Fifty States,
Thirty-Five Countries

The device, taken literally

A chart circulated widely this year. It set Germany's social statistics against the fifty American states and gave each one a rank: life expectancy, first; homicide, first; gun deaths, first; nine first places in all, and then, at the bottom, GDP per capita, fiftieth. On the measures the chart used, Germany would be the best state in the union.

The device works because it answers the standard objection in advance. Comparisons between countries invite the reply that America is too large and too various for an average to mean anything, and that Mississippi and Massachusetts are different countries in all but name. Ranking against the distribution grants that and continues. A country that beats Massachusetts has beaten the best case, not the average.

What follows takes the device literally rather than rhetorically. Thirty-five countries were placed against the fifty states on thirteen metrics, using, wherever both sides existed, sources that measure the same thing on both sides of the comparison. Rank one means better than all fifty states. Rank fifty-one means worse than every one of them. The District of Columbia is reported where it is informative and never counted among the fifty: it is a city, and a city of a particular kind.

The thirty-five are a selection, not a population, and it is worth saying so rather than inventing a rule they satisfy. They comprise the larger European economies, the Nordic and Baltic states, and the non-European rich democracies — Japan, South Korea, Canada, Australia, Israel. To these were added seven countries that sit outside that circle and were chosen to show where the pattern ends: Russia, Serbia, Bosnia and Herzegovina, Mexico, Brazil, South Africa and India. Several OECD members are absent, among them Iceland, Luxembourg, Slovakia, Slovenia, New Zealand, Chile and Türkiye, as are Bulgaria and Romania. Whether their inclusion would change anything was not tested, so nothing here is claimed about it. The selection is richer and more Western than the world is, and a reader should carry that into everything below.

The results fall into three parts. On one set of metrics the gap is wide enough that no objection about definitions or reference years touches it. On a second set the ordering reverses. And a third set of comparisons cannot be made at all, including several that look straightforward. Working out which was which took most of the effort behind this piece.

Where the answer is not close

On the metrics a legislature sets, the result admits almost no exception. On the metrics that accumulate over decades, it admits a specific and coherent set.

Firearm deaths among children: twenty-nine of twenty-nine. Counting ages one to nineteen, every country with a figure ranks ahead of every American state. The margins are the widest in the dataset. Pooling five years so that the smaller states have a stable rate, the lowest American state records 1.0 deaths per 100,000, the median 5.6, Louisiana 14.7. Canada records 0.73, Sweden 0.44, Germany 0.05, the United Kingdom 0.03. Japan, Hungary, Ireland and the three Baltic states record none at all. Germany's margin over the safest American state is a factor of twenty; Poland's is thirty-three; set against Louisiana the German figure is smaller by a factor of 294.

That row has a gap. It was compiled for the thirty rich democracies, before Mexico, Brazil, South Africa, India and Serbia were added, and those five carry no figure in it. Three of them would probably not rank first. The result holds for the countries the row covers, not for the world.

Firearm deaths, all ages: twenty-nine of thirty-two. The safest state, Hawaii, records 3.9 per 100,000; the median state 14.8; Mississippi 27.6. The United Kingdom records 0.15, Poland 0.27, the Netherlands 0.36. Switzerland, with one of Europe's highest rates of civilian gun ownership, records 2.53 and still comes first. The figure includes suicides, which are the majority on both sides — roughly two thirds of the American total.

Imprisonment: thirty-one of thirty-four. Counting prisons and jails together, including people held before trial, the median American state holds 610 people per 100,000. Massachusetts, the lowest, holds 241; Louisiana 1,067. Finland holds 53, the Netherlands and Sweden 63, Germany 77. Only three countries fall inside the American range — South Africa at 286, Brazil at 354, Russia at 416 — and all three still rank ahead of most states.

Life expectancy: twenty-one of thirty-four. Here the exceptions begin. The best American state is Hawaii at 80.0 years, the worst West Virginia at 72.2, the median 77.3. Spain reaches 83.2, Switzerland 83.7. Fourteen countries fall inside or below the American range, in two groups. The post-communist states: Czechia 79.0, Estonia 78.1, Croatia 77.7, Poland 77.2, Hungary 76.0, Lithuania 75.8, Serbia 75.2, Latvia 74.5, Russia 72.5. And the middle-income additions: Mexico 75.2, Brazil 74.9, India 71.7, South Africa 65.5 — the last two below every American state.

Homicide: twenty-four of thirty-three. New Hampshire records 1.3 per 100,000, the median state 5.4, Mississippi 18.6. Switzerland records 0.38, Germany 0.39. The exceptions are the Baltic states — Latvia 3.24, Lithuania 2.22, Estonia 1.96 — Russia at 4.98, two narrow misses in Canada (1.43) and Israel (1.56), and then a different order of magnitude entirely: Brazil at 21.24 and Mexico at 29.15, both worse than every American state, Mississippi included.

So the line does not run between Europe and America. It runs three ways. On what a legislature decides — who may own a firearm, who goes to prison — Latvia beats every American state as comfortably as Norway does, and Brazil beats most of them on imprisonment while being four times more lethal. On what accumulates over decades, the formerly communist east sits inside the American distribution and the middle-income countries below it. And on lethal violence, Mexico and Brazil occupy a category the American range does not reach.

What the first group has in common is worth stating plainly, because it is the finding rather than a gloss on it. Each is the product of arrangements a government chooses and can change: how guns are regulated, how often the state imprisons and for how long, how roads and cars are engineered and policed. None of them is a fact about geography, climate or national character.

Where it reverses

On economic output the ordering inverts, and it inverts almost completely. Of thirty-five countries, exactly one ranks ahead of any American state on GDP per capita. That country is Ireland, and its figure is an artefact.

The poorest American state by output is Mississippi, at $55,877 per head in 2025. Mississippi is a byword for American poverty; it is also richer per head than France ($48,930), Italy ($43,270), Czechia ($35,678), Portugal ($32,166), Greece ($27,041) and Poland ($28,374). Germany, at $60,439, clears Mississippi and nothing else.

The obvious objection is that a dollar buys more in Jackson than in Manhattan, and it is a fair one. The American statistical agencies publish exactly what is needed to answer it: a price level for each state against the national average. Dividing each state's output by its own price level puts both sides in dollars at American average prices, the unit international purchasing-power figures already use.

The adjustment moves things, but not in the direction most people expect. Mississippi's price level is 87.0, so its adjusted output rises to $64,261 — further above France, not below it. The countries it helps are the expensive ones: Germany moves from fiftieth to forty-third, Denmark from thirty-fifth to twenty-third, Norway from twelfth to third. Switzerland, where prices are high, falls from second to seventh. The Netherlands land twenty-seventh, just below the median state — the one country in the set that is both squarely mid-table in output and first in the safety metrics. Italy, Spain, Poland and Czechia do not move at all: they remain below every American state on both measures.

Which leaves Ireland. Its nominal output per head is $130,652, and its price-adjusted figure $152,632. On both it ranks first, ahead of New York and North Dakota respectively. The number is not wrong in the sense of being miscalculated; it is wrong in the sense of not measuring what its name suggests. A large share of it is intellectual property booked in Ireland, aircraft leasing capital, and retained profits of companies redomiciled there. Ireland's own statistical office says so, and publishes a corrected measure for exactly this reason: modified gross national income. That figure is 55.4% of GDP — the two differ by a factor of 1.80, so roughly 45% of the recorded output disappears once the globalisation effects are removed. On the corrected measure Ireland ranks twenty-eighth, one place below the Netherlands.

First or twenty-eighth, from the same economy in the same year, depending on which of two official numbers you use. The Dutch and Belgian figures were checked for the same distortion and do not show it; Ireland is a particular case, not a general feature of corporate hubs.

So the device cuts both ways, and the second cut is the one that rarely appears in the version that circulates. A country can be first in life expectancy and fifty-first in output at the same time, and both numbers are correct.

The metric that fits neither family

One measure belongs to neither group. The pandemic was neither decided in a legislature nor accumulated over fifty years. It arrived, and within two years it had done its sorting. The measure is not confirmed deaths from covid — that figure depends on how much a country tested and how it coded, so a state that tested less records fewer deaths without anyone having survived. The measure is excess mortality: deaths above the number that would otherwise have been expected, whatever they were later called.

On every metric so far the countries clustered on one side of the American distribution or the other. On this one they are spread right across it.

Four came through better than every American state: Australia at 91.9 excess deaths per 100,000, Denmark 109.9, Japan 123.9, South Korea 124.6, against Hawaii's 130.0. Four came through worse than every state: Russia 948.9, Serbia 920.2, Lithuania 919.7, Bosnia 797.4, against Mississippi's 655.0. Everything else is in between — including the United States itself at 390.6, against a state median of 363.7. Worse than most of Western Europe, better than all of Eastern Europe, and better than Italy.

The three groups this piece has been using — rich democracies, the post-communist east, the middle-income countries — do not survive contact with this metric. The rich democracies, which sit close together on almost everything else, run from Australia at 91.9 to Italy at 464.9, a fivefold spread. The middle-income countries, far below everyone on output and far above them on violence, land in the middle rather than at the bottom: Brazil, at 379.3, finished ahead of Italy and ahead of the United States. Only the post-communist east stays together, and it stays together badly, from Estonia at 396.1 to Russia at 948.9. Mexico finished worse than Poland. Lithuania lost 5.3 times as many people above baseline as Denmark, and both are small northern European democracies in the EU with universal health systems. Even the Nordic bloc, which behaves as one unit everywhere else in this comparison, splits across half the range: Denmark 109.9, Norway 139.3, Sweden 172.8, Finland 240.4.

Three cautions belong with the figure. India's value is a model output rather than a count: India has no complete series of total deaths for the calculation to start from, and estimates of its pandemic excess differ by several multiples between research groups. It is shown because it comes from the same model as every other country here, which makes it consistent, not reliable, and no claim in this piece rests on it. Nothing here is age-standardised either, so comparing Japan with Utah is partly a comparison of who was there to die. And the baseline — the estimate of deaths that would have happened anyway — moves the small numbers a long way: between the two standard approaches the gap runs to 63% for Denmark, and it is largest exactly where the excess is smallest. Between Australia and Ireland lie seven countries inside the range the baseline choice alone can move, and no order among them should be asserted. What holds is the spread, not the sequence.

The income question

Output per head is not income, and the gap between the two is wider in the United States than anywhere else in this set. American GDP contains more capital income, more hours worked per person, and a medical system that costs twice as much and counts as economic activity while doing so. So the natural next question is what a worker is actually paid — and this is where the exercise becomes genuinely difficult.

Start with average wages, converted at purchasing-power rates and with each state's output divided by its own price level, as before. On that measure the United States leads at $81,116, Switzerland is close behind at $80,069, and everyone else is some distance back: Denmark $68,167, Germany $66,770, France $56,523, Poland $47,837. Ranked against the states, Germany lands forty-seventh and France fifty-first. Mississippi again sets the floor, and again it is higher than most people expect: an average wage of $53,223 at a price level of 87.0 is $61,209 in purchasing power — above France, above Sweden, above Finland.

Then take the median instead, and the picture changes sign.

American wages are distributed far more unequally than any other country's here. The ratio of the ninth decile to the first — what the top tenth earns against the bottom tenth — is 4.79 in the United States. Poland follows at 3.52, the United Kingdom at 2.95, Germany 2.89, Sweden last at 2.21. The distance between the United States and Poland is larger than the distance between Poland and Sweden. And the dispersion is not only at the top: the ratio of the median to the first decile is 1.93 in the United States against 1.69 in Poland, so the bottom tenth sits further below the middle as well.

Applied to the wage figures, that dispersion does most of the work. Against Germany, an American advantage of 14.3% at the mean becomes a deficit of 9.0% at the median: the typical German worker earns more than the typical American one. Against the United Kingdom, 29.9% falls to 10.0%; against France, 42.9% to 25.0%; against Poland, 82.3% to 61.3%. The advantage does not vanish, but it shrinks by roughly a third to two thirds, and against the largest European economy it reverses.

That calculation deserves its caveat in the open. The median levels are derived — average wage multiplied by a published median-to-mean ratio — and the ratio exists only for countries with a statutory minimum wage, which excludes Italy, Switzerland and the Nordics. It is a magnitude, not a precision figure. No single source publishes comparable median wage levels for every country in this set, which is why the comparison appears here as prose and not as a rank.

A correct number that measures something other than its name promises is a general hazard, not a quirk of wage statistics. A weekly survey of 250,000 people overstated vaccine uptake by seventeen points, because size does not correct for selection; Mass Is Not a Method works through why a large biased sample can be worth less than a small random one.

What the traps cost

Every metric in this exercise exists in several versions, and the versions do not agree. The differences are usually larger than the differences between countries, which means the choice of version decides the result. Five cases, each of which would have produced a wrong answer if taken from the nearest available source.

Obesity is measured two ways, and no constant converts between them. The American state figures come from a telephone survey in which people report their own height and weight; respondents shade both. International figures are often measured by clinicians. The error cannot be adjusted away: the gap between the two readings is 7.6 points for the United States, 6.6 for Germany, 12.8 for Portugal and 0.3 for Latvia. The instruments have to match rather than be reconciled. That they do can be checked: the OECD's self-reported figure for the United States, 34.2%, is exactly the median of the fifty state values from the telephone survey. On the measured series Germany reads 24.5%; on the self-reported series, 17.9%. Only the second belongs in this comparison.

Homicide counts run in opposite directions on either side of the Atlantic. In Germany the police figure (0.75) is nearly double the death-certificate figure (0.39); in the United States the police figure (4.93) is below the death-certificate figure (5.80). Mixing the two sources therefore distorts twice, once in each direction. Since the American state data are death certificates, the country figures must be too.

Imprisonment doubles depending on which institutions you count. Counting only state prisons, the American median is 299 per 100,000. Counting prisons and jails together, including people held before trial, it is 610. International prison statistics include pre-trial detention as a matter of course, so the second figure is the comparable one.

Advertised wage replacement is not paid wage replacement. Thirteen American states now pay parental leave, and their headline replacement rates look generous. Washington D.C. advertises 90%. Because a weekly cap overrides the sliding scale at average earnings, the rate actually received by an average earner in D.C. is 43.7% — the worst in the country, behind the state with the lowest advertised rate. In seven of the thirteen programmes the cap, not the published percentage, determines the outcome.

Weeks of leave without the replacement rate inverts the ranking. The United Kingdom grants 39 weeks of maternity leave against Germany's 14, and pays 29.9% against Germany's 100%. Measured in full-rate equivalent weeks, the United Kingdom provides 11.7 and Germany 14.0. Plotting weeks alone would have shown the opposite of the truth.

In each case the misleading version is the one a competent person would reach for first.

The pattern is not confined to official statistics. In marketing measurement the same spending has been shown to return +4,100% when estimated by regression and −63% when measured by a randomised field experiment — two methods, one budget, opposite signs. That case is worked through in The Second Experiment, on rhinegold Insights.

What cannot be compared

Eight metrics were prepared, checked and then left out. They are worth listing, because the reasons are not technicalities and because a chart that included them would look more complete while being worth less.

Median household income and the poverty rate cross their own definitions. The international figures are net, equivalised for household size, and measure poverty relative to each country's own median. The American state figures are gross household income and an absolute poverty threshold. Poland's 9.1% poverty rate does not describe less hardship than the American 18.1%; it describes a flatter distribution.

The Human Development Index has no state-level counterpart at all.

Maternal mortality is published by thirty-nine states. The other eleven record too few deaths for the statistical agency to release a rate. A first place against an incomplete field is not a first place.

Statutory paid vacation is genuinely zero in all fifty states, which was verified rather than assumed: federal law requires no paid time off, and no state mandates any. Three states — Illinois, Maine and Nevada — require forty hours of paid leave usable for any purpose, which is five days by another name. But a rank in which every state scores zero carries no information.

Life satisfaction, which would have been the most valuable row of all, because it checks the whole exercise against what people say about their own lives. State-level figures on the same eleven-point ladder do exist, in one academic source pooling 2008–17. They cannot be joined to the country figures, and the arithmetic is unusually clean about why. The same survey house asked the same question of the same country over the same years through two different vehicles and got 7.08 and 6.90 — an instrument gap of 0.18 ladder points. The entire spread across all fifty-one American jurisdictions is 0.52. The median gap between neighbouring states is 0.005, and every one of the fifty neighbouring gaps is smaller than the instrument gap; thirty-six of the fifty-one states fit inside a single 0.18 window. Between countries the measure survives — their spread, 3.06 points, is seventeen times the gap.

Wages, in every form. The average misleads, as shown. The median has no source covering all countries. And wage dispersion, the measure that would explain both, fails for a reason that no amount of care can repair.

That reason is the most important thing this exercise produced. Measured from the American occupational survey, the ratio of top to bottom decile for the United States as a whole is 4.12. Measured from the OECD's earnings database, the same country in the same period reads 4.79. The method alone accounts for 16% — more than the distance between most of the countries in the ranking. But even with that offset corrected, the comparison would remain invalid, because a state is not a nation. Germany's dispersion contains the distance between Munich and Mecklenburg. Vermont's contains Vermont. For life expectancy or homicide this does not matter; each is a rate that a smaller unit can hold as meaningfully as a larger one. For a measure of spread, the size of the unit is the measurement.

The device, in other words, has a domain. It works for quantities that are per-head and policy-made. It breaks on quantities that describe how a population is distributed — and it breaks silently, producing numbers that look like the others.

Four ways of not reporting

Four countries in the set cannot be read straight, and each fails differently. Sorting out how took more of the work than any other part of this exercise, and it is the part most likely to be useful to someone running a different comparison.

Russia ranks nowhere. Life expectancy 72.5 years, fiftieth of fifty-one. GDP per head fifty-first on both measures. Minimum wage fifty-first, at $1.86 an hour. Nothing first. Inside that there is a stranger figure: Russian infant mortality ranks fifth and maternal mortality third, while its overall life expectancy is the worst in the set. The explanation is in the sex difference — Russian men live 67.6 years, Russian women 77.8, a gap of ten years that no Western European country approaches. Russia is not a country where people die young. It is a country where adult men do. Four metrics have no Russian value at all, and its cause-of-death reporting stops at 2019, so for the war years there is nothing to read.

Bosnia and Herzegovina is the harder case, because Bosnia does report. For 2019 it recorded 38,829 deaths. Of those, the number coded as road traffic deaths is zero. The number coded as homicide is one. Deaths of undetermined intent: 918 of the 927 externally caused deaths in the file. Deaths of unknown cause: 4,966. A country of three million people did not have one homicide and no road deaths in 2019. What the file records is not a peaceful country but a cause-of-death coding system that leaves the category unfilled. Read without checking, Bosnia would publish as the safest country on earth by a wide margin — and the numbers would look exactly like the numbers from Norway.

South Africa is the third case, and the most subtle. The category is filled in — 6,435 road deaths are coded for 2019, so this is not Bosnia. But of 7,596 deaths coded as homicide, 6,800 sit under “sharp object” and eleven under the firearm codes. Meanwhile 8,340 of 8,369 firearm deaths sit under “discharge from an unspecified firearm”, which is the accident code. The weapon is recorded; the intent is not. The result is a death-certificate homicide rate of 12.85 against a police rate of 35.79.

India fails earliest of the four. Ninety-four per cent of deaths are registered, so registration is not the problem; but only 22% of registered deaths carry a medical certification of cause, ranging from 100% in Goa to 5.5% in Bihar. In the international mortality file India appears not at all — zero rows, every year.

India also supplies the clearest warning against reading a rank without reading what sits behind it. Its imprisonment rate is 36 per 100,000, the lowest in the set, well below Finland's 53. That looks like restraint. But 72.6% of Indian prisoners are held without trial, the highest share here by twenty-two points, and the prisons stand at 112.7% of capacity. India imprisons about ten convicted people per 100,000, a quarter of Finland's rate, while holding 26 per 100,000 who have been convicted of nothing — more than Finland and Norway hold in total. Japan makes the contrast: a similar rate of 33, but 15.8% pre-trial and prisons at 47.3% of capacity. The same number, two unrelated situations.

Russia withholds, and the absence is visible to anyone who looks. Bosnia supplies figures that pass every format check and mean nothing. Bosnia's kind of failure is the one that gets published.

What these files record is the coding system, not the country — an instrument can only report the layer it was built to observe. The same problem in a different field is set out in Measuring the Wrong Layer.

What the exercise measures

The thirteen metrics fall into two families, and nothing in the data connects them. Excess mortality, the fourteenth, belongs to neither.

One family is decided in legislatures: who may own a firearm, how many people are imprisoned and for how long, how many weeks of leave are paid and at what rate, what a road is engineered to survive. On these, the countries in this set beat the American states almost without exception, and the margins are large enough that no dispute about reference years or definitions touches them. Firearm deaths among children admit no exception at all across the twenty-nine countries measured. Imprisonment admits three, and all three still rank ahead of most American states.

The other family is produced by markets and compounds over decades: output per head, wages, and — more slowly and more indirectly — how long people live. On these the ordering reverses. Thirty-four of thirty-five countries produce less per head than the poorest American state, and the thirty-fifth is a bookkeeping artefact. Of the fourteen that fail to beat every state on life expectancy, nine spent the twentieth century on the other side of the Iron Curtain and four are middle-income countries outside Europe.

The argument between the two systems is normally conducted by each side selecting its family. Read one way, the American states are unmatched in what they produce. Read the other, they are outliers among rich societies in how many of their people are shot, imprisoned or killed on the road. Both readings survive contact with the data. Neither is a rebuttal of the other, because they are not claims about the same quantity.

What the exercise does establish is narrower and harder to argue with. The variety defence — that America is too large and too various to be compared with a single country — does not work on the first family. The American average is not being dragged down by its poorest states. Massachusetts, the least punitive state in the union, imprisons at four times the Dutch rate, and Hawaii, the safest, records firearm deaths at ten times the Dutch rate.

And on the second family, the honest version is less flattering to Europe than the chart that started this. A German worker's median wage is higher than an American one's; a German state's output per head is not, and neither is a French, Italian, Spanish, Polish or Czech one.

The obvious move at this point is to explain the exceptions away by endowment: Norway has oil, Switzerland has banks. That move does not survive contact with the map. Norway, the United Kingdom, the Netherlands and Denmark all drew hydrocarbons from the same North Sea. Norway banked the proceeds in a sovereign fund and ranks third in price-adjusted output; Denmark ranks twenty-third; the Netherlands, which had a syndrome named after its gas in the 1970s, ranks twenty-seventh; the United Kingdom, which spent the money, ranks fiftieth. Dutch production ended on 1 October 2024, when Groningen was closed by law after drilling-induced earthquakes. British output continues, 77% below its 1999 peak. One basin, four countries, four different places in the ranking. And on firearms, prisons and roads all four rank first, alongside countries with no hydrocarbons whatever. Whatever the North Sea explains, it does not explain that.

The device is worth using for what it can carry: rates, per head, produced by decisions. It should not be asked to carry distributions, and it should never be asked to carry a country whose statistical system does not fill in the category being measured. Most of the work in this exercise went into finding out which was which.

Sources and method

Every metric was taken from one source covering every country in that row, and from one source covering every state, with the two chosen to measure the same thing. Where that was impossible the metric was dropped rather than patched. Reference years differ between metrics and are stated per row; within a row they are held constant.

Country side

State side

Stated limits

Reference years are not uniform across metrics: violent deaths and road deaths are 2019 on the country side against 2024 on the state side, and imprisonment is 2018 against a mixed 2022–24 vintage. Where a country's margin is a factor of two or more this is immaterial; where it is a few per cent it is noted in the text.

Obesity appears only for countries whose survey year is within two years of the American one, which excludes France, Poland and Russia rather than excluding the metric. The row for firearm deaths among children covers thirty countries and not Mexico, Brazil, South Africa, India or Serbia. That the two sides of that row measure the same thing was verified rather than assumed, by reproducing the American 2019 count from both sources independently — 3,378 deaths from each, on the same denominator.

Purchasing-power figures chain two adjustments: an international parity from the country to United States average prices, and a regional price parity from those prices to the individual state. Both carry their own error, and the second was built for comparisons inside the United States rather than across borders.

The median wage comparison is derived from published average wages and published median-to-mean ratios, not from a single published series of median levels, which does not exist for this set of countries. South Africa's homicide rate appears in the text at its police value and is absent from the ranked column. India has no entry in the international mortality file at all. Neither gap was filled from elsewhere.

No figure on this page is age-standardised. The claim that firearms are the leading cause of death among American children traces to Goldstick, Cunningham and Carter, NEJM 2022, and was re-checked against the underlying data rather than cited from the paper.

Take the data

Everything here is reproducible. The figures below are the ones the charts and cards are drawn from, not a summary of them.

All 35 countries

One card per country, in series order. A dash means the country does not report that figure — it is not a zero. Countries with fewer ranked rows report less, not better.

The last column is the margin: how far the value would have to move before the rank changes. It turns a place into a distance. Germany is first on life expectancy at ×1.009 and first on firearm deaths among children at ×20 — the same word, two different orders of confidence. A margin near 1 means a revision, a later reference year or a definitional quibble could take the place away; a margin of 20 means nothing plausible can. Where the ranking in this piece is called comfortable or narrow, this column is what was being read.

Legal notice

Responsible for this site

Michael Zachrau
Putstraat 29
3630 Maasmechelen
Belgium

E-mail: michaelhzachrau@gmail.com

Responsible publisher (verantwoordelijke uitgever / éditeur responsable): Michael Zachrau, at the address above.

This is a private, non-commercial publication. It sells nothing, takes no advertising and carries no enterprise number. Identification is given under Book XII of the Belgian Code of Economic Law.

What this site collects

Nothing. There are no cookies, no analytics, no tracking pixels and no embedded third-party content. Fonts are served from this domain rather than from a font provider, so no request leaves for another server and no visitor IP address is passed to anyone. The site is static: it has no database, no forms and no login, and it receives no personal data to store.

The hosting provider (GitHub Pages, GitHub Inc., 88 Colin P. Kelly Jr. St., San Francisco, CA 94107, USA) records server log data including IP addresses for the operation and security of the service. That processing is GitHub's, described in the GitHub Privacy Statement, and is not accessible to the operator of this site.

Under the General Data Protection Regulation you may ask what data is held about you, have it corrected or erased, and lodge a complaint with a supervisory authority — in Belgium, the Data Protection Authority (Autorité de protection des données / Gegevensbeschermingsautoriteit). As this site holds no personal data, such a request would return nothing.

Accuracy and reuse

Every figure on this page carries its source and reference year, and the dataset is published in full so that any number can be checked against its origin. Where a comparison could not be made honestly, it was left out and the reason stated. Mistakes are possible all the same; write to the address above and corrections will be made and marked.

The text and charts on this page may be quoted and reproduced with attribution to mapgap.eu. The underlying statistics belong to the institutions that publish them and carry their own terms, which are linked in the dataset.