World Pulse · methodology
How to read anything on this site
This site is built on one commitment: never show a number it cannot source, and never fill a gap with a guess. What follows is what that means in practice — including the places where it makes the site look worse than a competitor that quietly interpolates.
Two datasets, deliberately separate
The world
21 indicators across 217 countries · 156,366 observations, from the World Bank's World Development Indicators. One consistent method across every country is what makes them comparable at all.
Pakistan
A separate, hand-verified snapshot from national sources — PBS, SBP, OGRA, NEPRA, the Economic Survey, NNS, PDHS, World Bank and UNICEF. Closer to the ground, but not comparable with other countries.
They are kept apart on purpose. A country page reads the global series so the comparison is fair; Pakistan's own pages read national sources so the detail is right. Where the two disagree, both are shown — see below.
The rules, and what they cost
Nothing is interpolated or carried forward
If a country reported in 1990 and again in 2004, this site shows two points and an empty gap — never a line implying fourteen years of trend nobody measured. Sparklines physically break across gaps longer than five years. The cost is that our charts look sparser than charts that smooth over the same holes.
Absence is not zero
A country with no figure is drawn as nothing — on the globe it is a hole you can see through. It is never a pale tint, because a pale tint reads as “low” and would invent a value we do not have.
Every rank carries its denominator
You will always see “169 of 217”, never “169th”. Denominators differ by indicator and by year: in 2024, 217 countries reported life expectancy but only 196 reported under-5 mortality. A bare rank would silently imply the larger field.
Ranks are computed within a single year
A country is ranked only against countries that reported in the same year it did. Ranking each country's latest value against every other country's latest value would quietly compare 2019 with 2024. Indicators with no better end — inflation, CO₂, population — are not ranked at all, because ordering them would assert a judgement this site does not hold.
Computed is labelled computed
Regional, income-group and world medians are arithmetic this site performs over real values. They are legitimate, and they are never presented as figures any institution published. Anywhere one appears it says so.
Every figure carries its vintage
Multilateral statistics run years behind. A figure is shown with the year it describes and how old that is in plain words — “2018 · 8 years old” — rather than a bare year on a current-looking page.
The world is read at its most recent well-covered year
Neither obvious rule works. Reading the newest year is wrong: child stunting's newest year has six reporters, and a global median from six countries is a confident number that means nothing. Reading the best-covered year is wrong in a way that is easier to miss: internet use is best covered in 1990, when nearly every country reported it as roughly zero. So each indicator is read at the most recent year that still holds 80% of its best coverage on record, and both years are shown on the dashboard.
Where sources disagree, both are shown
The World Bank puts Pakistan's 2018 child stunting at 37.6%; the National Nutrition Survey 2018 puts it at 40.2%. Same year, both sourced. This site does not average them, pick a winner, or let you find two different numbers on two pages with no explanation — it shows both and says they are unreconciled.
Evidence grades
Readings on the Pakistan surfaces carry a grade describing what kind of evidence stands behind them. A grade is about provenance, not confidence: a Grade A figure can still be old, and the freshness label says so separately.
- A
- Official direct measurement
- B
- Trusted institutional dataset
- C
- Reputable non-official source
- D
- Modelled from official inputs
- E
- Experimental estimate
Freshness
Current · Recent · Aging · Old baseline. Applied to how far a reading sits from its expected publication cadence, not from today.
Dataset version
Snapshot v2026.07
Pakistan's hand-verified indicators were last checked against primary sources on 21 July 2026. Population 241.5 million. The world spine is refreshed separately from the World Bank API and each figure carries its own retrieval date.
The editorial dateline on the Pakistan front page is the day that page was written, and is deliberately distinct from the date the underlying data was verified.
Corrections
Figures are corrected in place when a source revises them, and material corrections are noted in the reading's caveats rather than removed. If a number here disagrees with a source you trust, the disagreement is the interesting part — both are worth recording.