← All experiments

Child deaths in Colombia and Brazil: how sure can I be, and where is the bigger gap?

Under-5 mortality with its uncertainty intervals, and the gap between the poorest and richest fifth of families. An explainer that will not rank the two countries when the data cannot tell them apart.

Data
Explainer

Updated5 Oct 2026

The problem

Colombia's under-5 mortality rate is 11.5 per 1,000 live births and Brazil's is 14.2. Read as a league table, Colombia does better. I wanted to know whether the data supports that.

What was out there

Rankings compare the point estimates and leave out the uncertainty around them, and they rarely show the gaps inside a country.

What I built

An explainer that puts the uncertainty intervals and the gap between the poorest and richest families next to the headline numbers, with 27 self-tests checking every value it shows.

Step 2 of the explainer in its National view: under-five mortality of 11.5 for Colombia with a 90 percent interval of 7.1 to 19.0, and 14.2 for Brazil with 12.2 to 16.6, the two intervals overlapping — Colombia's best estimate is lower, but the two uncertainty intervals overlap, so the ranking is not confirmed.

Step 2 of the explainer in its By wealth fifth view: under-five mortality with intervals for each wealth fifth, from 15.8 to 8.2 in Colombia and from 21.0 to 8.6 in Brazil — The bigger gap is inside each country: in Brazil the poorest fifth's interval does not even touch the richest fifth's.

Equipment

WHO Global Health Observatory Data sources
Python and uv Code and infrastructure
Claude Code Harnesses and agents
Claude (Opus 4.6 to 5.5) AI models
Codex CLI Harnesses and agents
GPT (GPT-5.5 to GPT-6.1 in Codex) AI models

Results

The 90% uncertainty intervals of the two countries overlap, so the ranking is not confirmed. The larger gap is inside each country: in Brazil, children in the poorest fifth of families die at 21.0 per 1,000 against 8.6 in the richest fifth.

What I learned

Numbers of deaths in a what-if are illustrative, so I label them that way instead of calling them deaths averted.

Lab notes

5 Oct 2026

I built a second explainer from the same Colombia and Brazil comparison, on under-5 mortality, with 27 self-tests and an independent Codex review.