EN ES
Contact

← All insights

A person analyzing data on a computer screen.

Plate · Salt flat, Salinas Grandes, Argentina — deciding where nothing is measured

Investigation

Deciding in statistical silence

Much of the world governs without reliable data — missing censuses, unmeasured informal economies, indicators that simply do not exist. Honest decision tools must carry that silence, not paper over it.

February 18, 2025 · Enpath · 3 min read

Every serious decision tool rests on a quiet assumption: that the data exists. For much of the world, it doesn’t — or it exists with gaps and distortions large enough to change the answer.

This matters to us because a decision-intelligence system that treats thin data as solid ground is not neutral. It is confidently wrong in exactly the places that can least afford it. So this piece looks at what the silence actually consists of, and what an honest system does about it.

How much is actually missing

The scale of the gap is documented, not anecdotal. The World Bank’s Statistical Performance Indicators — which score countries on the data they produce, disseminate, and use — show a persistent divide: high-income countries cluster near the top of the index while many low- and middle-income countries lack recent data for basic economic and social indicators (World Bank SPI). The UN’s own progress reports on the Sustainable Development Goals repeatedly note that for a large share of indicators, many countries have either no data at all or nothing recent enough to steer by (UN SDG progress reports).

And the silence is not only in poor countries’ statistics offices. It follows a pattern: the less formal, the less measured.

The informality blind spot

The International Labour Organization estimates that about two billion people — roughly six in ten workers worldwide — work in the informal economy (ILO, 2018). In Latin America the share is around half of all employment; in parts of Africa and South Asia it is far higher.

Informal work is, almost by definition, work that administrative data cannot see. A ministry planning employment policy from payroll records is planning for the minority of workers it can observe. Any AI system trained or grounded on those records inherits the same blind spot — and then states its conclusions fluently, as though the unobserved majority did not exist.

Even the ground truth moves

Foundational statistics are less settled than they look. Colombia’s 2018 census initially counted about 44 million people; after reviewing coverage, the national statistics office, DANE, estimated the true population at over 48 million — an omission of several million people that had to be reconciled through post-census adjustment, with real consequences for how resources and political representation were allocated (DANE, Censo 2018). Colombia’s statistical system is among the region’s strongest, which is the point: if the anchor number of a well-run system can move by millions, the precision of everything derived from it is softer than it appears.

Similar stories exist on every continent — census disputes, rebased GDPs that shift by double digits, survey frames a decade old. None of this is scandal. It is what measuring a complex society honestly looks like.

What an honest system does with silence

The wrong response is the common one: fill the gap with a model’s best guess and present the result at the same confidence as measured data. The silence disappears from view, and with it the decision-maker’s ability to weigh it.

We hold ourselves to a different set of rules:

  1. Silence is information. “No reliable data exists for this” is a finding, stated as plainly as any number.
  2. Every figure carries its vintage and coverage. A 2012 survey extrapolated to 2025 is labelled as exactly that.
  3. Estimates are never dressed as measurements. Modelled values are marked, with what the model assumed.
  4. Uncertainty must reach the decision. If the data could be wrong by enough to change the answer, the person deciding sees that before they decide.

A tool that follows these rules will sound less impressive than one that doesn’t. It will say “we don’t know” in places a competitor would print a number. We think that is precisely what makes it fit to decide with — because the people governing in statistical silence deserve instruments that tell them where the ground actually is, not instruments that paint a floor over the gap.

Sources