Form
When the figure is a table
A table is the right form when a reader needs to read values off rather than compare shapes. What keeps it a figure rather than markup is that a column can hold its own marks: a sparkline beside the number it summarises says how a row got where it is, which the number alone cannot.
US deaths against the pooled rate of 21 peer countries, by age. Sortable — the default order is by excess deaths.
| Age group | 2021 | per 1,000 | per 1,000 | 1 = parity | 1980–2021 | |
|---|---|---|---|---|---|---|
| 65–74 | 722,827 | 21.59 | 14.10 | 1.53× | 250,817 | |
| 55–64 | 477,915 | 11.04 | 5.85 | 1.89× | 224,651 | |
| 75–84 | 827,017 | 49.96 | 37.85 | 1.32× | 200,538 | |
| 45–54 | 216,344 | 5.37 | 2.30 | 2.33× | 123,613 | |
| 85+ | 936,703 | 150.82 | 133.31 | 1.13× | 108,791 | |
| 35–44 | 125,358 | 2.91 | 0.95 | 3.05× | 84,208 | |
| 25–34 | 82,598 | 1.79 | 0.49 | 3.62× | 59,772 | |
| 15–24 | 38,523 | 0.90 | 0.30 | 2.97× | 25,550 | |
| 0–4 | 23,755 | 1.24 | 0.68 | 1.81× | 10,660 | |
| 5–14 | 5,978 | 0.14 | 0.07 | 1.95× | 2,917 |
Bor J, Stokes AC, Raifman J, et al. "Missing Americans: Early death in the United States—1933–2021." PNAS Nexus 2023;2(6):pgad173. Data: doi:10.17605/OSF.IO/BAWU6. · US deaths, US and pooled-peer death rates per 1,000, their ratio, and excess deaths vs the peer-rate counterfactual, per age group, 2021. Sparkline: the rate ratio 1980–2021.
The prose around a figure
A chart page is mostly prose, and the prose has jobs a paragraph cannot do alone.
The comparison holds the peer set fixed across every age band, so the ratio column is comparable down the table as well as across it.
Half the missing Americans of 2021 were under 65.