About This Case Study

When Hurricane María struck Puerto Rico in September 2017, it knocked out power to the entire island, some of it for the better part of a year. The government's official death toll settled at 64, and it stayed there for nearly a year — even as funeral directors, journalists, and researchers insisted the true number was far higher. Independent studies eventually put the death toll in the thousands: a study by researchers at the George Washington University, commissioned by Puerto Rico's own governor, estimated roughly 2,975 excess deaths, and the government revised its official figure to about 3,000. A separate study published in the New England Journal of Medicine by researchers at Harvard estimated a still-higher number, with a wide range of uncertainty. However the exact figure is drawn, the gap is staggering: an official count of 64 against credible estimates nearly fifty times larger. This case study asks how that is possible — and works from the studies themselves to answer it.

The answer is that a death toll is not a fact you read off a ledger; it is produced by a method, and the choice of method is shaped by politics. There are two basic ways to count a disaster's dead. The first counts direct deaths — deaths a medical examiner attributes to the storm itself, such as drowning or trauma; this is what Puerto Rico's official 64 measured. The second counts excess mortality — the gap between the number of people who actually died in a period and the number who would have been expected to die had the disaster not occurred; this is what the independent studies estimated. Excess mortality captures the deaths a direct count misses: the dialysis patient who dies when the power fails, the heart attack during months without electricity, the untreated infection when hospitals go dark. The analytical move in this case study is to stop treating "the death toll" as a single objective number and start treating the count as an instrument — to ask what each method measures and misses, why the official number stayed so low for so long, what the low number did politically, and how Puerto Rico's status as a U.S. territory without full political voice shaped both the disaster response and the authority to count its dead. You will reconstruct the two ways of counting from the studies, work out why they diverge, and take a position on how the dead of a disaster should be counted, and by whom.

Before You Begin

Have ready:

• Kishore, Nishant, et al., "Mortality in Puerto Rico after Hurricane Maria", New England Journal of Medicine (2018) — the Harvard household-survey estimate.
• Milken Institute School of Public Health, George Washington University, "Ascertainment of the Estimated Excess Mortality from Hurricane Maria in Puerto Rico" (commissioned by the Governor of Puerto Rico) — the study behind the revised official toll.
• A note-taking surface, and a calculator or spreadsheet if you want to check any figures — paper, a document, or a shared doc if you are working in a group.

A note on the sources: the two studies use different methods and arrive at different estimates, and neither is "the answer" in the way the official 64 pretended to be. Read them for how they count, not only for the headline number — and note that even careful estimates come with a range of uncertainty rather than a single exact figure.

Before you begin, write down, in a sentence or two each:

• When you hear that a disaster "killed 64 people," what do you picture that number including? Does it include someone who died three weeks later because their hospital had no power?
• Who should get to decide the official death toll of a disaster — the government responsible for the response, or independent researchers? What is at stake in the answer?

The Exercise

Phase 1: Orientation — The Official Number (6–8 minutes)

Fix the shape of the dispute in your own words. The government's official toll was 64, and it held for nearly a year before being revised to roughly 3,000. State the gap plainly, and the puzzle it poses: how can an official death toll be off by a factor of nearly fifty? Note that both figures claim to count the dead of the same event — so the disagreement is not about what happened but about what counts as a death caused by the disaster.

For groups: agree on a one-paragraph statement of the puzzle before going further.

Phase 2: Two Ways of Counting (12–15 minutes)

Work through the two methods, using the studies. The point is to understand why sincere, competent counts can differ so enormously.

1. Direct deaths. What does the official method count — deaths a medical examiner certifies as caused directly by the storm? Explain why this method produces a low number, and list the kinds of death it leaves out: the person on dialysis who dies when the power fails, the heart attack during the blackout, the untreated illness when a hospital goes dark. Why are these hard to certify as storm deaths one by one?
2. Excess mortality. What do the independent studies count instead — the difference between the deaths that actually occurred after the storm and the deaths that would have been expected in a normal period? From the studies, note each estimate and its range: the GWU study's figure of roughly 2,975, and the Harvard study's higher estimate with its wide confidence interval. Why does this method capture the indirect deaths the official count misses?
3. Why they diverge. In your own words, explain why the two methods produce numbers so far apart. Which question is each one answering — and which better captures what the hurricane actually did?

For groups: one or two members present the direct-death logic and another the excess-mortality logic before the group compares.

Phase 3: Why the Count Mattered — Politics and Colonial Status (10–12 minutes)

A death toll is not only a measurement; it is a political fact with consequences. Work through what the low number did, and the conditions that produced it:

• What did the official count of 64 do politically while it stood? Consider how the apparent scale of a disaster shapes the federal response it receives, the public urgency behind relief, and the accountability that attaches to the officials in charge. What does an undercount do to a disaster that is still killing people months after landfall?
• How did Puerto Rico's status shape both the response and the count? Puerto Ricans are U.S. citizens, but the island is a territory whose residents cannot vote for president and have no voting representation in Congress, and which depends on federal decisions it cannot fully influence. How might that political position have affected the resources sent, the scrutiny applied, and the authority to say how many had died?
• The strain test. Reading the death toll as a produced, political number is clarifying — but where does that reading strain? The excess-mortality estimates are themselves uncertain: they come as ranges, rest on assumptions, and disagree with one another. Treating every count as "political" could slide into the claim that no number is truer than any other — which would let the official 64 off the hook. How do you hold both at once: that death tolls are produced and contested, and that some counts are nonetheless far more accurate than others?

For groups: move quickly through the first two questions and spend the most time on the strain test.

Phase 4: Reach a Position (8–10 minutes)

Take a position in a short written verdict. How should the dead of a disaster be counted, and by whom?

• Should an official toll be based on excess mortality rather than certified direct deaths? What would be gained, and what — precision, timeliness, certainty about individuals — might be lost?
• Who should hold the authority to produce the number: the government responsible for the response, independent researchers, or a mandated independent body? Why?
• Name the evidence that most moved you and the one thing you would still want to know to be confident in your judgment.

For groups: hold a short structured debate — assign members to argue for different answers — then see whether the group can converge, and notice what the disagreement turns on.

Phase 5: The Count Elsewhere (optional, 5–7 minutes)

The question this case raises — who counts the dead, and by what method — reaches well beyond María. Excess mortality is the measure researchers reach for whenever official counts are suspected of missing indirect deaths: in the COVID-19 pandemic, in extreme heat waves, in other hurricanes. Choose one such case you know or can look up and ask: was the "official" toll likely an undercount, by the same logic that undercounted María? Who benefited from the lower number, and who did the work of producing a truer one? Notice that this is the same move the unit's social-autopsy method makes — looking past the official count to the deaths a disaster really caused.

Closing Reflection

In two or three sentences, complete this thought:

The official count and the independent estimates diverged so sharply because __________. The undercount mattered because __________. On who should count the disaster dead, my position — and the hardest part of holding it — is __________.

Write the most precise version you can; the value is in stating exactly why the methods differ and owning the judgment about which count a just system should use.

A Note on Modes

Solo mode. Work through the phases in order, keeping your figures and definitions in brief notes. The two ways of counting from Phase 2 and your explanation of the divergence and its stakes in Phase 3 are the central artifacts; keep them for use with the unit's questions on disaster, data, and accountability.

Group mode (3–6 people). Designate a timekeeper. Members present the direct-death and excess-mortality logics in Phase 2; Phase 3's strain test works well as full-group discussion; Phase 4 can be a short structured debate before a synthesis. If time is short, the closing reflection can be individual writing after the session.

After the Case Study

• The GWU/Milken report documents its methodology in full; reading how excess mortality is estimated — the expected-deaths baseline, the assumptions, the uncertainty range — shows what such a number can and cannot claim.
• The move at the center of this case — counting the deaths a disaster really caused rather than the ones officially attributed to it — is the same one Eric Klinenberg makes in Heat Wave: A Social Autopsy of Disaster in Chicago (University of Chicago Press, 2015), in the unit's Key Scholarship; read together, the two cases show that who counts the dead, and how, is itself a question of power.
• The politics of counting recurs across crises — pandemics, heat waves, wars — wherever an official toll and an excess-mortality estimate diverge; the pattern of who prefers the lower number is worth tracing across cases.