Chapter 44 — Exercises

How to use these. No answers are provided, and that is deliberate. Almost every question below is about a future nobody has observed, which means there is no answer key in the world either. What these exercises train is not recall but the discipline of §44.1: separating what is known from what is inferred, and both from what is asserted.

Items marked † are extended work — an hour or more, often requiring you to go and find something. They are the ones worth doing if you only do a few.

Where an exercise asks you to evaluate a source, do not name private individuals in your written answer. Characterize the claim by type, as the chapter does. This is a habit worth building, not a formality.

Part A — The limits of the method (§44.1)

A. In your own words, state why a randomized controlled trial cannot answer the question "what happens to a society on metabolic drugs." Do not use the word ethics in your answer — the reason is structural, not moral.

B. Define counterfactual. Then describe what the counterfactual would have to be for the claim "these drugs changed what my country eats," and explain why it cannot be constructed.

C. A report states that in regions with the highest prescribing rates for GLP-1 receptor agonists, sales in a particular food category fell over two years. List four explanations for that observation other than "the drug caused the decline." Rank them by how easy each would be to rule out.

D. Explain the ecological fallacy using an example that has nothing to do with medicine. Then give a version of it that could plausibly appear in coverage of this drug class.

E. §44.1 lists natural experiments, interrupted time series, and difference-in-differences as honest tools for population-level questions. For one of them, describe a situation involving these drugs where it could actually be applied, and name the assumption it would depend on. Then say where you would place it on Chapter 5's rung ladder relative to a consulting firm's market projection.

F. † Find a published claim about the population-level effects of GLP-1 drugs that includes a specific number — a market size, a projected decline, a prevalence estimate, a date. Trace the number backward as far as you can: who produced it, from what data, on what assumptions. Write one page on what you found, including the point at which the trail went cold. Almost every trail goes cold. Note where yours did and what was on the other side of it.

Part B — The food industry (§44.2)

G. State the mechanism by which reduced population food intake would propagate into changes in food products. Use no more than four sentences, and make each one a causal step.

H. The chapter says the direction is predictable and the magnitude is not. Explain the difference between those two kinds of claim, and say which one a rating can be attached to.

I. Four plausible industry responses are listed in §44.2. For each, describe one observation you could make yourself — in an ordinary shop, without any data access — that would be weak evidence for or against it. Then say why each observation is weak.

J. Write the strongest possible version of the argument that these drugs will have a negligible effect on the food industry. Do not strawman it. Then write the strongest version of the opposite.

K. † Over four weeks, record the package sizes and protein content of five specific products you buy regularly. Write down today's values with today's date. This is a baseline. It will be worth more in three years than any forecast you could read today, and it took you twenty minutes.

L. §44.2 raises the possibility that reformulation changes the food environment for people who never took the drug. Name this kind of effect using the term from the Key Terms list, and give one example of the same structure from a completely different domain.

Part C — Insurance and markets (§44.3)

M. Explain, in language a non-specialist would follow, why a chronic therapy that works creates a harder budget problem than one that does not.

N. Define cost offset and budget impact, and explain why a therapy can be strong on the first and still fail on the second.

O. Draw the payer-churn diagram from §44.3 from memory. Then modify it for a health system with a single lifetime payer, and describe how the incentive changes.

P. The chapter argues that naming the churn problem changes what remedy you would look for. Explain that argument, then name two remedies that follow from the structural diagnosis and one that follows from the "cruelty" diagnosis.

Q. † Find out how coverage for this drug class actually works in your own health system: what is covered, under what indication, with what documentation, and how appeals proceed. Write one page. Then note which parts you could establish from published policy and which parts you could only learn from someone's account — and what that difference implies about how well anyone can reason about the system from outside it.

Part D — The stigma question (§44.4)

R. State the reduction mechanism in four sentences. State the intensification mechanism in four sentences. Neither may contain the word obviously.

S. Both mechanisms run through the same psychological variable. Name it, and explain precisely why the drug pushes it in opposite directions depending on what is being attributed.

T. §44.4 argues that wider access may strengthen the intensification mechanism. Reconstruct that argument in your own words. Then say whether you think it is a reason to want narrower access, and defend your answer.

U. Give an example — from any domain, not necessarily health — of blame relocating from having a problem to not fixing it once a fix became available. Identify what made the relocation possible.

V. Explain why the chapter refuses to rate the stigma claim, and why a non-informative rating is argued to be worse than no rating. Do you agree? If a rating were forced, which would you give and what would you lose by giving it?

W. † Find four descriptions of this drug class in the wild: an advertisement, a video, a comment thread, and a news article. For each, record (1) whether weight is framed as a condition or as a choice, (2) whether anyone's body is discussed without their disclosure, and (3) which of the two §44.4 mechanisms the framing would tend to reinforce. Name no one. Characterize by type. Write half a page on the pattern.

X. §44.4's risk callout argues that stigma is a clinical variable. List the documented routes by which it affects health outcomes, and explain why that makes it a legitimate subject for a science book rather than a culture essay.

Part E — Medicalization and equity (§44.5–44.6)

Y. Build a two-column table of what medicalization gives and what it costs. Then mark each entry according to whether it is primarily an effect on individuals, on institutions, or on politics.

Z. † Explain the crowding-out mechanism, and say explicitly why it requires no one to intend it. Then find and describe one historical example from public health outside this drug class, and identify what evidence would show that crowding-out actually occurred rather than merely being plausible.

AA. † Chapter 12 argued for the food environment as a causal contributor; §44.5 argues that a medical frame treats the resulting condition one body at a time. Write a 1,000-word essay arguing that these are compatible. Then write a 300-word rebuttal to yourself that you actually find uncomfortable.

AB. Reproduce the four-phase access curve. Mark where you believe this drug class currently sits, state your reasoning, and — critically — state what observation would show you had placed it wrong.

AC. The chapter states that these conditions are concentrated in lower-income populations and that early access is concentrated in higher-income ones. Explain why that combination is predicted by the access curve rather than by anyone's malice, and why the "no malice" point does not make it less of a problem.

AD. † Read Case Study 1 on insulin. Then list three features of the insulin story that make it a good analogy for this drug class, and three that make it a poor one. A good answer will find the poor-analogy features harder to generate — say why.

Part F — The two cases, and the method (§44.7–44.9)

AE. Explain the difference between a prediction and a falsifiable condition. Then convert one prediction you have heard about these drugs into the second form.

AF. §44.7 and §44.8 each contain five conditions. Pick the two — one from each list — that you believe are most likely to be resolved by evidence within five years. Justify both choices by describing the study or dataset that would resolve them.

AG. † The chapter constructs the two cases symmetrically and refuses to weight them. Write the argument that this is intellectually honest, then the argument that it is a dodge. Decide which is stronger, and write a paragraph on what your answer reveals about what you want from a book like this.

AH. Explain what a values question is, why the rating system does not apply to one, and why recognizing that is described as part of the method rather than an exception to it.

AI. † Take a confident claim from an area completely outside this book — a supplement, a training protocol, a financial product, a policy proposal, a consumer technology — and run all six questions from §44.9's diagram against it in writing. Then answer two meta-questions: how long did it take, and which of the six did you most want to skip? The one you wanted to skip is the one to watch.

AJ. † Write your dated social prediction from the Dossier section, with a falsifier and a check date, and put it somewhere you will actually find it. Then write one paragraph on what you expect to learn from checking it that you could not learn any other way — and, in two sentences, state the book's thesis as §44.9 gives it, and why holding both halves at once is harder than believing either.