About the data
Methodology & sources
How each test works, where its data comes from, and where it falls short.
Sources
- World Bank Poverty and Inequality Platform (PIP) — percentile thresholds for each country, from the
world_100bin_revisedtable, 2021 PPP round. Public domain (CC0). - World Bank PA.NUS.PPP — purchasing power parity conversion factors used to translate local currency into international dollars. CC-BY 4.0.
- world-countries (mledoze/countries) — used only to look up each country's currency code. ODbL-1.0.
- Human Benchmark — the published median reaction time (273 ms) that the reaction-time curve is anchored to. Reported statistic only; we hold none of the underlying data.
Definitions
The percentile curves are built from per-capita income or consumption — household welfare divided evenly by the number of people in the household, with no adjustment for household size or composition. Values are expressed in 2021 international dollars (PPP), meaning a dollar buys roughly the same basket of goods no matter which country it's spent in. The World Bank publishes this as a daily figure, which we convert to an annual one by multiplying by 365. The dataset covers 162 countries.
How the calculation works
Everything runs in your browser. Your income is never uploaded or sent to a server. If you enter a household income, we divide it by the number of people in your household (per-capita) before comparing it to the percentile curves — the same per-capita basis the World Bank data uses. This is a simple even split, not an equivalence scale like the ones used in some official household-income statistics, which give extra weight to adults and less to additional household members. That means our numbers for people in larger households will read lower than figures from sources that use equivalized income.
Limitations
This calculator is a rough placement, not a precise measurement. A few things it can and can't account for:
- Some countries' figures are based on consumption surveys rather than income surveys, because that's what their national statistics offices collect. Income and consumption aren't directly comparable, but the World Bank uses them together as the best available measure of household welfare.
- What counts as "income" — pre-tax or post-tax, for instance — varies by country's source survey. There's no single consistent definition applied everywhere.
- Household surveys are known to undercount very high incomes: wealthy respondents are harder to reach and less likely to report accurately. The top percentiles here likely understate how much the richest people actually earn.
- Surveys aren't run in every country every year, so the underlying data was collected at different points in time across countries, not all in the same year.
Reaction time
The reaction-time percentile is an approximation, not a measurement against a dataset of our own. Your five valid trials are averaged and placed on a log-normal curve — the shape reaction times are conventionally modelled with — centred on the 273 ms median that Human Benchmark publishes from its own sample. That sample is large but self-selected: it is people who chose to take a reaction test, not a random draw of humanity.
Human Benchmark publishes no standard deviation, so the curve's spread (sigma = 0.30) is fitted so that roughly 200 ms lands near the top decile and roughly 350 ms near the bottom quartile — landmarks read off the same published histogram. Real reaction times are more skewed than a pure log-normal, so treat the number as a placement rather than a precise rank. Your equipment counts too: a wireless mouse and a slow display can add tens of milliseconds that have nothing to do with you.