São Tomé and Príncipe: Poverty & Inequality

A standing is a place in a list, not a mark out of anything. This one is part of the Country Index 2026, and it is worked out like this.

1. A few key indicators decide it; the rest are context. 2 indicators on this topic are keys and 13 are charted alongside them without a vote. That split is the point: if every statistic voted, importing five more of one kind would quietly change what the topic is about.
2. Each key is read at one edition. All of them describe 2023.
3. Every country’s published value is turned into a place within that one key’s own field, and then expressed as a share of that field — how far down this key’s list the country sits, rather than how many countries are above it. Keys where a higher published number is worse (terrorism, hunger, disaster risk) are turned round first, so the good end is always the same end.
4. Dividing by the field first is the whole reason this works. A place carries the size of its list inside it: 22nd of 44 and 108th of 215 are the same showing, and averaging the numbers 22 and 108 says loudly that they are not. Fields on this site run from about 44 countries to about 215, so averaging raw places would reward a country for happening to be measured in small ones. Dividing each place by its own field removes the size and leaves only the showing.
5. The shares are averaged, plainly. One key, one vote — no key is weighted above another, because any weighting would be this site’s opinion presented as a measurement. Countries are then put in order by that average, and the average itself is never published. It is our arithmetic with nothing anchoring it, so printing it would invite it to be read as a measurement somebody made. The position it produces, the band, and the keys it was built from are all published instead — every one of them checkable against the tables on this page.
6. A country needs 2 of this topic’s 2 keys — half, rounded up, and never fewer than 2. A standing built from one reading, printed beside one built from four, is two different things sharing a column heading. 159 countries clear the bar here.
7. A key has to earn its vote every edition. It must measure at least 120 countries (below that it starts measuring who was surveyed rather than the world), it must not have more than 10% of its field tied at either end, and it must not repeat another key in the same topic — Spearman correlation above 0.92 counts as repeating. A topic needs at least 2 surviving keys to place anyone at all; this one has 2.
8. A standing is relative, and the grouping is ours. It says where a country places against the others measured alongside it, so it moves when they move. The publishers listed on this page did not agree to be read together, and each measures its subject its own way: the underlying numbers are theirs, the ordering on top is this site’s — the same method as the rest of the Country Index, applied to one subject instead of all of them.