapped, around 2015. In 2024, the last year DUKES records any coal (2.0 TWh), wind out-generated it roughly forty times over.
• So: the honest chart is the one that picked its four fuels by a rule chosen before looking at which story it produced. Both rules are defensible. Only one of them was written down first.
• This is the Week 2 refrain arriving in their own work: compared to what? And the Week 4 refrain: audit the choices, not just the numbers.
• Also note the importance of colour: this uses the Durham palette. My first draft swapped the colours of coal and gas - semantic linkage made this hard to read!

The briefing assignment

• Recap
• Writing for a policy audience
• Write a summary
• How to lie with statistics
• “What’s missing?”
• Effective figures for a briefing

The briefing assignment

📋🎓 Assignment and traitors

• Wrap-up

Your assignment

Write a ~2-page policy briefing answering:

“Does UK biomass power help reach net zero, and should we keep paying for it?”

Due: committed to your GitHub repo by the end of the application session.

Structure: summary → evidence (with figures) → caveats → conclusions.

• [12:40–12:50] Hand out the assignment details. This is formative — it doesn’t count toward the final grade, but it’s essential practice.

Peer review

You will each be assigned two classmates’ briefings.

One Issue per briefing — clarity, evidence, uncertainty, verdict.

The review form opens with two drop-downs a script reads:

Verdict   FAITHFUL / SUSPECT
Argues    KEEP-OR-EXPAND / KEEP-ONLY-WITH-CCS / PHASE-OUT

• Labour the two drop-downs. They are how the reveal gets built. Repos accepted before 2026-10-05 have no form: those reviewers copy the block from the Peer review template page, and a mangled answer there drops that review out of the class data.
• Verdict and Argues are separate on purpose — flag that now, come back to it in the reveal.

Two questions, not one

Do you believe it? — and, separately — what is it arguing?

A briefing can reach the conclusion you already hold and still be a traitor’s.

• This is the whole reason Argues exists as a field. Students reliably conflate “I agree with the conclusion” with “this is honest”, and the reveal chart is built to show them doing it.

The traitors

Some of you will write deliberately misleading briefings.

You are not choosing. If you are a traitor, you have already been told — privately, and only you were told.

Everyone else: read carefully. Not everything is what it seems — including the things you agree with.

• Traitors are allocated, not volunteers, and notified privately before the session. Do not state the traitor rate: not knowing how many there are is part of the exercise.
• Nobody will be asked to identify themselves, in this session or the next. Say so — it lowers the stakes and improves the writing.
• Keep this light. The traitors are doing the class a favour.

Wrap-up

• Recap
• Writing for a policy audience
• Write a summary
• How to lie with statistics
• “What’s missing?”
• Effective figures for a briefing
• The briefing assignment

Wrap-up

Any questions we missed?

Submit questions:

PollEv.com/geol

text geol to 07480 781235

• Continuation of the same Poll Everywhere Q&A activity opened at the start of the session — this is not a new poll, so don’t reset/archive it yet.
• Surface any questions that have come in since the early Questions slide, especially ones raised during today’s content.
• Once you’ve finished with this slide, archive (reset) the activity at the end of the session so questions don’t carry over.

Next time

Application session: “Write, review, reveal”

Three phases:

1. Write your briefing (45 min)
2. Review two others’ (30 min)
3. The reveal (25 min)

Come ready to write. Your figures from Weeks 2–3 are your evidence.

The reveal is built from the reviews you file — so file them properly.

• Check the Q&A — surface any top-voted questions before finishing.
• Students should have their scenario analysis and figures ready to reuse. The application session is about synthesis and writing, not new analysis.