We have made an update of the indicator Life expectancy at birth. To see this indicator together with Income per capita, follow this link.
The biggest change is that we now show this indicator for 155 countries back to 1800, although in most cases, the early estimates are based on a very rough model. This full dataset is not suitable for statistical analysis. Please consult the documentation for information about sources and data quality. A spreadsheet with detailed source information will be added later.
You can also see some tentative information about data quality in the graph, look for this under “For advanced users” -> “Data quality”, or use this link. Red is “very poor quality data” while blue is “very good quality data”. Read more about our data quality ranking in this previous blogpost.
To be able to show Life expectancy at birth during 200 years, the indicator now contains:
a) estimates of average life expectancy from different sources
b) where no estimates are available, a constructed, simple model for life expectancy before 1950
See the full documentation for more details (you can find it here).
PLEASE NOTE: the model is a very simplified way of showing the approximate level of life expectancy.
The data for the 19th century are for many countries based on regional averages, which in themselves are only more or less qualified guesses. Hence, it is not advisable to draw any conclusions from the relative position in the graph of most of the individual countries in the 19th century.
What can then be seen in this graph? It is an illustration of the fact that in all countries health was considerably more poor 200 years ago, and that all have gone through more or less sustained periods of decreasing mortality and improving health. For the countries where we don’t have data showing exactly when health began to improve, we base the assumption of the timing of this “health transition” on the online bibliography of history Professor James Riley and his other published works (see the documentation). But any mistakes or errors are the sole responsibility of Gapminder.
When you play the graph over time, you will see that some countries move around a lot, while others move very smoothly or not at all. When the bubbles move a lot it is because we have estimates each year or at least most years, so the bubbles show the variation over time. In the cases where they move more smoothly, it is often because we only have data each five years, for example. And most countries during the early years do not move at all, which is because this is where we only make a rough assumption of their approximate average life expectancy during this period. It is reasonably safe to assume that for the initial period, when life expectancy was low, the mortality actually did vary strongly from year to year, due to vulnerability to starvation and disease (see the example of Sweden, where we have registration data for the whole period).
For most countries, the data for 2007 is based on a forward projection from earlier estimates.