Measuring child well-being, and what rankings do with it

Comparative reports, notably those published by Unicef's research centre, combine heterogeneous dimensions: health, education, monetary poverty, reported satisfaction, relations with peers. That centre, based in Florence, has published comparative reports on rich countries for more than two decades.

Aggregation requires converting those dimensions onto a common scale and then weighting them. Weightings are argued choices, never data, and the reports say so in their methodological annexes.

Sensitivity analyses show that a country's rank can move several places depending on the weights chosen. A country placed well on health and poorly on reported satisfaction changes position with the importance given to each.

That instability does not invalidate the exercise, but it forbids the most common reading, which consists in commenting on a rank as if it measured a single reality. A rank remains ordinal information: it says that one country is ahead of another, not by how much.

A frequent response is to publish the ranking for each dimension alongside the overall rank. It puts the composite result in perspective, but most coverage retains only the composite.

Asking a child to rate their life on a scale produces answers influenced by local response norms, by the translated wording and by desirability. The same score on a scale from zero to ten does not necessarily mean the same thing in two languages, nor in two cultures where it is more or less usual to say one is very satisfied.

International comparisons of those scores are therefore the most fragile of the set, although they are the most widely repeated because they produce spectacular gaps between similar countries.

Health and monetary poverty data are more comparable, because they rest on harmonised definitions. They remain sensitive to conventions: a relative poverty line measures position in the national distribution, not absolute living standards. The threshold of sixty per cent of median income, commonly used in Europe, is the best-known example.

A country may therefore appear to struggle on a relative indicator while having a higher standard of living than a better-ranked country, which is an arithmetical result and not an anomaly.

Their main contribution is not the final ranking but the series by dimension, which allow a country to be tracked over time on a stable measure. Following the evolution of the same indicator, defined in the same way, tells us more than a one-off comparison between countries with different statistical systems.

That is also what the authors themselves recommend, and their methodological warnings almost systematically disappear from public coverage of the rank obtained.